Monday, 5 July
List of Abstracts
Monday, 5 July
11.30 – 12.00
Opening Ceremony
Stable Markov Decision Processes Using Simulation Based
Predictive Control
Zhe Yang, Nikolas Kantas, Andrea Lecchini-Visintini, Jan M.
Room 1
08.40 – 09.00
Welcome Addresses
Plenary Lecture
Room 1
Chair: Giorgio Picci
09.00 – 10.00
Models of Real-World Networks: Inhomogeneous Random
Graphs and Convergent Graph Sequences
Béla Bollobás
Abstract – In this paper we investigate the use of Model
Predictive control for Markov Decision Processes under weak
assumptions. We provide conditions for stability based on
optimality of a specific class of cost functions. These results are
useful from both a theoretical and computational perspective.
When nonlinear non-Gaussian models for general state spaces
are considered, the absence of analytical tools makes the use of
simulation based methods necessary. Popular simulation based
methods like stochastic programming and Markov Chain
Monte Carlo can be used to provide open loop estimates of the
optimisers. With this in mind we provide conditions under
which such an approach would yield stable Markov Decision
12.00 – 12.30
Stochastic Model Predictive Control
(Invited Session)
Room 1
Organizers: Peter Hokayem, Debasish Chatterjee, John
Chair: Peter Hokayem
10.30 – 11.00
Stochastic Tube MPC with State Estimation
Mark Cannon, Qifeng Cheng, Basil Kouvaritakis, Saša V.
Abstract – An output feedback Model Predictive Control
(MPC) strategy for linear systems with additive stochastic
disturbances and probabilistic constraints is proposed. Given
the probability distributions of the disturbance input, the
measurement noise and the initial state estimation error, the
distributions of future realizations of the constrained variables
are predicted using the dynamics of the plant and a linear state
estimator. From these distributions, a set of deterministic
constraints are computed for the predictions of a nominal
model. The constraints are incorporated in a receding horizon
optimization of an expected quadratic cost, which is formulated
as a quadratic program. The constraints are constructed so as
to provide a guarantee of recursive feasibility, and the closed
loop system is stable in a mean-square sense.
Stable Stochastic Receding Horizon Control of Linear
Systems with Bounded Control Inputs
Peter Hokayem, Debasish Chatterjee, Federico Ramponi,
Georgios Chaloulos, John Lygeros
Abstract – We address stability of receding horizon
control for stochastic linear systems with additive noise and
bounded control authority. We construct tractable and
recursively feasible receding horizon control policies that
ensure a mean-square bounded system in closed-loop if the
noise has bounded forthorder moment, the unexcited system is
stabilizable, the system matrix A is Lyapunov stable, and there
is large enough control authority.
Information and Markov Dynamics
(Regular Session)
Room 2
Chair: Zsolt Talata
10.30 – 11.00
Finite Memory Estimation of Infinite Memory Processes
Imre Csiszár, Zsolt Talata
11.00 – 11.30
Constrained Control Design – A Simulation-Based Scenario
Maria Prandini, Marco C. Campi
Abstract – This paper deals with constrained control
design for linear systems affected by stochastic disturbances.
The goal is to optimize the control performance while
guaranteeing that the constraints are satisfied for most of the
disturbance realizations, that is with probability 1 − . In
mathematical terms, this amounts to solve a “chanceconstrained” optimization program and we introduce here a
randomized approach to this problem that builds on certain
recent results in robust convex optimization.
11.00 – 11.30
The Information Inequality on Function Spaces Given a
Singular Information Matrix
Tzvetan Ivanov, Pierre-Antoine Absil, Michel Gevers
Abstract – In this work we extend the scope of the
classical Cramér-Rao lower bound, or information inequality,
from Euclidean to function spaces. In other words we derive a
tight lower bound on the autocovariance function of a
function estimator. We do this in the context of system
identification. Two key elements of system identification are
experiment design and model selection. The novel information
inequality on function spaces is important for model selection
because it allows the user to compare estimators using different
model structures. We provide a consistent treatment of the
List of Abstracts
case where the Fisher information matrix is singular. This
makes it possible to take into account that in optimal
experiment design one tries to mask those parts of the system
non-identifiable, which are irrelevant for the application.
11.30 – 12.00
Minimum Relative Entropy State Transitions in Linear
Stochastic Systems: the Continuous Time Case
Igor G. Vladimirov, Ian R. Petersen
Abstract – This paper develops a dissipativity theory
for dynamical systems governed by linear Itô stochastic
differential equations driven by random noise with an
uncertain drift. The deviation of the noise from a standard
Wiener process in the nominal model is quantified by relative
entropy. The paper discusses a dissipation inequality for the
noise relative entropy supply. The problem of minimizing the
supply required to drive the system between given Gaussian
state distributions over a specified time horizon is considered.
This problem, known in the literature as the Schrödinger
bridge, was treated previously in the context of reciprocal
processes. The paper obtains a closedform smooth solution to a
Hamilton-Jacobi equation for the minimum required relative
entropy supply by using nonlinear algebraic techniques.
12.00 – 12.30
Deconvolution of Quantized-Input Linear Systems: analysis
via Markov Processes of a Low-Complexity Algorithm
Sophie M. Fosson, Paolo Tilli
Abstract – This paper is concerned with the problem of
the deconvolution, which consists in recovering the unknown
input of a linear system from a noisy version of the output.
The case of a system with quantized input is considered and
a low-complexity
techniques, is introduced to tackle it. The performance of such
algorithm is analytically evaluated through the Theory of
Markov Processes. In this framework, results are shown which
prove the uniqueness of an invariant probability measure of a
Markov Process, even in case of non-standard state space.
Finally, the theoretic issues are compared with simulations’
(Regular Session)
Room 3
11.00 – 11.30
Output Feedback Observers and Control under NonGaussian Types of Noise
Alexander B. Kurzhanski, Irina A. Digailova
Abstract – The
mathematical problems on systems and control emphasize
increased interest in feedback control under realistic system
output information on the basis of observations corrupted by
various types of disturbances (noise). The present text deals
with several situations of such type where the noise is confined
not only to Gaussian descriptions but also allows
nonprobabilistic interpretation. The text is restricted to
systems with original linear structure which, after being
subjected to feedback control, may turn out to become
nonlinear. Indicated here are some concise descriptions for an
array of problems and results with detailed solution versions to
appear. Beginning with description of connections between
stochastic Gaussian and set-membership bounding approaches,
it further proceeds with problems on output feedback control
under control dependent input stochastic noise additively
combined with unknown but bounded disturbances. This
produces problems under statistical uncertainty. The next item
is a discussion of observers under discrete-time measurements
that occur at random instants of time subjected to a Poisson
and measurement noise are indicated when the produced
solution is asymptotically consistent.
11.30 – 12.00
Internal Observers for Linear Systems with Time-Varying
Mustapha Ait Rami, Jens Jordan, Michael Schönlein
Abstract – This paper considers linear observed systems
with time-varying delays, where the state as well as the
observation of the state is subject to delays. It is assumed that
the delays are unknown but stay below a certain bound.
Similar to the case of uncertainties in the systems parameters
we aim to derive upper and lower estimates for the state of the
system under consideration. A pair of estimators providing
such bounds is called an interval observer. In particular, the
case where the estimators converge asymptotically is of notable
interest. In this case the interval observer is said to be
convergent. In this paper we derive necessary and sufficient
conditions for the existence of a convergent interval observer
for linear observed systems with time-varying bounded delays.
Chair: José Mario Araújo
10.30 – 11.00
Conditioned-Invariant Polyhedral Sets for Observers with
Error Limitation in Discrete-Time Descriptor Systems
José Mario Araújo, Péricles Rezende Barros, Carlos Eduardo
Trabuco Dórea
Abstract – This works aims to establish a characterization of
conditioned-invariant polyhedral sets applied in the context of
state estimation in linear discrete-time descriptor systems. It is
shown that, assuming causality, the existing conditions for
linear systems in the standard form can be extended
to descriptor ones, by rewriting the state equation in a suitable
form. To this end, a specific descriptor structure for the
observer is proposed, whereby limitation of the estimation
error can be achieved by the computation of an as small as
possible conditioned invariant polyhedron that contains the set
of possible initial errors, which is also characterized, together
with the corresponding output injection. The effectiveness of
methodology is then illustrated by numerical examples.
Algebraic Systems Theory, Behaviors, and
Codes: Stabilization and Interconnection
(Invited Session)
Room 4
Organizers: Eva Zerz, Heide Glüsing-Lürssen
Chair: Eva Zerz
10.30 – 11.00
A Behavioral Approach to Modeling Electrical Circuits
Jan C. Willems
Abstract–The aim of this contribution is to present a
hierarchical way of modeling electrical circuits.
A circuit is a device that interacts with its environment
through wires, usually called terminals. Associated with each
terminal there is a potential and a current. One can also start
from the voltage difference between the terminals, but it can be
shown that Kirchhoff's voltage law implies that the potential
description is equivalent. We view the modeling as proceeding
in two steps, first modeling an individual circuit, followed by
modeling the interconnection of circuits.
Monday, 5 July
First we outline a method of obtaining a model of an
individual circuit. Assume that the circuit consists of elements
with two-terminal ports, as R, L, C's, transformers, and
gyrators. Modeling aims at specifying the behavior of the
terminal potential/current vector from the circuit architecture
and the values of the circuit elements. We define the
architecture of the circuit as a graph with leaves, with ports in
the edges, connectors in the vertices, and external terminals in
the leaves. We show how in the linear case, this leads to a
systematic way to obtain the circuit behavior by introducing as
latent variables the potentials of the vertices and the currents
in the branches.
Next we turn to the interconnection of multi-terminal. The
architecture of a configuration is formalized as a graph with
leaves, with the vertices corresponding to subcircuits, the edges
to the connected terminals of the subcircuits, and the leaves to
external terminals. The model then follows from the behavior
of the subcircuits in the vertices and the interconnection laws
for the edges. This architecture is hierarchical since the
subcircuits can in turn be interconnected circuits in their own
11.30 – 12.00
Control of 2D Behaviors by Partial Interconnection
Paula Rocha, Diego Napp Avelli
Abstract – In this paper we study the stability of two
dimensional (2D) behaviors with two types of variables:
the variables that we are interested to control (the to-becontrolled variables) and the variables on which we are allowed
to enforce restrictions (the control variables). We derive
conditions for the stabilization of the to-be-controlled variables
by regular partial interconnection, i.e., by imposing nonredundant additional restrictions to the control variables.
12.00 – 12.30
Robustly Stable Multivariate Polynomials
Martin Scheicher
There is a striking difference between the modeling
procedure of the individual circuits and the modeling of the
interconnection of circuits. In the former case, the elements are
in the edges and the vertices correspond to the connections. In
the latter case, the elements (the subcircuits) are in the vertices
and the edges correspond to the connections.
11.00 – 11.30
The Design and Parametrization of Stabilizing Feedback
Compensators via Injective Cogenerator Quotient Signal
Ingrid Blumthaler, Ulrich Oberst
Real Algebraic Geometry and Applications – 1
(Invited Session)
Room 6
Organizers: William J. Helton, Pablo Parrilo
Chair: William J. Helton
10.30 – 11.00
Input-Output Systems Analysis Using Sum Of Squares
Programming: a Decomposition Approach
James Anderson, Antonis Papachristodoulou
Abstract – A method for estimating the L2 gain of a
nonlinear system using sum of squares (SOS) programming
and dynamical system decomposition is presented. Typically
SOS approaches to systems analysis are only computationally
tractable for systems of a modest state dimension comprising of
low order vector fields. We present a dynamical system
decomposition approach that extends the class of systems that
the L2 gain can be estimated for using SOS methods to include
those of large state dimension with high vector field degree.
11.00 – 11.30
The Grothendieck Problem With Rank Constraint
Jop Briët, Fernando Mário De Oliveira Filho, Frank Vallentin
Abstract – Finding a sparse/low rank solution in
linear/ semidefinite programming has many important
applications, e.g. combinatorial optimization, compressed
sensing, geometric embedding, sensor network localization.
Here we consider one of the most basic problems involving
semidefinite programs with rank constraints: the Grothendieck
problem with rank-k-constraint. It contains the MAX CUT
problem as a special case when k = 1. We perform a complexity
analysis of the problem by designing an approximation
List of Abstracts
algorithm which is asymptotically optimal if one assumes the
unique games conjecture.
11.30 – 12.00
Programmable Antenna Design Using Convex Optimization
Javad Lavaei, Aydin Babakhani, Ali Hajimiri, John C. Doyle
rigorous, scalable methods for identifying and characterizing
their vulnerabilities. This paper presents an approach for
analyzing the dynamics of complex networks in which the
network of interest is first abstracted to a much simpler, but
mathematically equivalent, representation, the required
analysis is performed on the abstraction, and analytic
conclusions are then mapped back to the original network and
interpreted there. We begin by identifying a broad and
important class of complex networks which admit
vulnerability-preserving, finite state abstractions, and develop
efficient algorithms for computing these abstractions. We then
propose a vulnerability analysis methodology which combines
these finite state abstractions with formal analytics from
comprehensive vulnerability analysis process for networks of
realworld scale and complexity. The potential of the proposed
approach is illustrated with a case study involving a realistic
electric power grid model and also with brief discussions of
biological and social network examples.
11.30 – 12.00
Analysis of Complex Networks
Angel Garrido
Abstract – Our paper analyzes some new lines to advance
on quickly evolving concepts, the so-called Complex Networks,
represented by graphs in general. It will be very necessary to
analyze the mutual relationship between some different
concepts and their corresponding measures, with very
interesting applications, as the case may be of Symmetry or
Entropy, and Clustering Coefficient, for example.
12.00 – 12.30
12.00 – 12.30
On the Real Multidimensional Rational K-Moment Problem
Jaka Cimprič, Murray Marshall, Tim Netzer
Cost Optimisation of Electric Power Transmission Networks
Using Steiner Tree Theory
Kevin Prendergast, Doreen Thomas
Abstract – This paper introduces a new approach to
electricity transmission network planning, which optimises the
network with respect to capital cost. The approach is limited to
three terminals at this stage, but it does illustrate the basic
building blocks of a cost optimised network of larger size.
Optimisation is achieved by means of a weighted Steiner tree
method, in which the weight of a transmission line is the per
unit length cost function. The angle between two connected
lines is found, such that for a smaller angle a star network
connecting the three terminals is more cost effective.
Graph Processes
(Regular Session)
Room 7
(Regular Session)
Room 8
Chair: Thomas P. Cason
10.30 – 11.00
Chair: Doreen Thomas
A Unified Framework for Affine Local Graph Similarity
Thomas P. Cason, Pierre-Antoine Absil, V.D. Blondel, Paul
van Dooren
10.30 – 11.00
Abstract – In this work, we review and classify
several similarity measures on undirected graphs. We show
that these measures can be rewritten in terms of fixed points of
a scaled affine transformation. Finally, we propose a novel
definition that avoids undesirable degeneracy of the similarity
11.00 – 11.30
Vulnerability Analysis for Complex Networks Using
Aggressive Abstraction
Richard Colbaugh, Kristin Glass
Abstract – Large, complex networks are ubiquitous in
nature and society, and there is great interest in developing
A New Algorithm for the Euclidean k-Bottleneck Steiner
Marcus Brazil, Charl Ras, Doreen Thomas
Abstract – We consider the problem of adding a fixed
number of relays to a WSN in order to minimise the length of
the longest transmission edge. Since routing subnetworks are
often trees we model the problem as a Euclidean k-bottleneck
Steiner tree problem (k-BSTP). We then propose a new
iterative approximation algorithm for the k-BSTP, based on an
exact solution to the 1-BSTP, and compare our heuristic (via
simulation) to the currently best performing heuristic in the
literature, namely the minimum spanning tree heuristic
(MSTH). We observe that our algorithm performs up to 8%
better than MSTH on uniformly distributed node-sets.
Monday, 5 July
11.00 – 11.30
11.00 – 11.30
Delay-Based Connectivity of Wireless Networks
Martin Haenggi
Chaos via Two-Valued Interval Maps in a Piecewise Affine
Model Example for Hysteresis
Rudolf Csikja, Barnabás M. Garay, János Tóth
Abstract – Interference in wireless networks causes
intricate dependencies between the formation of links. In
current graph models of wireless networks, where vertices
represent transceivers and edges represent links, such
dependencies are not included. In this paper we propose a
random geometric graph that explicitly captures the effect of
interference. The graph connects nodes which can
communicate with a certain maximum expected delay. We
analyze some basic properties of the graph where nodes form a
Poisson point process and use ALOHA as the channel access
11.30 – 12.00
Covering Codes and Invariant Sets
Emerson L. Monte Carmelo
Abstract – A standard piecewise affine model of hysteresis
in two dimensions is reconsidered. Periodic orbits without
selfintersection are studied and, in terms of the two real
parameters, their full bifurcation analysis is given. The main
tool is a piecewise smooth, two-valued Poincaré mapping with
only four points of discontinuities, the first return map with
respect to the line connecting the two equilibria. Earlier, singlevalued Poincaré mappings for the same model were associated
with the switching lines and had an infinite number of
discontinuities. The present paper ends with bifurcation curves
responsible for larger/smaller supports of absolutely
continuous invariant measures.
11.30 – 12.00
Temporal Logic Control for Piecewise-Affine Hybrid Systems
on Polytopes
Luc Habets, Calin Belta
12.00 – 12.30
Classification Results for Non-Mixed and Mixed Optimal
Covering Codes: a Survey
Gerzson Kéri
Abstract – A survey is given that comprises the known
classification results on non-mixed and mixed optimal
covering codes. Several new, so far unpublished results of the
author are declared and proved as well. The generality of the
specified classification results are quite different. The proof of
the more or less general new results are mostly combinatorial.
In contrast with the latter, computer-aided proofs are given for
numerous individual cases. The individual (old and new)
classification results are merged and summarized in a set of
Abstract – In this paper, a method is proposed for the
design of control laws for hybrid systems with continuous
inputs. The objective is to influence their behavior in such a
way that the discrete component of the closed-loop system
satisfies a given condition, described by a temporal logic
formula. For this purpose, a transition system is constructed,
by abstracting from the continuous dynamics of the hybrid
system. It is shown that a controller for this transition system,
realizing the given control objective, corresponds to a
controller for the original hybrid system, realizing the same
objective, and vice versa.
12.00 – 12.30
Observability Reduction of Piecewise-Affine Hybrid Systems
Mihály Petreczky, Jan H. van Schuppen
Abstract – We
observability of piecewise-affine hybrid systems. We also
propose an observability reduction algorithm for transforming
a piecewiseaffine hybrid system to a hybrid system of possibly
smaller dimension which satisfies the formulated necessary
condition for observability.
Advanced Linear Algebra – 1
(Regular Session)
Piece-Wise Affine Systems
(Regular Session)
Room 10
Room 9
Chair: Madalena Chaves
10.30 – 11.00
Transition Probabilities for Piecewise Affine Models of
Genetic Networks
Madalena Chaves, Etienne Farcot, Jean-Luc Gouzé
Abstract – In
trajectories evolve among hyperrectangles in the state space. A
qualitative description of the dynamics - useful for models of
genetic networks - can be obtained by viewing each
hyperrectangle as a node in a discrete system, so that
trajectories follow a path in a transition graph. In this paper, a
probabilistic interpretation is given for the transition between
two nodes A and B, based on the volume of the initial
conditions on hyperrectangle A whose trajectories cross to B.
In an example consisting of two intertwinned negative loops,
this probabilistic interpretation is used to predict the most
likely periodic orbit given a set of parameters, or to find
parameters such that the system yields a desired periodic orbit
with a high probability.
Chair: Rodolphe Sepulchre
10.30 – 11.00
Rank-Preserving Geometric Means of Positive Semi-Definite
Silvere Bonnabel, Rodolphe Sepulchre
Abstract–The generalization of the usual geometric mean
of two positive numbers a and b to positive definite matrices A
and B has attracted considerable attention since the seminal
work of Ando, and finds an increasing number of applications
in signal and image processing. Building upon some recent
work of the authors, the present paper proposes a
generalization of any geometric mean defined on the interior of
the cone of positive definite matrices, that is, for full rank
matrices, to a rankpreserving geometric mean defined on the
boundary of the cone, that is, for fixed-rank positive
semidefinite matrices. The work is motivated by signal
on low-rank
highdimensional spaces. The paper will discuss the reasons
why the proposed definition is sound and relevant in
List of Abstracts
11.00 – 11.30
11.00 – 11.30
Gauss Elimination without Pivoting for Positive Semidefinite
Matrices and an Application to Sum of Squares
Carla Fidalgo
Derived Cones to Reachable Sets of Semilinear Differential
Aurelian Cernea
Abstract – We consider a semilinear differential inclusion
and we prove that the reachable set of a certain variational
inclusion is a derived cone in the sense of Hestenes to the
reachable set of the semilinear differential inclusion. This
result allows to obtain a sufficient condition for local
controllability along a reference trajectory.
11.30 – 12.00
Identification of Nonlinear Processes in Microfluidic Bubble
Florinda Schembri, Francesca Sapuppo, Luigi Fortuna,
Maide Bucolo
11.30 – 12.00
Algorithm to Compute Minimal Nullspace Basis of a
Polynomial Matrix
Swanand R. Khare, Harish K. Pillai, Madhu N. Belur
Abstract – In this paper we propose a numerical algorithm
to compute the minimal nullspace basis of a univariate
polynomial matrix of arbitrary size. In order to do so a
sequence of structured matrices is obtained from the given
polynomial matrix. The nullspace of the polynomial matrix can
be computed from the nullspaces of these structured matrices.
12.00 – 12.30
On the Minimum Rank of a Generalized Matrix
Approximation Problem in the Maximum Singular Value
Kin Cheong Sou, Anders Rantzer
Abstract – In this paper theoretical results regarding a
generalized minimum rank matrix approximation problem in
the maximum singular value norm are presented. Using the
idea of projection, the considered problem can be shown to be
equivalent to a classical minimum rank matrix approximation
which can be solved efficiently using singular value
decomposition. In addition, as long as the generalized problem
is feasible, it is shown to have exactly the same optimal
objective value as that of the classical problem. Certain
comments and extensions of the presented theorem are
included in the end of the paper.
Non-Linear Dynamics
(Regular Session)
Room 11
Chair: Luigi Fortuna
Abstract – An approach based on nonlinear dynamical
systems theory is used in this work to identify the
complex temporal patterns in air bubbles flow carried by water
in a snake microfluidic channel with two inlets. Air and
water were pumped in with periodic flow. Different
experimental campaigns have been designed varying the
frequency of the flow rate alternatively for the water and for
the air and maintaining fixed the other fluid flow. Microfluidic
bubble flows were optically acquired by means of a
photodiode-based system and converted into time series. In
relation to the input control parameters (flow rate, frequency),
the diversity of bubbles’ temporal dynamic patterns was
identified through nonlinear methodologies. Relationships
between nonlinear parameters, volume fraction of fluids and
capillary number were found suggesting the chaotic behavior
of the system. This work is a fundamental step toward the
control of bubble based operations in microfluidics.
12.00 – 12.30
Interconnection of Dirac Structures and Lagrange-Dirac
Dynamical Systems
Hiroaki Yoshimura, Henry Jacobs, Jerrold E. Marsden
Abstract – In the paper, we develop an idea of
interconnection of Dirac structures and their associated
Lagrange- Dirac dynamical systems. First, we briefly review
the Lagrange- Dirac dynamical systems (namely, implicit
Lagrangian systems) associated to induced Dirac structures.
Second, we describe an idea of interconnection of Dirac
structures; namely, we show how two distinct Lagrange-Dirac
systems can be interconnected through a Dirac structure on the
product of configuration spaces. Third, we also show the
variational structure of the interconnected Lagrange-Dirac
dynamical system in the context of the Hamilton-Pontryagind’Alembert principle. Finally, we demonstrate our theory by
an example of mass-spring mechanical systems.
10.30 – 11.00
Bergson's Time and Strange Attractors
Arturo Buscarino, Luigi Fortuna, Mattia Frasca
Abstract – In this work the recurrence times of
strange chaotic attractors are studied in relation to Bergson’s
view on time. According to the thought of this philosopher,
time is a continuous flow of unique states of the consciousness.
We define quantitative parameters characterizing recurrence
times in chaotic attractors and show how these can account for
the Bergson’s concept of time.
Interpolation and Approximation in Linear
Systems – 1
(Invited Session)
Room 12
Organizers: Andrea Gombani, Martine Olivi
Chair: Andrea Gombani
10.30 – 11.00
Minimal Symmetric Darlington Synthesis: the Real Case
Laurent Baratchart, Per Enqvist, Andrea Gombani, Martine
Abstract – We consider the symmetric Darlington
synthesis of a p × p rational symmetric Schur function S with
the constraint that the extension is of size 2p × 2p and
Monday, 5 July
we investigate what happens when we impose that S and
its extension have real coefficients. In this case, under
the assumption that S is strictly contractive in at least one
point of the imaginary axis, we determine the an upper
bound for the McMillan degree of the extension. A
constructive characterization of all such extensions is provided
in terms of a symmetric realization of S and of the outer
spectral factor of Ip − SS*.
11.00 – 11.30
Convergent Rational Interpolation to Cauchy Integrals on an
Laurent Baratchart, Maxim Yattselev
Abstract – We
interpolation schemes to functions defined as Cauchy integrals
of complex densities over open analytic arcs, under
mild smoothness assumptions on the density. The
interpolation points must be chosen according to the geometry
of the arc, and the convergence is locally uniform outside of
the arc. The result essentially settles the convergence issue
of multipoint Padé approximants to functions with
connected singular set of dimension 1.
In this approach, one works in the white noise space (or in the
Kondratiev space, which is an inductive limit of Hilbert
spaces which contains it) and replaces the usual product by the
Wick product. Most, if not all, of classical system theory
extends to this setting. In the talk we review this setting, and
present some new results on stochastic linear systems.
11.00 – 11.30
Discrete Multiscale Systems: Stability Results
Daniel Alpay, Mamadou Mboup
Abstract – We introduce discrete time-scale filtering by
the way of certain double convolution systems. We prove
stability theorems for these systems and make connections with
function theory in the poly-disc.We also make connections with
the white noise space framework.
11.30 – 12.00
Stochastic integration for a Class of Gaussian Processes
Daniel Alpay, Haim Attia, David Levanony
11.30 – 12.00
Approximative Covariance Interpolation
Per Enqvist
Abstract – When methods of moments are used for
identification of power spectral densities, a model is matched to
estimated second order statistics such as, e.g., covariance
estimates. If the estimates are good there is an infinite family of
power spectra consistent with such an estimate and in
applications, such as identification, we want to single out the
most representative spectrum. We choose a prior spectral
density to represent a priori information, and the spectrum
closest to it in a given quasi-distance is determined. However, if
the estimates are based on few data, or the model class
considered is not consistent with the process considered, it may
be necessary to use an approximative covariance interpolation.
Two different types of regularizations are considered in this
paper that can be applied on many covariance interpolation
based estimation methods.
12.00 – 12.30
LMI Conditions of Strictly Bounded Realness on a StateSpace Realization to Bi-Tangential Rational Interpolation
Yohei Kuroiwa
Abstract – We present LMI conditions to characterize
the strictly bounded realness of the state-space realization of
the solution to the bi-tangential rational interpolation
problem with McMillan degree constraint.
Linear Stochastic Systems, the White Noise
Space, and Related Topics
(Invited Session)
Room 13
Organizer: Daniel Alpay
Chair: Daniel Alpay
10.30 – 11.00
An Introduction to White Noise Theory and Linear
Stochastic Systems
Daniel Alpay, David Levanony
Abstract – In a recent paper which appeared in Acta
Applicandae Mathematicae, the authors developed a new
approach to linear stochastic systems and proved a number of
stability theorems for linear time invariant stochastic systems.
12.00 – 12.30
On a Schur Class of Functions whose Values are Operators
between Banach Spaces
Dan Volok
Abstract – In this talk we shall introduce a Schur class of
functions whose values are operators between Banach spaces
and present a characterization of the related de Branges Rovnyak spaces. This is a joint work with D. Alpay, O.
Timoshenko and P. Vegulla.
LDPC and Applications
(Invited Session)
Room 14
Organizers: Joachim Rosenthal, Marcus Greferath
Chair: Joachim Rosenthal
10.30 – 11.00
Exploration of AWGNC and BSC Pseudocodeword
Jens Zumbrägel, Mark F. Flanagan, Vitaly Skachek
List of Abstracts
11.00 – 11.30
11.00 – 11.30
Cross-Correlation of Costas Arrays: the Current Status
Konstantinos Drakakis, Scott Rickard
A New Approach to Strong Practical Stability and
Stabilization of Discrete Linear Repetitive Processes
Pawel Dabkowski, Krzysztof Gałkowski, Olivier Bachelier,
Eric Rogers, James Lam
Abstract – We study the cross-correlation of pairs of
Costas arrays, and more specifically its maximal value over the
families of Golomb and Welch permutations. We record the
numerical results found, analyze them, formulate conjectures
summarizing our findings, and present the current progress
towards a rigorous proof of these conjectures.
11.30 – 12.00
LDPC Codes from Matrix Equations
Ariel Amir, Abigail Mitchell, Joachim Rosenthal
Abstract – Different constructions of LDPC codes based
on matrix equations are investigated. The parameters such as
the dimension, rate and distance are computed. The classical
Tanner graph representation known for LDPC codes are
described. The main difference between standard LDPC codes
and the LDPC codes based on matrix equations lies in the
structure of their codewords. Whereas in the classical situation
codewords are simply vectors, the codewords in this new
setting will be two-dimensional vectors or matrices. This
implies that the parity-check constraints must be satisfied in
both perpendicular directions of the codeword. Therefore, a
codeword may be interpreted as a two-dimensional array
which is suitable for recording on two-dimensional patternoriented storage media.
12.00 – 12.30
On Binary Self-Dual Extremal Codes
Wolfgang Willems
Abstract – There is a large gap between Zhang’s
theoretical bound for the length n of a binary extremal selfdual doublyeven code and what we can construct. The largest n
is 136. In order to find examples for larger n a non-trivial
automorphism group might be helpful. In the list of known
examples extended quadratic residue codes and quadratic
double circulant codes have large automorphism groups. But
in both classes the extremal ones are all known. They are
exactly those which are in the list; hence of small length. The
investigations we have done so far give some evidence that for
larger n the automorphism group of a putative extremal selfdual doubly-even code may be very small, if not trivial. Thus
the code merely seems to be a big combinatorial object and
therefore possibly hard to find.
2D Systems
(Invited Session)
Room 15
Organizers: Marek Majewski, Dariusz Idczak
Chair: Marek Majewski
10.30 – 11.00
Positive Switched 2D Linear Systems Described by the
Roesser Models
Tadeusz Kaczorek
Abstract – In this paper the positive switched 2D linear
system described by the Roesser models will be considered. We
shall analyzed the following question: When is a positive
switched 2D linear system defined by a linear Roesser models
and a rule describing the switching between them
asymptotically stable. It is well known [1, 3] that a necessary
and sufficient conditions for stability under arbitrary switching
is the existence of a common Lyapunov function for the family
of subsystems. This result will be extended for positive
switched 2D linear systems described by the Roesser models.
Abstract – Repetitive processes are a distinct class of
2D systems of both theoretical and practical interest. The
stability theory for these processes originally consisted of two
distinct concepts termed asymptotic stability and stability
along the pass respectively where the former is a necessary
condition for the latter. Recently applications have arisen
where asymptotic stability is too weak and stability along the
pass is too strong for meaningful progress to be made. This, in
turn, has led to the concept of strong practical stability for such
cases, where previous work has formulated this property and
obtained necessary and sufficient conditions for its existence
together with Linear Matrix Inequality (LMI) based tests,
which then extend to allow control law design. This paper
develops considerably simpler, and hence computationally
more efficient, stability tests that also extend to allow control
law design.
11.30 – 12.00
2D Systems with Controls and Some their Applications
Dorota Bors, Stanisław Walczak
Abstract – We consider the 2D continuous counterpart
of Marchesini-Fornasini model of the process of gas
filtration. The continuous version of the discrete model
constitutes the hyperbolic boundary value problem. Our main
result is finding sufficient conditions for the existence of an
optimal solution for the process of gas filtration minimizing the
cost functional.
12.00 – 12.30
Fractional Differential Repetitive Processes
Dariusz Idczak, Rafał Kamocki
Monday, 5 July
Semiplenary Lecture
Semiplenary Lecture
Room 1
Room 15
Chair: Floyd B. Hanson
Chair: Peter E. Caines
14.00 – 15.00
14.00 – 15.00
Modelling High Dimensional Time Series by Generalized
Factor Models
Manfred Deistler
(joint work with B.D.O. Anderson, A. Filler, W. Chen)
Some Problems with Connecting Renewable Energy Sources
to the Grid
George Weiss, Qing-Chang Zhong
Abstract – We discuss and analyze generalized linear
dynamic factor models. These models have been developed
recently and they are used to model high dimensional time
series in order to overcome the “curse of dimensionality”. The
basic idea in factor models is to seperate “comovement”
between the variables (caused by a relatively small number of
factors) from individual (idiosyncratic) variation. Here factor
analysis is considered in a time series context, where
concentration of information is performed in the crosssectional and in the time dimension. The models considered are
linear dynamic in nature and stationarity of the processes is
assumed. As opposed to the classical case, in the generalized
case considered here, a certain form of weak dependence of the
noise components is permitted. In the core part of the paper,
we are concerned with structure theory, namely with realizing
the singular rational spectral density of the latent variables by
a linear system. Special emphasis ! is laid on the autoregressive
case, which is generic in our setting. These autoregressions may
have a singular innovation variance, which may cause multiple
solutions for the Yule Walker equations. Finally, identification
procedures, using a suitable denoising procedure and
estimators suggested by our structure theory, are discussed.
Abstract – In this paper, we review some challenges
resulting from the grid connection of powerful renewable
energy generators that produce randomly fluctuating power
and have no mechanical inertia. We propose and develop the
idea of operating an inverter to mimic a synchronous
generator. We call the inverters that are operated in this way
synchronverters. Using synchronverters, the well-established
theory/algorithms used to control synchronous generators can
still be used in power systems where a significant proportion of
the generating capacity is inverter-based. We describe the
dynamics, implementation and operation of synchronverters.
The real and reactive power delivered by synchronverters
connected in parallel and operated as generators can be
automatically shared using the well-known frequency and
voltage drooping mechanisms. Synchronverters can be easily
operated also in island mode and hence they provide an ideal
solution for microgrids or smart grids.
Continuous-Time Model Identification
(Invited Session)
Room 1
Organizer: Toshiharu Sugie
Chair: Toshiharu Sugie
Semiplenary Lecture
Room 14
Chair: Yutaka Yamamoto
14.00 – 15.00
Smith-Predictor Type Structure for a Class of InfiniteDimensional Systems: Optimal Control and Performance
Limitation Formula
Kenji Kashima
Abstract – In this talk we investigate control problems
for infinite-dimensional systems whose transfer matrices are
expressible in terms of a rational transfer matrix and a
scalar (possibly irrational) inner function. This class of systems
is capable of describing many practical control problems,
when weighting functions are rational and plants have at most
a finite number of unstable modes or zeros. In the first half of
this talk the concept of Smith-predictors, that was originally
used for I/O delay systems, is extended to the aforementioned
class of systems. This allows us to reduce the optimal control
problems to easily checkable criteria that do not require the
solution of operator-valued equations. Furthermore, the
obtained (stabilizing or suboptimal) controllers are shown to
have the structure of Smith-predictors, or their dual. In the
second half of the talk we derive a new expression for the H²
performance limit, based on state-space representation. The
resulting formula, given as a functional of the inner function,
helps us to understand how achievable H² performance
deteriorates due to the plant’s nonminimum phase properties
or unstable modes. The example of a linear quantum control
system suffering from feedback delay is given to illustrate the
15.30 – 16.00
Continuous-Time Model Identification and State Estimation
Using Non-Uniformly Sampled Data
Rolf Johansson
Abstract – This contribution reviews theory, algorithms, and
validation results for system identification of continuous-time
state-space models from finite inputoutput sequences. The
algorithms developed are autoregressive methods, methods of
subspace-based model identification and stochastic realization
adapted to the continuous-time context. The resulting model
can be decomposed into an input-output model and a
stochastic innovations model. Using the Riccati equation,
we have designed a procedure to provide a reducedorder stochastic model that is minimal with respect to
system order as well as the number of stochastic inputs,
thereby avoiding several problems appearing in standard
application of
validation problem. Next, theory, algorithms and validation
results are presented for system identification of continuoustime state-space
sampled input-output sequences. The algorithms developed
are methods of model identification and stochastic
realization adapted to the continuous-time model context
using non-uniformly sampled input-output data. The
resulting model can be decomposed into an input-output
model and a stochastic innovations model. For state
estimation dynamics and Kalman filters, we have designed a
procedure to provide separate continuous-time temporal
update and error feedback update based on nonuniformly sampled input-output data.
List of Abstracts
16.00 – 16.30
Frequency Domain Total Least Squares Estimator of TimeVarying Systems
John Lataire, Rik Pintelon
Abstract – An
continuoustime, time-varying systems is presented. The model
considered is an ordinary differential equation whose
coefficients are polynomials in time. The model equation is
evaluated in the frequency domain (thus allowing a simple
selection of the frequency band of interest) from sampled, finite
length records of the input and output signals. The timefrequency transformations are performed using the Discrete
Fourier Transform and its inverse. The leakage and alias
errors (due to the nonperiodicity of the system’s response) are
shown to be easily captured by adding a polynomial to the
model equation. The identification procedure is formulated as
a total least squares estimation problem. The estimator is
illustrated on a simulation example.
16.30 – 17.00
EM Identification of Continuous–Time State Space Models
from Fast Sampled Data
Juan I. Yuz, Jared Alfaro, Juan C. Agüero, Graham C.
proposed in our ECC 09 paper. Resetting takes place if the
estimator process hits the boundary of a pre-specified compact
domain, or if the rate of change, in a stochastic sense, of the
parameter process would hit a fixed threshold. An outline of a
proof of convergence almost surely and in Lq was given, under
realistic conditions. In the present paper we show that the
RML estimator differs from the off-line estimator by an error
of the magnitude of logT/T in an appropriate sense. With this
result a conjecture formulated back in 1984 has been settled.
16.30 – 17.00
Mean-Square Minimization in Mathematical Finance with
Control and State Constraints
Andrew J. Heunis
Abstract – We study a problem of optimal stochastic control
from mathematical finance. The problem involves both a
control constraint (on the portfolio) together with an
almostsure state constraint (on the wealth process), giving a
rather challenging combination of constraints. We demonstrate
existence of a Lagrange multiplier, show that this is a pair
comprising a finitely additive measure (for the state constraint)
and an Ito process (for the portfolio constraint), and construct
an optimal portfolio in terms of the Lagrange multiplier.
17.00 – 17.30
Abstract – In this paper we apply the ExpectationMaximization (EM) algorithm to the identification
of continuous-time state-space models from fast sampled data.
We modify the standard EM formulation, using a
parametrization of the sampled-data model in incremental
form. This model recovers the underlying continuous-time
system when the sampling period goes to zero. Also, the use of
the incremental model parametrization shows better numerical
behavior for fast sampling rates. We also consider the case of
non–uniform sampling and a robust identification procedure
that can be applied in the time or frequency domain.
Convergence of the MAP Path Estimator in HMMs with
Continuous Hidden State Space
Pavel Chigansky, Ya’acov Ritov
Filtering, Estimation and Control
(Regular Session)
Room 2
Chair: Lorenzo Finesso
15.30 – 16.00
Two-Step Nonnegative Matrix Factorization Algorithm for
the Approximate Realization of Hidden Markov Models
Lorenzo Finesso, Angela Grassi, Peter Spreij
Abstract – We propose a two-step algorithm for the
construction of a Hidden Markov Model (HMM) of assigned
size, i.e. cardinality of the state space of the underlying Markov
chain, whose n-dimensional distribution is closest in divergence
to a given distribution. The algorithm is based on the
factorization of a pseudo Hankel matrix, defined in terms of
the given distribution, into the product of a tall and a wide
nonnegative matrix. The implementation is based on the
nonnegative matrix factorization (NMF) algorithm. To
evaluate the performance of our algorithm we produced some
numerical simulations in the context of HMM order reduction.
16.00 – 16.30
Recursive Identification of Continuous-Time Linear
Stochastic Systems - An Off-Line Approximation
László Gerencsér, Vilmos Prokaj
Abstract – We consider multi-variable continuous-time
linear stochastic systems given in innovation form, with
system matrices depending on an unknown parameter that is
locally identifiable. A computable continuous-time recursive
maximum likelihood (RML) method with resetting has been
PDE Systems
(Regular Session)
Room 3
Chair: Joao R. Branco
15.30 – 16.00
Integro-Differential IBVP versus Differential IBVP: Stability
Joao R. Branco, José Augusto Ferreira
Abstract – The aim of this work is the qualitative analysis
from theoretical and numerical points of view of an integrodifferential initial boundary value problem where the reaction
term presents a certain memory efect. Stability results are
established in both cases. As in certain cases the integrodifferential initial boundary value problem can be seen as a
Monday, 5 July
differential initial boundary value problem, the results
obtained for the integro-differential formulation are compared
with the correspondent results stated for the diferential initial
boundary value problem. Numerical results illustraing the
theoretical results are also presented.
16.00 – 16.30
On the Weight Hierarchy of Certain Grassmann Codes
Arunkumar R. Patil, Harish K. Pillai
16.00 – 16.30
Distributed Source Identification for Wave Equations: an
Observer-Based Approach
Marianne Chapouly, Mazyar Mirrahimi
Abstract – In this paper, we consider a wave equation on
a bounded interval where the initial conditions are known
(are zero) and we are rather interested in identifying an
unknown source term q(x) thanks to the measurement output y
which is the Neumann derivative on one of the boundaries. We
use a back and forth iterative procedure and construct wellchosen observers which allow to retrieve q from y in the
minimal observation time.
16.30 – 17.00
16.30 – 17.00
Block Preconditioned Methods in Solution of Hyperbolic
Ahmad Shayganmanesh (Golbabai), M.M. Arabshahi
The Shortest-Basis Approach to Minimal Realizations of
Linear Systems
G. David Forney Jr.
17.00 – 17.30
The Best State Space for the SCOLE Model
Xiaowei Zhao, George Weiss
Abstract – It is well-known that the SCOLE model (a
beam coupled to a rigid body) is not exactly controllable in the
energy state space with L2 input signals, since the control
operator is compact from the input space to state space. In this
paper, we derive its exactly controllable space for L2 input
signals and we prove its well-posedness and regularity in this
Algebraic Systems Theory, Behaviors, and
Codes: New Developments Beyond Classical
Algebraic Coding Theory
(Invited Session)
17.00 – 17.30
Correlations in Stream Ciphers: a Systems Theory Point of
Sara Diaz Cardell, Gerard Maze, Joachim Rosenthal, Urs
Abstract – Given a sequence of some autonomous
behavior, this sequence can be computed as the output of a
linear system. If one receives a highly noisy sequence
correlated with such a linear sequence, the problem we study is
how to obtain the input of the linear system. We explain known
correlation attacks in this general setting and we show types of
autonomous behaviors which should be avoided.
Room 4
Organizers: Heide Glüsing-Lürssen, Eva Zerz
Chair: Heide Glüsing-Lürssen
Systems Theory and the Economics of Pricing in
New Markets
(Invited Session)
15.30 – 16.00
New Improvements on the Echelon-Ferrers Construction
Anna-Lena Trautmann, Joachim Rosenthal
Abstract – We show how to improve the echelonFerrers construction of random network codes introduced in
[3] to attain codes of larger size for a given minimum distance.
Room 6
Organizers: Robert Martin, Clyde F. Martin
Chair: Robert Martin
15.30 – 16.00
Systems and Markets: Instability and Irrationality
Robert Martin, Clyde F. Martin, Xuyao Lin
Abstract – New financial products are difficult to price.
Often the products suffer through an initial period of price
List of Abstracts
volatility as the market searchers for an equilibrium value. In
this paper, we extend our previous results by showing that the
rate of convergence is extremely slow, less than sqrt(N), where
N is the number of draws. As a result with a sufficiently low
discount rate, an investor gambling on draws from the urn
faces possibility of time-relevant unbounded losses. Why do
these losses fail to prevent the market from forming? Are
investors then inherently irrational? Because the market price
is an average of the beliefs of many investors rather than a
reflection of the truth. Therefore, any investor who believes
their priors better than the market's also believes they have
unbounded earning potential. As a result, the market thrives,
volatility persists, and some investors win at the expense of
16.00 – 16.30
A General Theory of Markovian Time Inconsistent Stochastic
Control Problems
Tomas Björk, Agatha Murgoci
Abstract–We develop a theory for stochastic control problems
which, in various ways, are time inconsistent in the sense that
they do not admit a Bellman optimality principle. We attach
these problems by viewing them within a game theoretic
framework, and we look for Nash subgame perfect equilibrium
points. For a general controlled Markov process and a fairly
general objective functional we derive an extension of the
standard Hamilton-Jacobi-Bellman equation, in the form of a
system of non-linear equations, for the determination for the
equilibrium strategy as well as the equilibrium value function.
All known examples of time inconsistency in the literature are
easily seen to be special cases of the present theory. We also
prove that for every time inconsistent problem, there exists an
associated time consistent problem such that the optimal
control and the optimal value function for the consistent
problem coincides with the equilibrium control and value
function respectively for the time inconsistent problem. We
also study some concrete examples.
16.30 – 17.00
A Model for Multiscaling and Clustering of Volatility in
Financial Indexes
Alessandro Andreoli, Francesco Caravenna, Paolo Dai Pra,
Gustavo Posta
Abstract – We propose a stochastic model which matches
some relevant stylized facts observed in time series of financial
indexes, and that are not fully captured by the models most
often used in this context. These stylized facts concern with the
distribution of the log-returns (increments of the logarithm of
the index). This distribution is not Gaussian, and its moments
obey peculiar scaling relations (multiscaling). Moreover,
absolute values of log-returns in disjoint time intervals are
positively correlated (clustering of volatility): their correlation
has slow (sub-exponential) decay for moderate time distances
(up to few months), and have a faster decay for larger
distances. The simplicity of the model allows sharp analytic
results, statistical estimation of its few parameters, and
low computational effort in simulations, allowing its
concrete use in applications such as option pricing.
options in a bivariate factor model which can be easily
computed numerically by calculating four line integrals.
Biological Networks
(Regular Session)
Room 7
Chair: Siamak Taati
15.30 – 16.00
Gene Regulatory Networks: the Impact of Distance between
Gilles Bernot, Jean-Paul Comet, Enrico Formenti, Siamak
Abstract – We analyze the basic building block of gene
regulation networks using a simple stochastic model. We
consider a network consisting only of two interacting genes: an
activator (or repressor) gene that produces proteins of type S
and a target gene that is activated (or repressed, respectively)
by proteins of type S. We identify the role of distance
between the two interacting genes by calculating the relative
density of those activator proteins that until time t have
succeeded in reaching the vicinity of the target gene via an
unbiased threedimensional Brownian motion. The latter
quantity seen as a function of time has a sigmoidal shape (like a
simple delay line) that is sharper and taller when the two genes
are closer to each other. This suggests an evolutionary pressure
towards making the interacting genes closer to each other to
make their interactions more efficient and more reliable.
16.00 – 16.30
Controlling Gene Regulatory Networks by Means of Control
Systems Theory Principles and Microfluidic Devices
Filippo Menolascina, Mario di Bernardo, Diego di Bernardo
16.30 – 17.00
Estimation of Protein Networks for Cell Cycle in Yeast Based
on Least–Squares Method Using Periodic Signals
Noriko Takahashi, Takehito Azuma, Shuichi Adachi
Abstract – In this
approach to
estimation problems of protein networks is proposed for
systems biology. Generally, it is difficult to estimate
complicated networks in molecular biology. Then, in order to
estimate complicated networks systematically, it is considered
to estimate the networks based on a control engineering
method. Considering that wave patterns of proteins are
periodic, the protein networks are estimated by the least–
squares estimation method. In this method, the networks can
be estimated by using just 1 cycle data of protein
concentrations. Moreover, this method is applied to an
estimation problem of protein networks for cell cycle in yeast,
and 9–dimensional protein networks are actually estimated.
Signal Processing
(Regular Session)
Room 8
17.00 – 17.30
Chair: Kenadall Gillies
Arbitrage-Free Multifactor Term Structure Models: a Theory
Based on Stochastic Control
Andrea Gombani, Wolfgang J. Runggaldier
15.30 – 16.00
Abstract – We present an alternative approach to the
pricing of bonds and bond derivatives in a multivariate linearquadratic model for the term structure that is based on the
solution of a linear-quadratic stochastic control problem. We
focus on explicit formulas for the computation of bond
Convergence of Bayesian Posterior Distributions
Kenadall Gillies, Robert Martin, Shan Sun, Clyde F. Martin
Abstract–This paper will compare the model of determining
the true price of a product to determining the proportion of
black balls in a bottomless, rotating urn. In doing this it is seen
that as long as the ratio of black balls converge to the true
proportion, the Bayesian updating method for the rth moment
will converge to the true proportion raised to the rth power.
Monday, 5 July
Thus the posterior distributions converge to unit mass at the
true proportion. A theory for speeding up the convergence rate
for the Bayesian updating method was then tested since it now
known that the Bayesian updating method converges.
16.00 – 16.30
High Resolution Analysis via Sparsity-Inducing Techniques:
Spectral Lines in Colored Noise
Lipeng Ning, Tryphon T. Georgiou, Allen R. Tannenbaum
Abstract – The impact of sparsity-inducing techniques
in signal analysis has been recognized for over ten years
now and has been the key to a growing literature on the
subject– commonly referred to as compressive sensing. The
purpose of the present work is to explore such sparsityinducing techniques in the context of system identification.
More specifically we consider the problem of separating
sinusoids in colored noise while at the same time identifying the
dynamics that generate the wide-bandwidth noise-component.
Our formalism relies on modeling the data as a superposition
of a few unknown sinusoidal signals together with the output of
an auto-regressive filter which is driven by white noise.
Naturally, since neither the underlying dynamics nor any
possible sinusoids present are known, the problem is ill-posed.
We seek a sparse selection of sinusoids which together with the
auto-regressive component can account for the data-set and, to
this end, we propose a suitable modification of sparsityinducing functionals (a la LASSO/Basis pursuit/etc.) which can
generate admissible solutions-their sparsity being determined
by tuning parameters.
16.30 – 17.00
Probabilistic Current-State Opacity is Undecidable
Anooshiravan Saboori, Christoforos N. Hadjicostis
Abstract – Increasing concerns about security and privacy in
applications of discrete event systems have led to
various notions of opacity for systems that are modeled as
(possibly non-deterministic) finite automata with partial
observation on their transitions. Specifically, a system is
current-state opaque if the entrance of the system state to a
given set of secret states remains opaque (uncertain), until the
system leaves this set of secret states, to an intruder who
observes system activity through some projection map. While
this notion has been shown useful for security requirements in
many applications (including encryption using pseudo-random
generators and coverage properties in sensor networks), it does
not provide a quantifiable measure for characterizing the
security of a given system. In this paper, we extend this
framework to systems that can be modeled as probabilistic
finite automata, obtaining in the process the probability of
observing sequences of observations that violate current-state
opacity. We then introduce and analyze the notion of
probabilistic current-state opacity which can be used to
provide a measure of a given system’s opacity. We show that
verifying probabilistic currentstate opacity is undecidable in
general, though it can become decidable in specific settings.
Switched Systems
(Regular Session)
Room 9
Chair: Koichi Kobayashi
15.30 – 16.00
Free-Variable Analysis of Finite Automata Representations
for Hybrid Systems Control
Koichi Kobayashi, Jun-ichi Imura
Abstract – As is well known, the computational complexity in
the mixed integer programming (MIP) problem is one of the
main issues in model predictive control of hybrid systems such
as mixed logical dynamical systems. To overcome this issue, the
authors have proposed a new method to represent
a deterministic finite automaton as a linear state equation with
a relatively smaller number of (free) binary input variables,
which thus makes the number of binary variables in the
resultant MIP problem smaller. This paper continues upon the
above approach, and presents theoretical aspects on input
binary variables in the linear state equation model such as the
upper bound of the number of the binary input variables.
16.00 – 16.30
Disturbance Decoupling Problems with Quadratic Stability
for Switched Linear Systems via State Feedback
Naohisa Otsuka
Abstract – In
problems without stability and with quadratic stability for
switched linear systems are formulated in the framework of the
so-called geometric approach. Firstly, necessary and sufficient
conditions for the problem without stability to be solvable are
given. Secondly, sufficient conditions for the problem with
quadratic stability to be solvable are given. Further, for
switched linear systems composed of two subsystems necessary
and sufficient conditions for the problem with quadratic
stability to be solvable are also investigated. Finally, an
illustrative example is shown.
16.30 – 17.00
Output Feedback Stabilizing Control and Passification of
Switching Diffusion Systems
Pavel V. Pakshin, Dimitri Peaucelle
Abstract – A parametric description of static output
feedback stabililizing controllers for diffusion systems with
Markovian switching is presented. This description is
expressed in terms of coupled linear matrix equations and nonconvex quadratic matrix inequalities which depend on
parameter matrices similar to weight matrices in LQR theory.
A convexifying approximation technique is proposed to obtain
the LMI-based algorithms for computing of the gain matrix.
These are non-iterative and used computationally efficient SDP
solvers.The results are then applied to simultaneous
stabilization of a set of diffusion systems, robust stabilization
and stochastic passification problems. Finally, a numerical
example is provided to demonstrate the applicability and
effectiveness of the proposed method.
17.00 – 17.30
On Well-Posedness of Piecewise Affine Bimodal Dynamical
Le Q. Thuan, Kanat Camlibel
Abstract – The theory of differential inclusions provides
certain sufficient conditons for the uniqueness of Filippov
solutions such as one-sided Lipschitzian property or maximal
monotone condition. When applied to piecewise affine
dynamical systems, these conditions impose rather strong
conditions. In this paper, we provide less restrictive conditions
for uniqueness of Filippov solutions for the bimodal piecewise
affine systems.
List of Abstracts
Advanced Linear Algebra – 2
(Regular Session)
Room 10
Chair: Christophe Fonte
The original problem is restated as an eigenvalue minimization
problem and further transformed into a semidefinite
programming problem. We then relax our problem by
imposing a sum-of-squares constraint. Experimental results
show that our algorithm can obtain good results in a much
smaller computation time than the existing algorithm.
15.30 – 16.00
Conditions for Interpolation of Stable Polynomials
Christophe Fonte, Cédric Delattre
Abstract – This contribution addresses the problem of
the interpolation of a set of positive numbers by stable
real polynomials. It is shown that the interpolant preserves
local positivity, monotonicity, and convexity in order to satisfy
stability requirement of the interpolating polynomial. Then this
issue is formulated as a nonlinear system carrying on the
existence of negative real roots and positive real parameters.
By considering an extension of the Farkas’s Lemma and the
method of Fourier- Motzkin elimination, conditions are
explicitly produced for the existence of an Hurwitz polynomial
that passes through all the pairs of values to interpolate.
16.00 – 16.30
Convexity of Higher Dimensional Numerical Ranges
Michael Karow
New Results on Computation and Control
(Invited Session)
Room 11
Organizer: Bijoy K. Ghosh
Chair: Bijoy K. Ghosh
15.30 – 16.00
Periodic and Recursive Control Theoretic Smoothing Splines
Maja Karasalo, Xiaoming Hu, Clyde F. Martin
Abstract – In this paper, a recursive control theoretic
smoothing spline approach is proposed for reconstructing a
closed contour. Periodic splines are generated by minimizing a
cost function subject to constraints imposed by a linear
control system. The optimal control problem is shown to be
proper, and sufficient optimality conditions are derived for a
special case of the problem using Hamilton-Jacobi-Bellman
The filtering effect of the smoothing splines allows for
usage of noisy sensor data. An important feature of the method
is that several data sets for the same closed contour can be
processed recursively so that the accuracy is improved stepwise
as new data becomes available.
16.00 – 16.30
Smoothing Splines on the Torus
F. Egebrand, Magnus Egerstedt, Clyde F. Martin
Abstract – An
constructing smoothing splines on the torus. The algorithm
uses a particular representation of the torus and is
suboptimal. However it produces good trajectories and among
all possible smoothing spline it does very well in terms of the
cost function.
16.30 – 17.00
16.30 – 17.00
Real Radius of Controllability for the Systems Described by
Polynomial Matrices: SIMO Case
Swanand R. Khare, Harish K. Pillai, Madhu N. Belur
Abstract – In this paper we discuss the problem of
computing the real radius of controllability of the Single Input
Multi Output (SIMO) systems described by univariate
polynomial matrices. The problem is equivalent to computing
the nearest noncoprime polynomial matrix to the polynomial
matrix describing the system in some prescribed norm. A
particular case of this problem is to compute approximate
GCD of univariate polynomials. Further this problem is shown
to be equivalent to the Structured Low Rank Approximation
(SLRA) of a linearly structured resultant matrix associated
with the given polynomial matrix. The radius of controllability
is then computed by finding the nearest SLRA of this resultant
17.00 – 17.30
Computing the Controllability Radius for Higher Order
Systems Using Semidefinite Programming
Bogdan C. Şicleru, Bogdan Dumitrescu
Abstract – We propose here a new approach for
computing the controllability radius for higher order systems.
Control Aspects of a Finite Length Hubbard Chain
F. Assaad, Gunther Dirr, F. Goth, Uwe Helmke
Abstract – Solid state physics provides a rather new
application area for quantum control. Yet, the high
dimensional state spaces in this field require a thorough
analysis of the underlying Lie algebraic structures for
developing efficient control strategies. In this paper, we focus a
one-dimensional chain of quantum dots described by the
Hubbard model. The model plays an extremely important role
in solid state physics since it is the simplest model which
explicitly takes into account the interplay between Coulomb
repulsion of electrons and their kinetic energy. We begin with
an comprehensive description to the mathemtical tools need for
the appropriate state space construction. Based on these
concepts, we introduce the general Hubbard Hamiltonian and
discuss some aspects of is general Lie algebraic structure.
Finally, we present a toy example for illustrating the afore
established notions.
The tutorial part of this work is supposed to enhance the
collaboration between
mathematicians in the area of quantum control.
Monday, 5 July
17.00 – 17.30
16.30 – 17.00
To Tilt Your Head or Not To: Potentially
Bijoy K. Ghosh, Indika B. Wijayasinghe
On Tangential Matrix Interpolation
Paul A. Fuhrmann
Abstract – In this paper we study the human head
movement, when the head shifts its orientation between two
possible heading directions, as a simple mechanical control
system. Head movements obey Donders’ constraint (as opposed
to the Listing’s constraint for eye movement), which states that
the allowed orientations of the head are obtained by rotating a
fixed ‘primary heading direction’ by a subclass of rotation
matrices. These rotation matrices have their axes of rotation
restricted to a fixed surface, called the Donders’ surface.
Donders’ Law states that when head moves spontaneously
from left to right and back or from top to bottom and back, the
head rotation matrix has no torsional component. On the other
hand, when the head moves diagonally from the top left to the
bottom right and back or from the top right to the bottom left
and back, spontaneously, the axis of rotation has a small
torsional component. The torsional component effectively
rotates the head with respect to the frontal line of ‘heading’.
The head appears slightly tilted as a result of the slight torsion.
Defining a suitable Riemannian metric, we study dynamic
control of head movement when the head orientations satisfy
the Donders’ constraint throughout its entire trajectory. Head
movements are actuated by choosing a suitable potential
function and the oscillations are damped by adding a suitable
damping term. An important result of this paper is to show the
effect of the torsional component as head is allowed to move
between two ‘headings’. We show that when the peak value of
the allowable torsion is high, i.e. when the head is allowed to be
more tilted, transition time between two headings is shortened.
Abstract – The talk will present an algebraic approach, using
polynomial and rational mod- els over an arbitrary field, to
tangential interpolation problems, both by polynomial as well
as rational functions. Appropriate extensions of scalar
problems, associated with the names of Lagrange (first order),
Hermite (high order) and Newton (recur- sive) are derived.
The relation of tangential matrix interpolation problems to
the matrix Chinese remainder theorem are clarified. Some two
sided interpolation prob- lems are dealt using the theory of
tensored models. The polynomial results are then used as a
basis for the solution of the corresponding problems of
interpolation by rational matrix functions. We shall also try to
explain the role of tensored models in the study of the
polynomial version of the Sylvester equation and its
connection to model reduction by interpolation.
Interpolation and Approximation in Linear
Systems – 2
(Invited Session)
17.00 – 17.30
Polynomial Structure of 3 x 3 Reciprocal Inner Matrices
David Avanessoff, Martine Olivi, Fabien Seyfert
Abstract – The objective of our work is the derivation
of efficient algorithms for the synthesis of microwave
multiplexers. In our opinion, an efficient frequency design
process calls for the understanding of the structure of n x n
inner (or lossless) reciprocal rational functions for n > 2.
Whereas the case n = 2 is completely understood and a
keystone of filter synthesis very little seems to be known about
the polynomial structure of such matrices when they involve
more than 2 ports. We therefore start with the analysis of the 3
x 3 case typically of practical use in the manufacturing of
diplexers. Based on recent results obtained on minimal degree
reciprocal Darlington synthesis [6], we derive a polynomial
model for 3 x 3 reciprocal inner rational matrices with given
McMillan degree.
Room 12
Organizers: Andrea Gombani, Martine Olivi
Chair: Martine Olivi
Passive Network Synthesis
(Invited Session)
15.30 – 16.00
Parametrization of Matrix-Valued Lossless Functions Based
on Boundary Interpolation
Ralf Peeters, Martine Olivi, Bernard Hanzon
Abstract – This
parametrization issues for rational lossless matrix valued
functions. In the same vein as previous works, interpolation
theory with metric constraints is used to ensure the lossless
property. We consider here boundary interpolation and
provide a new parametrization of balanced canonical forms in
which the parameters are angular derivatives. We finally
investigate the possibility to parametrize orthogonal wavelets
with vanishing moments using these results.
16.00 – 16.30
On the Partial Realization Problem
Andrea Gombani, György Michaletzky
Abstract – We consider here a two sided interpolation
problem where we want to minimize the degree of the
interpolant. We show that this degree is given by the rank of a
particular solution to a Sylvester equation which, in some
particular cases becomes a Löwner or a Hankel matrix. We
consider an application to the usual partial realization
problem. The results are quite general and no particular
assumption on the location of the interpolating nodes are
Room 13
Organizer: Malcolm C. Smith
Chair: Malcolm C. Smith
15.30 – 16.00
Passive Synthesis of the Terminal Behavior of Circuits
Jan C. Willems
Abstract–The synthesis of passive electrical circuits has had
a profound influence in electrical engineering in general, and in
systems theory in particular. The seminal result in this area is
the 1932 theorem by Otto Brune in which it is proven that a
transfer function is the driving point impedance of a 2-terminal
one-port consisting of a finite number of passive resistors,
inductors, capacitors, and transformers if and only if the
transfer function is rational and positive real. The notion of
positive realness was in fact first introduced in precisely this
context. Later, Bott and Duffin showed that transformers are
not needed. Brune's result was generalized to multi-ports,
leading to the result that a transfer function is the driving point
impedance of a multi-port consisting of a finite number of
passive resistors, inductors, capacitors, transformers, and
gyrators if and only if the transfer function is a positive real
matrix of rational functions.
The importance of these results may be seen by the
relevance of positive realness and dissipativity in essentially all
areas of present-day control and systems theory. The aim of
this presentation is to examine some aspects of the classical
synthesis question. In another presentation (entitled "Ports and
Terminals") in this conference, we explain that the description
List of Abstracts
of circuits with a finite number of external wires through
which the circuit interacts with its environment leads to a
multi-terminal description, rather than to a multi-port
description. In addition, there is subtle difference, related to
controllability and common factors, between the impedance of
a circuit versus a specification of the external behavior. We
pose the synthesis problem of circuits in terms of a terminal
description. Necessary and sufficient conditions are given for
the terminal behavior of a circuit consisting of an
interconnection of positive resistors. In its simplest form, these
conditions require that the matrix that relates the voltage
vector to the current vector be symmetric hyperdominant with
zero excess. This result is readily generalized to circuits
containing only inductors and capacitors. Other cases will be
discussed as well. Finally, we examine the subtle difference
between synthesizing a linear time-invariant differential
behavior versus synthesizing a transfer function. We show that
the classical synthesis results apply only to controllable
behaviors, but that the synthesis of non-controllable systems is
basically an open area. In fact, the classical Bott-Duffin result
realizes a given transfer function in a non-controllable fashion.
This implies that the transformerless synthesis of a controllable
linear time-invariant differential one-port behavior remains an
open question.
16.00 – 16.30
Redundancies in Transformerless Network Synthesis
Jason Zheng Jiang, Malcolm C. Smith
Abstract – The purpose of this paper is to give a
compact summary of recent results based on the concept of a
regular positive-real function. We will list an efficient set of
networks which is capable of realising all biquadratics which
are realisable by 5-element networks or 6-element networks of
seriesparallel type. The structures are simpler than the full
Bott- Duffin synthesis, though there are some (non-regular)
positivereal biquadratics which cannot be realised by this
system of differential time-delay equations, system of
difference equations). Finding an equivalent representation of
a linear functional system containing fewer equations and
fewer unknowns generally simplifies the study of its structural
properties, its closed-form integration and different numerical
issues. The purpose of this paper is to present a constructive
approach to Serre’s reduction for linear functional systems.
16.00 – 16.30
Serre's Reduction of Linear Partial Differential Systems
Based on Holonomy
Thomas Cluzeau, Alban Quadrat
Abstract – Given a linear functional system (e.g., an
ordinary/ partial differential system), Serre’s reduction aims
at finding an equivalent linear functional system which
contains fewer equations and fewer unknowns. The purpose of
this paper is to study Serre’s reduction of underdetermined
linear systems of partial differential equations with analytic
coefficients whose formal adjoints are holonomic in the sense of
algebraic analysis. In particular, we prove that every analytic
linear system of ordinary differential equations with at least
one input is equivalent to a sole analytic ordinary differential
16.30 – 17.00
Purity Filtration of 2-Dimensional Linear Systems
Alban Quadrat
16.30 – 17.00
Network Optimization and Synthesis Using a Combined
Mechanical and Electrical System: Application to Vehicle
Suspension Control
Fu-Cheng Wang, Hsiang-An Chan
Abstract – This paper introduces a mechatronic network
and applies it to vehicle suspensions for performance
optimization. The mechatronic network consists of a ball-screw
and permanent magnet electric machinery (PMEM), such that
the system impedance is a combination of mechanical and
electrical impedances. We then apply the network to vehicle
suspensions, and demonstrate the performance benefits and
their sensitivities to parameter variations. The optimal
electrical impedances are constructed and experimentally
verified. Based on the results, the mechatronic network is
deemed effective.
New Mathematical Methods in Multidimensional
Systems Theory – 1
(Invited Session)
Room 14
Organizers: Alban Quadrat, Thomas Cluzeau
Chair: Alban Quadrat
15.30 – 16.00
Further Results on Serre’s Reduction of Multidimensional
Linear Systems
Mohamed S. Boudellioua, Alban Quadrat
Abstract – Serre’s reduction aims at reducing the number
of unknowns and equations of a linear functional system
(e.g., system of ordinary or partial differential equations,
17.00 – 17.30
A Polynomial-Algebraic Approach to Lyapunov Stability
Analysis of Higher-Order 2-D Systems
Paolo Rapisarda, Kiyotsugu Takaba, Chiaki Kojima
Abstract – We introduce a four-variable polynomial
matrix equation which plays an essential role in the stability
analysis of discrete 2-D systems and in the computation of
Lyapunov functions for such systems; we call this the 2-D
polynomial Lyapunov equation (2-D PLE). We also give
necessary and sufficient conditions for the stability of “square”
2-D systems based on solutions of the 2-D PLE satisfying
additional properties.
Monday, 5 July
Recent Developments in Multidimensional
Systems, Control and Signals – Theory and
Applications – 1
(Invited Session)
Room 15
Organizer: Krzysztof Gałkowski
Chair: Krzysztof Gałkowski
15.30 – 16.00
On a Nonlinear Two-Directionally Continuous Repetitive
Marek Majewski
Abstract – In the paper we consider a nonlinear,
continuous version of the well-known discrete and differential
repetitive process. A two-directionally linear continuous
version of repetitive process has been introduced in paper [9].
The aim of this paper is to introduce a nonlinear version of the
system considered in [9] and to prove the fundamental results
for such systems: the existence, uniqueness and the
continuous dependence of solutions on functional parameters
(controls), as it has been obtained in [9] for the linear system.
16.00 – 16.30
Controllability, Observability and Disturbance Attenuation
by Boundary Control of Repetitive Processes with
Teresa Azevedo-Perdicoúlis, Gerhard Jank
Abstract – In this paper, we present an explicit
representation of solutions for a specific class of linear
repetitive processes with smoothing. This representation then is
used to obtain direct criteria for controllability and
observability properties of this class of discrete time 2–D
systems with delays. We not only consider classical
controllability properties, where control is obtained by
but also
controllability of the system by steering the system
through boundary data control. From the point of view of
technical applications, for instance in high pressure gas
network modelling (see [1]), it seems to be more reliable to
consider boundary data controls. Therefore in this paper we
emphasise boundary control properties of the system. A
disturbed optimal boundary control problem with a quadratic
criterion is also solved.
16.30 – 17.00
On Asymptotically Reconstructing Local-State Trajectories
for 2-D Systems without Exploiting Input Information
Lorenzo Ntogramatzidis, Michael Cantoni
Abstract – The
detectability subspaces are developed within the context of
observer synthesis for two-dimensional (2-D) FornasiniMarchesini models. Specifically, the paper considers observers
which asymptotically estimate the local state, in the sense that
the error tends to zero as the reconstructed local state evolves
away from possibly mismatched boundary values, modulo a
detectability subspace. Ultimately, the synthesis of such
observers in the absence of explicit input information is
17.00 – 17.30
LQ Parametrization of Robust Stabilizing Static Output
Feedback Controllers for 2D Continuous Roesser Systems
Pavel V. Pakshin, Krzysztof Gałkowski, Eric Rogers
Abstract – The paper considers robust control of 2D
linear systems described by the Roesser model, where
information propagation in each the independent directions is
a function of a continuous variable, and an affine parallelotopic
type model for uncertainty is assumed. Sufficient optimality
conditions for the existence of an LQ state feedback controller
are developed in the absence of model uncertainty and then
used to characterize a set of stabilizing static output feedback
controllers for this system in the presence of parameter
in non-convex
parameterized by the weighting matrices of a quadratic cost
function. Replacing these conditions by convex approximations
leads to an algorithm for computing the stabilizing gain matrix
of the controller. The algorithm is non-iterative and uses
computationally efficient SDP solvers. A numerical example is
given to demonstrate the applicability and effectiveness of the
List of Abstracts
12.00 – 12.30
Tuesday, 6 July
Infinite Structure for Infinite-Dimensional Systems: a
Directional Approach
Petteri Laakkonen, Seppo Pohjolainen
Plenary Lecture
Room 1
Chair: Joseph A. Ball
09.00 – 10.00
In memory of Christopher I. Byrnes
Alberto Isidori, Joachim Rosenthal, Anders Lindquist,
Giorgio Picci
Distributed Parameter Systems I: System
(Invited Session)
Room 1
Organizers: Birgit Jacob, Michael A. Demetriou, Miroslav
Krstic, Kirsten Morris, Hans Zwart
Chair: Birgit Jacob
Abstract – In this article structure at infinity of
infinitedimensional linear time invariant systems with
finitedimensional input and output spaces is discussed. It is
shown that by appropriately restricting the paths approaching
infinity and under certain majorization conditions a diagonal
form that describes the behavior at infinity can be found. This
diagonal form is a generalization of the Smith-McMillan form
at infinity. It is then used to simplify certain solvability
conditions of a regulation problem. Examples on time-delay
and distributed parameter systems are given.
Probabilistic Methods
(Regular Session)
Room 2
Chair: Pavel Shcherbakov
10.30 – 11.00
10.30 – 11.00
Second-Order Systems with Acceleration Measurement
Birgit Jacob, Kirsten Morris
Boundary Oracles for Control-Related Matrix Sets
Pavel Shcherbakov
Abstract – A number of systems are modelled by partial
differential equations that include second-order derivatives
with respect to time. Flexible structures, acoustic waves in
cavities as well as coupled acoustic-structure systems are
examples of systems modelled by equations of this type.
Accelerometers are a very popular choice of sensor for these
systems. The systems theory for acceleration systems has not
been well-studied. In this talk, conditions under which these
systems are well-posed are established. We obtain a
representation for the input/output map and transfer function
for the situation where the control system may not be wellposed. We provide several examples to show that in general
using acceleration as the output leads to an ill-posed
system. We then develop a model for acceleration
measurements that incorporates a model for the microelectrical-mechanical systems (MEMS) devices used to measure
acceleration. With this more complex model, the control system
is in general well-posed with a natural choice of state space.
Abstract – This paper presents closed-form solutions for
the problem of finding the points of intersection of a 1D line
and the boundary of typical matrix sets encountered in
control; specifically, those defined by linear matrix inequalities.
This procedure is referred to as boundary oracle; it is the key
technical component of various random walk algorithms
exploited within the randomized approach to control and
optimization. In the paper, several such oracles are devised and
generalized to robust formulations where the coefficients of
matrix inequalities are subjected to uncertainties.
11.00 – 11.30
Riesz Basis for Strongly Continuous Groups
Hans Zwart
Abstract – Given a Hilbert space and the generator of
a strongly continuous group on this Hilbert space. If the
eigenvalues of the generator have a uniform gap, and if the
span of the corresponding eigenvectors is dense, then these
eigenvectors form a Riesz basis (or unconditional basis) of the
Hilbert space. Furthermore, we show that none of the
conditions can be weakened.
11.30 – 12.00
Optimal Control of Fractional Systems: a Diffusive
Denis Matignon
Abstract – Optimal control of fractional linear systems on
a finite horizon can be classically formulated using the
adjoint system. But the adjoint of a causal fractional integral
or derivative operator happens to be an anti-causal
operator: hence, the adjoint equations are not easy to solve in
the first place. Using an equivalent diffusive realization helps
transform the original problem into a coupled system of PDEs,
for which the adjoint system can be more easily derived and
properly studied.
11.00 – 11.30
Potentials and Limitations to Speed up MCMC Methods
Using Non-Reversible Chains
Balázs Gerencsér
Abstract – Mixing time is the quantity to measure the
speed of MCMC sampling. We compare the cases of using
reversible chains, which are better understood with nonreversible chains, which offer more degree of freedom. It turns
out that nonreversible chains can provide significant speedup
in some cases but no improvement in others.
11.30 – 12.00
A Stochastic Paradox in a Model for Reflected Brownian
Erik I. Verriest
Tuesday, 6 July
12.00 – 12.30
Distribution-Dependent Performance of the Good-Turing
Estimator for the Missing Mass
Mesrob I. Ohannessian, Munther A. Dahleh
Abstract – The Good-Turing estimator for the missing
mass has certain bias and concentration properties which
define its performance. In this paper we give distributiondependent conditions under which this performance can or
cannot be matched by a trivial estimator, that is one which
does not depend on observation. We introduce the notion of
accrual function for a distribution, and derive our conditions
from the fact that the latter governs the decay rate of the
mean of the missing mass. These results shed light on the
inner workings of the Good-Turing estimator, and explain why
it applies particularly well for heavy-tailed distributions such
as those that arise when modeling natural language.
(Regular Session)
Room 3
Chair: Paolo Rapisarda
10.30 – 11.00
State Maps from Bilinear Differential Forms
Paolo Rapisarda, Arjan J. van der Schaft
a representation of simultaneous stabilizers under the
assumption that the interconnection of these two behavior is
stable. By using this result, we address to derive a condition
under which a set of three linear behaviors is simultaneously
stabilizable. In this case, we show that: if one of the three
behaviors stabilizes the other two behaviors, then a set of these
three behaviors are simultaneously stabilizable. Moreover, a
representation for simultaneous stabilizer in this case is also
presented under this assumption.
12.00 – 12.30
Deterministic Identification of Lossless and Dissipative
Paolo Rapisarda, Harry Trentelman
Abstract – We illustrate procedures to identify a statespace representation of a passive or bounded-real system from
noisefree measurements. The basic idea underlying our
algorithms is to obtain a state sequence from a rank-revealing
factorization of a Gramian-like matrix constructed from the
data. The computation of state-space equations is then
performed solving a system of linear equations, similarly to
what happens in classical deterministic subspace identification
12.30 – 13.00
Passive Behaviors and their Passive State/Signal
Realizations in Continuous Time
Damir Z. Arov, Mikael Kurula, Olof J. Staffans
Abstract – State equations need often to be constructed
from a higher-order model of a system, resulting for example
from the interconnection of subsystems, or from system
identification procedures. In order to compute state equations
it is crucial to choose a state variable. One way of doing this is
through the computation of a state map, introduced in [4]. In
this paper we develop an alternative approach to the
algebraic characterization of state maps, based on the calculus
of bilinear differential forms (BDFs), see [8]. From this
approach stem a new algorithm for the computation of state
maps, and some new results regarding symmetries of linear
dynamical systems.
11.00 – 11.30
On the Problem of Model Reduction in the Gap Metric
Mark Mutsaers, Siep Weiland
Sigma-Delta Modulators
(Invited Session)
Room 4
Organizer: Paolo Rapisarda
Chair: Paolo Rapisarda
10.30 – 11.00
Incremental Data Converters
Gabor C. Temes, Yan Wang, Wenhuan Yu, Janos Markus
11.30 – 12.00
The Behavioral Approach to Simultaneous Stabilization
Osamu Kaneko
Abstract – Simultaneous stabilization is the problem of
finding a condition under which there exists a single
controller that stabilizes multiple (which is denoted with N in
this paper) plants. In this paper, we address the problem of
simultaneous stabilization in the behavioral framework. First,
we provide a new equivalent condition for a pair of linear
systems to be simultaneously stabilizable. We then also present
Abstract – Incremental data converters (IDCs) are highaccuracy oversampled analog-to-digital converters (ADCs).
They form a special subclass of the commonly used delta-sigma
ADCs. Unlike the latter, IDCs are only operated intermittently,
typically for a few hundred clock periods, and hence they
possess only a finite memory. They offer advantages in high
accuracy, stability, absence of idle tones, low power dissipation,
and ease of multiplexing. Hence, they are often used in sensor
and MEMS interfaces.
In this paper, some recent results on the theory and design
of IDCs are discussed, and illustrated with the description of
a recently implemented data converter.
List of Abstracts
11.00 – 11.30
A Constraint in Single-Feedback Sigma Delta ForceFeedback Loops with a Discrete-Time Loop Filter
Pieter Rombouts, Johan Raman, Ludo Weyten
Abstract – In this tutorial paper we review Sigma Delta force
Micro-ElectroMechanical (MEMS) inertial sensors with a discrete time loop
filter. First we focus on the single feedback structure (with only
mechanical feedback). It is shown that in this situation the
mechanical transfer function introduces a zero in the overall
loop gain. This gives rise to a constraint in the realizable NTF.
In theory this can be overcome by adding a pole to the
controller. An alternative solution is to add an extra electrical
feedback branch. The latter solution is considered more
beneficial in terms of power consumption and chip area.
problem involves rational expressions in these matrix variables
which have therefore the same form independent of the matrix
sizes. Hence the study of LMIs in systems and control leads not
so much to classical convex analysis and positivity, but rather
to the newly emerging areas of (free) noncommutative
convexity and noncommutative positivity, with the polynomials
and rational functions in commuting variables replaced by
noncommutative polynomials and noncommutative rational
functions. The purpose of this minicourse is to provide an
introduction to noncommutative rational functions and their
realization theory on one hand, and to noncommutative
convexity, including
noncommutative LMIs, on the other.
Multi-Agent Systems
(Regular Session)
11.30 – 12.00
Quantization Noise Conditioning Techniques for Digital
Delta-Sigma Modulators
Sudhakar Pamarti
Abstract – This
outstanding theoretical problems in delta-sigma modulator
based electronic digital-to-analog circuits and outlines
quantization noise conditioning techniques that are being
employed to address these problems. Both the problems and
the conditioning techniques are described in the context of a
special class of electronic circuits called frequency
12.00 – 12.30
Dynamics-Level Design for Discrete- and Continuous-Time
Band-Pass Sigma-Delta-Modulators for Micro-Machined
Jian Luo, Paolo Rapisarda, Michael Kraft
Room 7
Chair: Magnus Egerstedt
10.30 – 11.00
Sustainable Group Sizes for Multi-Agent Search-and-Patrol
Musad Haque, Magnus Egerstedt, Clyde F. Martin
Abstract – We identify sustainable sizes for a multiagent system that consists of two classes of agents: one class
is responsible for searching an area; the other for
providing perimeter security for that area. In this context,
sustainability means the ability of the system to accomplish the
task while balancing shared resources. Bio-inspired rules based
on the pride structures of African lions are developed to
determine the sustainability of a group size.
11.00 – 11.30
Dynamic Spectral Clustering
Amy LaViers, Amir Rahmani, Magnus Egerstedt
Abstract – Clustering is a powerful tool for data
classification; however, its application has been limited to
analysis of static snapshots of data which may be time-evolving.
This work presents a clustering algorithm that employs a fixed
time interval and a time-aggregated similarity measure to
determine classification. The fixed time interval and a
weighting parameter are tuned to the system’s dynamics;
otherwise the algorithm proceeds automatically finding the
optimal cluster number and appropriate clusters at each time
point in the dataset. The viability and contribution of the
method is shown through simulation.
11.30 – 12.00
Noncommutative Rational Functions and
Noncommutative Convexity – 1
Bounds and Approximations on the Reliability of Large
Tamás Szántai, József Bukszár, Edith Kovács
Room 6
Organizers: William J. Helton, Dmitry S. KaliuzhnyiVerbovetskyi, Igor Klep, Victor Vinnikov
10.30 – 12.30
Noncommutative Rational Functions and Noncommutative
William J. Helton, Dmitry S. Kaliuzhnyi-Verbovetskyi, Igor
Klep, Victor Vinnikov
Abstract – One of the biggest recent revolutions in
optimization, called
methodology for solving Linear Matrix Inequalities (LMIs)
proposed in 1994 by Nesterov and Nemirovski. As it turns out,
most optimization problems appearing in systems and control
are dimensionindependent. Namely, the natural variables are
matrices (rather than just collections of scalars) and the
Abstract–We present methods for calculating bounds
and approximations on the reliability of large networks. We
regard the most reliable few paths leading from the start node
to the terminal node of the network and then calculate lower
and upper bounds on the probability that at least one of these
paths is permeable. For this purpose we will use the so called
multitree upper and hypermultitree lower bounds on the
probability of union of events, developed earlier by J. Bukszár.
Chow- Liu’s dependence tree approximation of multivariate
discrete probability distributions was generalized earlier by T.
Szántai and E. Kovács to higher order dependence tree
approximations, called k-th order t-cherry junction tree
approximations. We will show how the Chow-Liu’s dependence
tree and the more general t-cherry junction tree graph
structures can be used for bounding and approximating the
network reliability. Some possible applications and numerical
results will also be presented.
Tuesday, 6 July
12.00 – 12.30
On the Geometry and Deformation of Switching Manifolds
for Autonomous Hybrid Systems
Farzin Taringoo, Peter E. Caines
Abstract – This paper provides a geometrical analysis
of autonomous hybrid optimal control systems (HOCS) by
studying the properties of switching manifolds in Euclidean
space and their associated optimal hybrid trajectories.
Motivational examples are to be found in the speed dependent
operation of automatic gear shift systems of heavy trucks [1].
In this paper the mathematical formulation of a hybrid system
is presented and then the Hybrid Maximum Principle (HMP)
necessary conditions for the optimality of a hybrid system
trajectory are given, (see [6],[7]). Second order optimality
conditions are given in terms of the Hessian matrix of the value
function and geometrical data involving the curvature of the
switching manifold at its intersection with an optimal
trajectory. At the end, the energy of a switching manifold
deformation mapping is defined and the hybrid cost
optimization is performed with respect to such deformation
Recent Developments in Multidimensional
Systems, Control and Signals – Theory and
Applications – 2
(Invited Session)
Room 8
Organizer: Krzysztof Gałkowski
Chair: Krzysztof Gałkowski
describe the stability regions in the parameter space using a
convex constraint.
11.30 – 12.00
An Output Control of a Class of Discrete Second-order
Repetitive Processes
Pawel Dabkowski, Krzysztof Gałkowski, Biswa Datta
Abstract – This paper presents new computationally
efficient LMI results on stability and feedback stabilization for
a class of ill-conditioned discrete, linear second-order repetitive
processes, including the uncertain case. The results are derived
via transformation of the second-order system to an
equivalent first-order descriptor system, thus avoiding the
necessity of inversion of an ill-conditioned leading coefficient
matrix of the system, which is allowed also to be uncertain.
This last feature, as frequently occurs, is of a great significance
but the known approaches do not provide the easy way to solve
this problem.
12.00 – 12.30
Approximation of ND Systems with Multiple Dependent
Femke van Belzen, Siep Weiland
Abstract – Multi-variable distributed systems describe
the evolution of multiple dependent variables over a domain
of independent variables. This paper considers model
reduction for this type of systems. The method of Proper
Orthogonal Decompositions (POD) is adapted using concepts
from tensors and tensor decompositions. The result is a model
reduction framework that is applicable to systems with an
arbitrary number of dependent and independent variables.
10.30 – 11.00
Existence of Optimal Solutions of Two-Directionally
Continuous Repetitive Process under Convexity Assumption
Marek Majewski
Abstract – In the paper a sufficient condition for the
existence of optimal solution to the optimal control problem
governed by two-directionally continuous repetitive process
and the integral cost functional is given. In the main theorem
the crucial assumption is the convexity of the so-called
generalized velocities set. The proof of the main theorem is
based on the lower closure theorem.
11.00 – 11.30
Distributed Stabilization of Spatially Invariant Systems:
Positive Polynomial Approach
Petr Augusta, Zdeněk Hurak
Abstract–The paper gives a computationally feasible
characterisation of
controllers stabilising a spatially invariant system. This gives a
building block for convex optimisation based control design for
these systems. Mathematically, such systems are described by
partial differential equations with coefficients independent on
time and location. In this paper, a situation with one spatial
and one temporal variable is considered. Models of such
systems can take a form of a 2-D transfer function. Stabilising
distributed feedback controllers are then parametrised as a
solution to the Diophantine equation ax + by = c for a given
stable bivariate polynomial c. This paper brings a
computational characterisation of all such stable 2-D
polynomials exploiting the relationship between a stability of a
2-D polynomial and positiveness of a related polynomial matrix
on the unit circle. Such matrices are usually bilinear in the
coefficients of the original polynomials. It is shown that a
factorisation of the Schur-Cohn matrix enables linearisation of
the problem, at least in a special instance of first-order systems.
Then the computational framework of linear matrix
inequalities and semidefinite programming can be used to
Stability and Switching
(Regular Session)
Room 9
Chair: Vahid S. Bokharaie
10.30 – 11.00
On the D-Stability of Linear and Nonlinear Positive Switched
Vahid S. Bokharaie, Oliver Mason, Fabian Wirth
Abstract – We present a number of results on D-stability of
positive switched systems. Different classes of linear
and nonlinear positive switched systems are considered and
simple conditions for D-stability of each class are presented.
11.00 – 11.30
On the Stabilizability of Discrete-Time Positive Switched
Ettore Fornasini, Maria Elena Valcher
Abstract – In this paper we consider the class of
discretetime systems switching between an arbitrary number p
of autonomous positive subsystems. Necessary and sufficient
conditions for the existence of (either linear or quadratic)
copositive Lyapunov functions, whose values can be decreased
in every positive state, by suitably choosing one of p
subsystems, are obtained. When these conditions are fulfilled,
state-dependent switching strategies, which prove to be
stabilizing, can be adopted. Finally, the performances of these
Lyapunov based strategies are compared.
11.30 – 12.00
Stability Criteria for Planar Linear Systems with State Reset
Svetlana Polenkova, Jan Willem Polderman, Rom Langerak
Abstract – In this work we perform a stability analysis for
a class of switched linear systems, modeled as hybrid
List of Abstracts
automata. We deal with a switched linear planar system,
modeled by a hybrid automaton with one discrete state. We
assume the guard on the transition is a line in the state space
and the reset map is a linear projection onto the x-axis. We
define necessary and sufficient conditions for stability of the
switched linear system with fixed and arbitrary dynamics in
the location.
12.00 – 12.30
On the Preservation of Co-Positive Lyapunov Functions
under Padé Discretization for Positive Systems
Annalisa Zappavigna, Patrizio Colaneri, Stephen Kirkland,
Robert Shorten
Abstract – In this paper the discretization of switched
and non-switched linear positive systems using Padé
approximations is considered. We show:
1) diagonal Padé approximations preserve both linear
and quadratic co-positive Lyapunov functions;
2) positivity need not be preserved even for arbitrarily
small sampling
approximations. Sufficient
approximations are given to preserve positivity of the discretetime system. Finally, some examples are given to illustrate the
efficacy of our results.
Algebraic Structures – 1
(Regular Session)
Room 10
Chair: Néstor Thome
10.30 – 11.00
11.30 – 12.00
Explicit Parameterization of All Solutions of Linear Periodic
Systems with Real-Valued Coefficients
Ichiro Jikuya, Ichijo Hodaka
Abstract – An extension is introduced to the recently
introduced representation for linear periodic systems with
realvalued coefficients. In order to parameterize all state
transition matrices, a state transition matrix is factored as a
multiplication of a T-periodic real-valued factor and two realvalued matrix exponential functions. By solving the matrix
equations which give an implicit parameterization, we study
the block structure of those factors and propose an explicit
parameterization. We also study the corresponding block
structure in the coefficient matrices of the systems.
12.00 – 12.30
Factorizations for Some Classes of Matrices Related to
Juan M. Peña
Abstract – We descrtibe factorizations for some classes
of matrices related to positivity and important in
applications. The classes of matrices considered include
nonsingular Mmatrices (matrices with nonpositive off-diagonal
entries with positive inverse) and totally nonnegative matrices
(matrices with all minors nonnegative). The considered
factorizations include rank revealing factorizations, LDUfactorization,
QRfactorization and
factorization. Applications of these factorizations are
Stability and Dissipativity
(Regular Session)
Nonnegativity of Descriptor Systems of Index 1
Alicia Herrero, Francisco J. Ramirez, Néstor Thome
Room 11
Chair: Masaki Ogura
10.30 – 11.00
Dissipativity of Pseudorational Behaviors
Masaki Ogura, Yutaka Yamamoto
Abstract – This paper studies dissipativity for a class
of infinite-dimensional systems, called pseudorational, in the
behavioral context. First a basic equivalence condition for
average nonnegativity of quadratic differential forms induced
by distributions is established as a generalization of the
finitedimensional counterpart. For its proof, we derive a
new necessary and sufficient condition for entire functions of
exponential type (in the Paley-Wiener class) to be
symmetrically factorizable. Utilizing these results we then
study dissipativity of pseudorational behaviors. An example is
given to illustrate results.
11.00 – 11.30
11.00 – 11.30
Approximation of Non-Negative Integer-Valued Matrices
with Application to Hungarian Mortality Data
Márton Ispány, György Michaletzky, Jenı Reiczigel, Gábor
Tusnády, Paula Tusnády, Katalin Varga
Cyclodissipativity and Power Factor Improvement for Full
Nonlinear Loads
Dunstano del Puerto-Flores, Romeo Ortega, Jacquelien M.A.
Abstract – Singular valued decomposition (SVD) is a
commonly applied technique for dimensionality reduction.
SVD implicitely minimizes an unweighted sum of squares
which may be inappropriate in several practical applications.
This paper gives generalizations of SVD to other loss functions,
e.g., weighted Frobenius distance and logistic loss, that are
better suited to the data. We describe algorithms for
minimizing these loss functions, and give an application to
Hungarian mortality data.
Abstract – In
characterization of the problem of power factor compensation
(PFC) for nonlinear loads with non-sinusoidal source voltage
has been presented. Using this characterization the classical
capacitor and inductor compensators can be interpreted in
terms of energy equalization. This brief note focuses on the
extension of this approach. In particular, one result is to show
that power factor compensation is equivalent to a new
cyclodissipativity condition. Another result is to consider
general lossless linear filters as compensators and to show that
the power factor is improved if and only if a certain
Tuesday, 6 July
equalization condition between the weighted powers
inductors and capacitors of the nonlinear load is ensured.
11.30 – 12.00
On Copositive Lyapunov Functions for a Class of Monotone
Björn S. Rüffer, Christopher M. Kellett, Peter M. Dower
Abstract – This paper considers several explicit formulas
for the construction of copositive Lyapunov functions for
global asymptotic stability with respect to monotone systems
evolving in either discrete or continuous time. Such monotone
systems arise as comparison systems in the study of
interconnected large-scale nominal systems. A copositive
Lyapunov function for such a comparison system can then
serve as a prototype Lyapunov functions for the nominal
system. We discuss several constructions from the literature in
a unified framework and provide sufficiency criteria for the
existence of such constructions.
12.00 – 12.30
Stability of Fluid Network Models and Lyapunov Functions
Michael Schönlein, Fabian Wirth
Abstract – We consider the class of closed generic
fluid networks (GFN) models. This class contains for example
fluid networks under general work-conserving and priority
disciplines. Within this abstract framework a Lyapunov
method for stability of GFN models was proposed by Ye and
Chen. They proved that stability of a GFN model is equivalent
to the property that for every path of the model a Lyapunov
like function is decaying. In this paper we construct statedependent Lyapunov functions in contrast to pathwise
functionals. We first show by counterexamples that closed GFN
models do not provide sufficient information that allow for a
converse Lyapunov theorem with state-dependent Lyapunov
functions. To resolve this problem we introduce the class of
strict closed GFN models by forcing the closed GFN model to
satisfy a concatenation and a perfectness condition and define a
statedependent Lyapunov function. We show that for the class
of strict closed GFN models a converse Lyapunov theorem
holds. Finally, it is shown that common fluid network models,
like general work-conserving and priority fluid network
models as well as certain linear Skorokhod problems define
strict closed GFN models.
Mechanical Systems
(Regular Session)
Room 12
Chair: Yoshiro Fukui
10.30 – 11.00
Real-Time Obstacle Avoidance of a Two-Wheeled Mobile
Robot via the Minimum Projection Method
Yoshiro Fukui, Hisakazu Nakamura, Hirokazu Nishitani
Abstract – This paper considers real-time obstacle
avoidance control of a two-wheeled nonholonomic mobile
robot. In this paper, we propose a discontinuous asymptotic
stabilizing state feedback control law for a real-time obstacle
avoidance problem via the minimum projection method. The
method guarantees asymptotic stability and reduces the
computational cost. The effectiveness of the proposed method is
comfirmed by experiments.
11.00 – 11.30
Dynamics and Control of 2D SpiderCrane: a RHC Approach
Atul K. Kamath, Faruk Kazi, Navdeep M. Singh
Abstract – In this paper we present modeling and control of
a multicable suspended mechanism called the ‘2D SpiderCrane’. A spidercrane does not have any conventional
heavy components and makes use of cables by virtue of which
high transfer speeds are achievable. The initial part of the
paper addresses the modeling of the mechanism, where the
cable and pulley dynamics are separated, and the payload is
viewed as a pendulum suspended from a cable whose
suspension point lies on a mass that moves in a twodimensional space. Using the Receding Horizon Control (RHC)
strategy a trajectory tracking controller is proposed.
Simulations were carried out in MATLAB.
11.30 – 12.00
Sampled-Data Cross-Track Control for Underactuated Ships
Hitoshi Katayama
Abstract – The sampled-data cross-track control problem
for an underactuated three degree-of-freedom ship is
considered. A line-of-sight guidance algorithm is used to design
surge and yaw control laws which make a ship track a desired
straightline reference trajectory while maintaining a desired
nonzero constant forward speed. Then applying the nonlinear
sampleddata control theory and the stability theory of
parametrized discrete-time cascade interconnected systems, it
is shown that sampled-data cross-track is achieved by the
designed control laws. Simulation results are also given to
illustrate the design method.
12.00 – 12.30
Intelligent Multiple AUV Path Planning
P.B. Sujit, João de Sousa, Bernardo Maciel
Abstract – One of the primary requirements of autonomous
underwater vehicle (AUV) navigation is obstacle avoidance
capability. Currently, AUVs have obstacle avoidance
capability, however, they do not have intelligence to decide if
narrow passages are safe for navigation or not. The AUVs
update their location using inertial navigation method
continuously which accumulates error with time. In order take
the position deviation of the vehicle into account we use reach
sets that defines safe region for the AUV. Determining reach
set for a given time horizon is computationally intensive,
therefore we use ellipsoidal toolbox that can provide
approximate reach set for a given computational time interval.
In this paper, we develop an intelligent path planner for
autonomous underwater vehicles navigating in a rough terrain
with limited sensing and communication capability using reach
sets based on ellipsoidal toolbox. Later, we describe how the
information can be shared between vehicles using micromodems to accurately map the region and cooperatively assign
those regions that are not covered. We present simulation
results to show how the regions can be cooperatively explored
and also study effect of sensing range, communication range
and number of vehicles through Monte-Carlo simulations.
List of Abstracts
Moment Problems, Maximum Entropy, and
Covariance Extension
(Invited Session)
Room 13
Organizer: Giorgio Picci
Chair: Giorgio Picci
10.30 – 11.00
A Maximum Entropy Approach to the Covariance Extension
Problem for Reciprocal Processes
Francesca Carli, Augusto Ferrante, Michele Pavon, Giorgio
Abstract – This paper addresses the problem of completing a
partially specified symmetric matrix, where the specified
entries lie on a single band centered on the main diagonal, in
such a way that the completed matrix is positive definite,
blockcirculant and with a banded inverse. This particular
completion has the meaning of the covariance matrix of a
reciprocal process stationary on the discrete circle ZN. This
problem, called the block-circulant band extension problem,
arises in the context of maximum likelihood identification for
such processes. This paper shows that the block-circulant band
extension problem can in fact be solved as a maximum entropy
problem. Indeed, the constraint that the inverse be banded can
be removed with a considerable theoretical and computational
simplification, as the maximum entropy block-circulant
extension can be shown to always enjoy this property.
Conditions for the feasibility of the problem are also provided.
11.00 – 11.30
On the Factorization Approach to Band Extension of BlockCirculant Matrices
Francesca Carli, Giorgio Picci
12.00 – 12.30
On the Maximum Entropy Completion of Circulant
Covariance Matrices
Francesca Carli, Tryphon T. Georgiou
Abstract – This paper deals with the positive-definite
completion of partially specified (block-) circulant covariance
matrices. In the absence of any constraint other than positivity,
the maximal-determinant completion of a partially specified
covariance matrix (i.e., the so-called maximum entropy
completion) was shown by Dempster to have an inverse with
zero-values at all locations where the original matrix was
unspecified–this will be referred to as the Dempster property.
In earlier work, Carli etal. [2] showed that even under the
constraint of a covariance being block-circulant, as long as the
unspecified elements are in a single band, the maximum
entropy completion has the Dempster property. The purpose of
the present paper is to prove that circulant, block-circulant, or
Hermitian constraints do not interfere with the Dempster
property of the maximum entropy completion. I.e., regardless
of which elements are specified, the completion has the
Dempster property. This fact is a direct consequence of the
invariance of the determinant to the group of transformations
that leave circulant, block-circulant, or Hermitian matrices
invariant. A description of the set of all positive extensions is
discussed and connections between this set and the
variables, facilitated by the circulant structure, is highlighted.
Codes and Rings
(Invited Session)
Room 14
Organizers: Joachim Rosenthal, Marcus Greferath
Chair: Marcus Greferath
10.30 – 11.00
On the Parameters of Two-Weight Codes over Rings
Alfred Wassermann
Abstract – Linear codes over rings are interesting
because there are examples which can correct more errors than
comparable linear codes over finite fields. Here, we look on the
special case of linear codes attaining exactly two nonzero
weights and give an exhaustive list on the weights of such
11.30 – 12.00
Graphical Models of Autoregressive Moving-Average
Enrico Avventi, Anders Lindquist, Bo Wahlberg
Abstract – Consider a Gaussian stationary stochastic
vector process with the property that designated pairs of
components are conditionally independent given the rest of the
components. Such processes can be represented on a graph
where the components are nodes and the lack of a connecting
link between two nodes signifies conditional independence.
This leads to a sparsity pattern in the inverse of the matrixvalued spectral density. Such graphical models find
applications in speech, bioinformatics, image processing,
econometrics and many other fields, where the problem to fit
an autoregressive (AR) model to such a process has been
considered. In this paper we take this problem one step
further, namely to fit an autoregressive moving-average
(ARMA) model to the same data. We develop a theoretical
framework which also spreads further light on previous
approaches and results.
11.00 – 11.30
Predictable Degree Property and Row Reducedness for a
System over a Semi Simple Ring
Mohammed El Oued, Patrick Solé
11.30 – 12.00
A New Series of Z4-Linear Codes of High Minimum Lee
Distance Derived from the Kerdock Codes
Michael Kiermaier, Johannes Zwanzger
Tuesday, 6 July
12.00 – 12.30
Constructions of Two-Weight Codes over Finite Rings
Eimear Byrne, Alison Sneyd
11.30 – 12.00
Asymptotic Values of Zero Sum Repeated Games: Evolution
Equations in Discrete and Continuous Time
Guillaume Vigeral
Max-Plus, Tropical and Idempotent Methods in
Control – 1
(Invited Session)
Room 15
Organizers: John S. Baras, William M. McEneaney
Chair: John S. Baras
10.30 – 11.00
Routing in Equilibrium
João Luís Sobrinho, Timothy G. Griffin
Abstract – Some path problems cannot be modeled
using semirings because the associated algebraic structure is
not distributive. Rather than attempting to compute globally
optimal paths with such structures, it may be sufficient in some
cases to find locally optimal paths-paths that represent a
stable local equilibrium. For example, this is the type of
routing system that has evolved to connect Internet Service
Providers (ISPs) where link weights implement bilateral
commercial relationships between them. Previous work has
shown that routing equilibria can be computed for some nondistributive algebras using algorithms in the Bellman-Ford
family. However, no polynomial time bound was known for
such algorithms. In this paper, we show that routing equilibria
can be computed using Dijkstra’s algorithm for one class of
non-distributive structures. This provides the first polynomial
time algorithm for computing locally optimal solutions to path
problems. We discuss possible applications to Internet routing.
11.00 – 11.30
Stochastic Perturbations of Deterministic Optimization
Problems with Applications to a Spin Control Problem
(Control of a Two-Level Atom)
Vassili N. Kolokoltsov
12.00 – 12.30
Local Pruning for Information Dissemination in Dynamic
Networks for Solving the Idempotent Semiring Algebraic
Path Problem
Kiran K. Somasundaram, John S. Baras
Abstract – We present a method, inspired from routing
in dynamic data networks, to solve the Semiring Algebraic
Path Problem (SAPP) for dynamic graphs. The method can be
used in dynamic networks such as Mobile Ad Hoc Networks,
where the network link states are highly dynamic. The
algorithm makes use of broadcasting as primary mechanism to
recompute the SAPP solution. The solution suffers from
broadcast storm problems, and we propose a selective
broadcasting mechanism that reduces the broadcast storm. We
call this method local pruning and prove its correctness.
Semiplenary Lecture
Room 1
Chair: László Gerencsér
14.00 - 15.00
Variable Robustness Control: Principles and Algorithms
Marco C. Campi, Simone Garatti
Abstract – Robust control is grounded on the idea that
a design should be guaranteed against all possible occurrences
of the uncertain elements in the problem. When this
philosophy is applied to securing a desired performance, it
often leads to conservative, low performing, designs because
emphasis is all placed on the worst-case situation. On the other
hand, in many applications a 100%-guarantee is not necessary,
and it may be convenient to opt for a small compromise in the
List of Abstracts
guarantee level, say 99%, in favor of a (possibly significant)
improvement in the performance. While the above reasoning
sets a sensible principle, to date the real stumbling-block to its
practical use is the lack of computationally-tractable
algorithmic methods to trade guarantees for performance. This
paper aims to open new directions to address this problem, and
we show that this result can be achieved through
16.00 – 16.30
Absolute Stability of a System with Distributed Delays
Modeling Cell Dynamics in Leukemia
Hitay Özbay, Houda Benjelloun, Catherine Bonnet, Jean
Chair: Lars Grüne
Abstract – In this paper we consider a mathematical
model proposed recently by Adimy et al. (2008) for studying
the cell dynamics in Acute Myelogenous Leukemia (AML). By
using the circle and Popov criteria, we derive absolute stability
conditions for this nonlinear system with distributed delays.
Connections with the earlier results on stability of the
linearized model are also made. The results are illustrated with
a numerical example and simulations.
14.00 - 15.00
16.30 – 17.00
Synthesis of Electrical and Mechanical Networks
Malcolm C. Smith
Admissibilty for Volterra Systems with Scalar Kernels
Bernhard H. Haak, Birgit Jacob
Abstract – The synthesis of electrical networks whose
driving-point immittance is some prescribed positive-real
function has given rise to a rich classical theory including the
celebrated results of Brune, Darlington, Bott and Duffin etc.
After the 1970s, there was a decline in interest due to the
increasing prevalence of active circuits. Despite the relative
maturity of the field, some basic questions remained
unanswered, e.g. on the most efficient realisations for
transformerless synthesis. The latter question becomes
important again in the context of mechanical networks with the
introduction of the inerter as an ideal two-terminal analogue of
the capacitor (in contrast to the mass element which is
analogous only to the grounded capacitor). This talk will
discuss the motivation for passive mechanical network
synthesis, survey some classical results of electric circuit
synthesis, and discuss recent progress on the concept of regular
positive-real functions and its application to transformerless
Abstract – Volterra observations systems with scalar
kernels are studied. New sufficient conditions for admissibility
of observation operators are developed.
Semiplenary Lecture
Room 14
Semiplenary Lecture
Room 15
Chair: Uwe Helmke
14.00 - 15.00
Quantify the Unstable
Li Qiu
Abstract – The Mahler measure, a notion often appearing in
the number theory and dynamic system literature, provides a
way to quantify the instability in a linear discrete-time system.
17.00 – 17.30
Multiscale Dynamics Optimal Control of Parabolic PDE with
Time Varying Spatial Domain (Crystal Growth Process)
James Ng, Ilyasse Aksikas, Stevan Dubljevic
Abstract – This paper considers the multi-scale optimal
control of the Czochralski crystal growth process. The
temperature distribution of the crystal is realized by heat input
at the boundary and by the force applied to the mechanical
subsystem drawing the crystal from a melt. A parabolic partial
differential equation (PDE) model describing the temperature
distribution of the crystal is developed from first-principles
continuum mechanics to preserve the time-varying spatial
domain dynamical features. The evolution of the temperature
distribution is coupled to the pulling actuator finitedimensional subsystem with dynamics modelled as a second
order ordinary differential equation (ODE) for rigid body
mechanics. The PDE timevarying spatial operator with natural
boundary conditions is characterized as a Riesz-spectral
operator in the L2(0; l(t)) functional space setting. The finite
and infinite-time horizon optimal control law for the infinitedimensional system is obtained as a solution to a timedependent and time-invariant differential Riccati equation.
2. System Identification
(Regular Session)
Room 2
Chair: Ignat Domanov
Distributed Parameter Systems II: System
Theoretical Properties
(Invited Session)
Room 1
Organizers: Birgit Jacob, Michael A. Demetriou, Miroslav
Krstic, Kirsten Morris, Hans Zwart
Chair: Birgit Jacob
15.30 – 16.00
Non-Dissipative Boundary Feedback for Elastic Beams
Chris Guiver, Mark R. Opmeer
Abstract – We show that a non-dissipative feedback that
has been shown in the literature to exponentially stabilize an
Euler- Bernoulli beam makes a Rayleigh beam and a
Timoshenko beam unstable.
15.30 – 16.00
Enhanced Line Search for Blind Channel Identification Based
on the Parafac Decomposition of Cumulant Tensors
Ignat Domanov, Lieven de Lathauwer
Abstract – In this paper we consider higher-order
cumulant based methods for the blind estimation of a singleinput single-output finite impulse response system driven by a
non- Gaussian signal. This problem can be interpreted as a
particular polynomial optimization problem. Using the link
between this problem and the parallel factor decomposition of
a third-order tensor we present a new representation of the
cost function and give an explicit expression for its complex
gradient. Then we explore convergence/non convergence of the
single-step leastsquares algorithm and improve it by enhanced
line/plane search procedures.
Tuesday, 6 July
16.00 – 16.30
Quantification of Model Uncertainty for a State-Space
Wafa Farah, Guillaume Mercère, Thierry Poinot, Jan-Willem
van Wingerden
Structural Properties of Realizations
(Regular Session)
Room 3
Chair: W. Steven Gray
Abstract – In this communication, the uncertainty
domain determination problem for multi-input multi-output
systems described with a linear time-invariant state-space
representation is adressed. The developed method is based on a
two-step approach. The first step consists in estimating the
nominal model using a particular least-squares subspace
algorithm. Then, the uncertainty domains are described by
using a bounded error approach. Simulations are used to
highlight the performance of the method.
15.30 – 16.00
16.30 – 17.00
16.00 – 16.30
On Asymptotic Properties of MOESP-Type Closed-Loop
Subspace Model Identification
Hiroshi Oku
On the Existence of Various Realizations
Jana Němcová, Jean-Baptiste Pomet
Abstract – Recently, MOESP-type closed-loop subspace
model identification (CL-MOESP) has been proposed by
the authors and its effectiveness has been demonstrated via
both numerical simulations and real-life systems, e.g., a cartinverted pendulum system. However, asymptotic properties of
CLMOESP has not yet been studied. The purpose of this
paper is to clarify the asymptotic properties of CL-MOESP
from the viewpoint of Two-stage closed-loop identification.
Moreover, it is shown that CL-MOESP minimizes a truncation
error due to a finite number of sampled data.
17.00 – 17.30
Regularized Parametric Models of Nonstationary Processes
Daniel Rudoy, Tryphon T. Georgiou
Abstract–In this article, we study two classes of
nonstationary processes respectively parameterized by timevarying autoregressions and time-varying lattice filters. The
processes considered are induced by solutions to certain convex
optimization problems with local or global constraints, and
are consistent with standard models of their stationary
counterparts. We show that an underlying nesting property
naturally leads to a family of hypothesis tests for stationarity
and provide a geometric interpretation of our results on the
manifold of allpole rational transfer functions.
17.30 – 18.00
Coupled Segmentation for Anatomical Structures by
Combining Shape and Relational Spatial Information
Ivan Kolesov, Vandana Mohan, Gregory Sharp, Allen R.
Abstract–We propose a sequential method to estimate
a shape prior using previously segmented structures as
landmarks. It is founded on probabilistic principal
component analysis and probabilistic canonical correlation
analysis. We derive equations in order to utilize these
techniques for prediction. At a given stage in a sequence of
segmentations, this approach predicts the most likely shape of
the structure being segmented based solely on the
segmentations of completed structures. Hence, the shape prior
is independent of the image information around the target.
This is applied to the problem of adaptive radiotherapy in
oncology. Structures of interest in the head and neck region
have insufficient image information and strictly image based
approaches fail. Such cases also present major problems for
methods that simultaneously perform segmentation and fitting
of a shape model to image data. The strength of our method is
the flexibility that it provides to the user in determining what
image information to trust. We demonstrate our technique on a
dataset that is illustrative of real-world data for our
applications in volume and in variance.
On the Rationality of the Feedback Connection
W. Steven Gray
Abstract – This paper presents a variety of necessary
conditions and sufficient conditions under which the
feedback interconnection of two rational input-output systems,
that is, systems having bilinear state space realizations,
produces a closed-loop system which is also rational.
Abstract – The aim of this paper is to characterize
the existence of polynomial, rational and Nash realizations
with respect to one another. The existence of realizations
within various classes of systems is the main topic of realization
theory. In this paper it is shown that if there exists a
polynomial realization of a response map then there exists also
its rational realization. To disprove the converse implication we
provide an example of a response map which is realizable by a
rational system but not by a polynomial system. Further, it is
shown that the existence of a rational realization implies the
existence of a Nash realization of the same response map. The
equivalence is proven for response maps defined on piecewiseconstant inputs the values of which are of a finite set. However,
the question whether the existence of a Nash realization implies
the existence of a rational realization generally is still open.
Additionally, we discuss the observability properties of
polynomial, rational and Nash systems.
16.30 – 17.00
External Dynamical Equivalence of Time-Varying Nonlinear
Control Systems on Time Scales
Zbigniew Bartosiewicz, Ewa Pawłuszewicz
Abstract – Theory of systems on homogeneous time
scales unifies theories of continuous-time and discrete-time
systems. The characterizations of external dynamical
equivalence known for continuous-time and discrete-time
systems with outputs are extended to time-varying systems on
time scales. The main result says that two nonlinear control
systems are externally weakly dynamically equivalent if and
only if their delta universes are properly isomorphic. The delta
operator associated to the given system on a time scale is a
generalization of the differential operator associated to a
continuous-time system and of the difference operator
associated to a discrete-time system.
17.00 – 17.30
Spaces of Nonlinear and Hybrid Systems Representable by
Recognizable Formal Power Series
Mihály Petreczky, Ralf Peeters
Abstract – The paper presents the manifold structure of
the spaces of those nonlinear and hybrid system which can
be encoded by rational formal power series. The latter
class contains bilinear systems, linear multidimensional
systems, linear switched and hybrid systems and jump-markov
linear systems.
List of Abstracts
New Mathematical Methods in Multidimensional
Systems Theory – 2
(Invited Session)
Room 4
Organizers: Alban Quadrat, Thomas Cluzeau
Chair: Thomas Cluzeau
15.30 – 16.00
Frobenius Method for Computing Power Series Solutions of
Linear Higher-Order Differential Systems
Moulay Barkatou, Thomas Cluzeau, Carole El Bacha
Abstract – Linear matrix differential systems appear in
many fields of mathematics and many applications in
mathematical physics and control theory. Computing power
series solutions of such systems around singularities can help in
the understanding of the underlying problem.
The first goal of the present paper is to give a survey on the
classical Frobenius method for computing power series
solutions of linear scalar differential equations. Then the first
contribution of the paper is to show how this Frobenius method
can be generalized to handle general linear higher-order
differential systems.The last part of the paper is dedicated to
the study of higher-order matrix differential control systems
for which we show how to use such a technique to compute
power series solutions.
17.00 – 17.30
Module Structure of Classical Multidimensional Systems
Appearing in Mathematical Physics
Thomas Cluzeau, Alban Quadrat
Abstract – In this paper, within the constructive
algebraic analysis approach to linear systems, we study
classical linear systems of partial differential (PD) equations in
two or three independent variables with constant
coefficients appearing
engineering sciences such as the Stokes and Oseen equations
studied in hydrodynamics. We first provide a precise
algebraic description of the endomorphism ring of the left Dmodule associated with a linear PD system. Then, we use it
to prove that the endomorphism ring of the Stokes and
Oseen equations in R2 is a cyclic D-module, which allows us to
conclude about the decomposition and factorization properties
of these linear PD systems.
Noncommutative Rational Functions and
Noncommutative Convexity – 2
Room 6
Organizers: William J. Helton, Dmitry S. KaliuzhnyiVerbovetskyi, Igor Klep, Victor Vinnikov
16.00 – 16.30
15.30 – 17.30
Controllability and Differential Flatness of Linear Analytic
Ordinary Differential Systems
Alban Quadrat, Daniel Robertz
Noncommutative Rational Functions and Noncommutative
William J. Helton, Dmitry S. Kaliuzhnyi-Verbovetskyi, Igor
Klep, Victor Vinnikov
Abstract – Based on an extension of Stafford’s classical
theorem in noncommutative algebra [24] obtained in [4],
the purpose of this paper is to show that every
controllable linear ordinary differential system with
convergent power series coefficients (i.e., germs of real analytic
functions) and at least two inputs is differentially flat. This
result extends a result obtained in [20], [21] for linear ordinary
differential systems with polynomial coefficients. We show how
the algorithm developed in [21] for the computation of
injective parametrizations and bases of free differential
modules with polynomial or rational function coefficients can
be used to compute injective parametrizations and flat outputs
for these classes of differentially flat systems. This algorithm
allows us to remove singularities which naturally appear in the
computation of injective parametrizations and bases obtained
by means of Jacobson normal form computations.
16.30 – 17.00
Extendability of Multidimensional Linear Systems
Alban Quadrat
Abstract – Within the algebraic analysis approach
to multidimensional linear systems defined by linear systems of
partial differential equations with constant coefficients, the
purpose of this paper is to show how to use
different mathematical results developed in the literature of
algebraic analysis to obtain new characterizations of the
concepts of controllability, in the sense of Willems and PillaiShankar, observability, flatness and autonomous systems
in terms of the possibility to extend (smooth or
distribution) solutions of the multidimensional system and of
its formal adjoint. Each characterization is equivalent to a
moduletheoretic property that can be constructively checked
by means of the packages OreModules and QuillenSuslin.
Networked Systems (Regular Session)
Room 7
Chair: Thomas E. Gorochowski
15.30 – 16.00
A Dynamical Approach to the Evolution of Complex
Thomas E. Gorochowski, Mario di Bernardo, Claire S.
Abstract – In this work we take a dynamical approach to the
evolution of complex networks using simulated output of the
full system dynamic to direct evolution of the
underlying network structure. Extending previous work, we
study the problem of enhanced synchronisation and the
generality of Type 2 features which have been shown to emerge
in regimes where full synchronisation is unstable. Networks are
evolved using a new computational tool called NetEvo which
aims to minimise a dynamical order parameter performance
measure. This process is performed for networks with several
alternative node dynamics, showing in all cases that
qualitatively similar Type 2 topologies emerge. Analysis of
these structures highlights variation in many of the network
statistics and motif frequencies, but helps to classify some key
characteristics exhibited by all Type 2 networks, regardless of
node dynamic.
16.00 – 16.30
Betweenness Centrality Dynamics in Networks of Changing
László Gulyás, Gábor Horváth, Tamás Cséri, Zalán
Szakolczy, George Kampis
Abstract – Dynamic networks are recently in the
foreground of interest in various fields that deal with complex
systems, such as sociology and biology (especially ecology and
Tuesday, 6 July
systems biology and epidemiology). Betweenness centrality of
nodes is a particularly valued concept as a tool for
characterizing large networks by way of selected nodes. In the
present study, we were interested in the effects of various
network changes on average betweenness (BW) centrality in
networks of changing densities. We applied two different
treatments in 100 trials on classic random networks models. In
treatment one, we created various instances of the studied
network models with different densities. We used classic
network model families such as Erdıs-Rényi (ER), BarabásiAlbert (BA) models, among others, and compared these with
empirical network data at various densities. In treatment two,
we studied the robustness of networks by simulating random
node failures and planned attacks according to two scenarios.
Observer Theory
(Invited Session)
Room 8
Organizer: Jochen Trumpf
Chair: Jochen Trumpf
15.30 – 16.00
Functional T-Observers
Ingrid Blumthaler
16.30 – 17.00
Motion Programs for Multi-Agent Control: From Specification
to Execution
Patrick Martin, Magnus Egerstedt
Abstract – This paper explores the process of turning
highlevel motion programs into executable control code for
multiagent systems. Specifically, we use a modified Motion
Description Language (MDL) for networked systems that can
specify motion programs for a collection of autonomous agents.
This MDL includes the network information dependencies
required for each agent to perform coordinated behaviors. We
discuss the design of this framework and the language theoretic
tools used to analyze the information dependencies specified
by these multi-agent motion programs. Additionally, we
develop a supervisor system that monitors the behavior of the
agents on the network, and prevents the agents from entering
into states where information dependencies are violated. We
demonstrate our framework using a simulated multi-robot
17.00 – 17.30
Minimal-time Uncertain Output Final Value of Unknown DTLTI Systems with Application to the Decentralised Network
Consensus Problem
Ye Yuan, Guy-Bart Stan, Ling Shi, Mauricio Barahona, Jorge
Abstract – For
timeinvariant (DTLTI) autonomous system, this paper
characterises the minimal number of steps necessary to
compute the asymptotic final value of an output observed with
uncertainty. We show that this minimal number of steps can
also be obtained directly from a graphical representation of the
DTLTI system using Mason’s rule. Moreover, we provide
heuristic algorithms to compute the final value in a minimal
amount of time with uncertain observations. The general
structure of these algorithms is as follows. Step one, by
introducing a one-step prediction error metric, we characterise
the minimal length of recursion for the outputs of the
considered DTLTI system. Step two, by constructing a new
data set “close” to the original uncertain output data set
satisfying certain conditions, we estimate the final value of the
original output set by computing the final value associated with
this new data set. Step three, we characterise the difference
between the estimated final values obtained from different
estimated data sets. Furthermore, we also consider systems
with time-delays and investigate how the delays affect the
minimal number of steps required to compute the final value.
These results find applications in minimal-time network
consensus problems with minimal and uncertain (e.g., noisy)
16.00 – 16.30
H∞ Observers Design for a Class of Continuous Nonlinear
Singular Systems
Mohamed Darouach, Latifa Boutat-Baddas, Mohamed
Abstract – This paper presents a new solution to the
H∞ observers design problem for a class of Lipschitz
continuous nonlinear singular systems. The approach is based
on the parameterization of the solution of generalized Sylvester
equations. Sufficient conditions for the existence of the
observers which guarantee stability and the worst case
observers error energy over all bounded energy disturbances is
minimized are given. The method also concerns the full-order,
reduced-order, minimal-order observers design A numerical
example is given to show the applicability of our results.
16.30 – 17.00
Observability of Partial States of Invariant Systems
Christian Lageman
Abstract – This paper considers the observability of
partial states of an invariant control system on a Lie group.
Specifically we consider in this paper left invariant systems
where the outputs and the partial states are given by actions of
List of Abstracts
the group on different manifolds. Depending of the type of
these actions we give characterizations of partial state
observability and for some cases additional sufficient
observability criteria.
17.00 – 17.30
Error Models for Nonlinear Observers
Jochen Trumpf, Robert Mahony, Stefano Stramigioli
Abstract – We revisit the concept of observation error
for nonlinear observers of nonlinear systems. In order to
obtain a coordinate free notion of such an error we define it
using fiber bundles over the system manifold. The new notion
ties in nicely with Brockett’s and Willems’ classical description
of nonlinear systems as bundles as well as with the related
description of observers as bundles due to van der Schaft. It
identifies the nonlinear observer design problem as the
problem of choosing a nonlinear connection form (horizontal
distribution plus affine offset) in this bundle with certain
desirable properties. We demonstrate how a particular solution
to this connection design problem is given by the invariant
observers of Bonnabel, Martin and Rouchon. In the case of
systems on Lie groups we recover the recent results by
Lageman and two of the authors.
Sampled Control
(Regular Session)
Room 9
Chair: Emilia Fridman
15.30 – 16.00
Sampled-Data Stabilization of a Class of Parabolic Systems
Emilia Fridman, Anatoly Blighovsky
Abstract – A semilinear scalar heat equation with the
control input in the right-hand side, coupled to the
homogenous Dirichlet or Neumann boundary conditions, is
considered. Such a system represents a class of reactiondiffusion equations that model many physical phenomena. It is
well-known that this system is stabilizable by a linear infinitedimensional state-feedback. For realistic design, finitedimensional discrete version realizations may be applied
leading to local results. In the present paper we suggest a
sampled-data controller design, where the sampled-data (in
time) measurements of the state are taken in the finite number
of fixed (sampled) spatial variables. It is assumed that the
sampling in time and the sampling in space (i.e. the distance
between the consequently sampled spatial variables) are
bounded. Our sampled-data feedback with a constant gain is
piecewise-constant in time and in space. Sufficient conditions
for the exponential stabilization are derived in terms of Linear
Matrix Inequalities (LMIs) depending on the controller gain.
By solving these LMIs, upper bounds on the sampling in time
and on the sampling in space are found that preserve the
exponential stability. The results are extended to the sampleddata in space and to delayed in time sampled-data
measurements. A numerical example illustrates the efficiency
of the method.
16.00 – 16.30
Truncated Norms and Limitations on Signal Reconstruction
Gjerrit Meinsma, Hanumant Singh Shekhawat
Abstract – Design of optimal signal reconstructors over
all samplers and holds boils down to canceling frequency
bands from a given frequency response. This paper discusses
limits of performance of such samplers and holds and develops
methods to compute the optimal L²-norm.
16.30 – 17.00
On the Consistency of L²-Optimal Sampled Signal
Gjerrit Meinsma, Leonid Mirkin
Abstract – The problem of restoring an analog signal from
its sampled measurements is called the signal reconstruction
problem. A reconstructor is said to be consistent if the
resampling of the reconstructed signal by the acquisition
system would produce exactly the same measurements. The
consistency requirement is frequently used in signal processing
applications as the design criterion for signal reconstruction.
System-theoretic reconstruction, in which the analog
reconstruction error is minimized, is a promising alternative to
consistency-based approaches. The primary objective of this
paper is to investigate, what are conditions under which
consistency might be a byproduct of the system-theoretic
design that uses the L² criterion. By analyzing the L²
reconstruction in the lifted frequency domain, we show that
non-causal solutions are always consistent. When
causality constraints are imposed, the situation is more
complicated. We prove that optimal relaxedly causal
reconstructors are consistent either if the acquisition device is a
zero-order generalized sampler or if the measured signal is the
ideally sampled state vector of the antialiasing filter. In other
cases consistency can no longer be guaranteed as we
demonstrate by a numerical example.
17.00 – 17.30
Stability Analysis of Aperiodic Sampled-Data Control
Systems: an Improved Approach Using Matrix Uncertainty
Yasuaki Oishi, Hisaya Fujioka
Abstract – Stability analysis is considered on a sampleddata control system with an uncertain/time-varying sampling
interval. A stability condition is given in a linear matrix
inequality, readily tested with the interior-point method.
Conservatism of the condition can be reduced to any degree by
dividing the region of the possible sampling intervals.
Reduction of the computational complexity as well as
generalization to design of a state-feedback gain is considered.
This approach is a simplification of the previous approach of
the same authors and does not need the real Jordan canonical
form, which is difficult to compute for a large-sized matrix.
Algebraic Structures – 2 (Regular Session)
Room 10
Chair: Kenji Fujimoto
15.30 – 16.00
Variational Symmetry of Discrete-Time Hamiltonian
Systems and Learning Optimal Control
Kenji Fujimoto, Soraki Ogawa
Abstract – Variational symmetry is a property of
Hamiltonian control systems. It is the basis for learning
optimal control. In order to apply a continuous-time control
strategy to a real plant, we need to discretize the data with
respect to time. This paper investigates the variational
symmetry for the discretized Hamiltonian systems. It is proved
that the discretization based on the midpoint rule preserves the
variational symmetry of the original system. Furthermore, a
learning optimal control method based on the midpoint rule is
Tuesday, 6 July
16.00 – 16.30
Reproducing Kernels Preserving Algebraic Structure: a
Duality Approach
Tzvetan Ivanov, Balázs Csanád Csáji
Abstract – From the classical reproducing kernel theory
of function spaces it is well-known that there is an inverse
relationship between inner-products and kernels. In
applications, such as linear system theory and machine
learning, these kernels are often highly structured. In order to
exploit algebraic structure, it is common to choose basis
functions and fall back to matrix representations. However, the
basis has to be chosen in a way that is compatible with the
algebraic structure, which is itself a nontrivial task. We
therefore choose a different approach and use standard duality
theory where additional algebraic structures form no obstacle.
This is demonstrated by examples from linear system theory,
namely two variable polynomials given by Bézoutians and
quadratic differential forms.
16.30 – 17.00
A Spinor Approach to Port-Hamiltonian Systems
Johannes G. Maks
Abstract – The key concept in the implicit definition of
a port-Hamiltonian system is the geometric notion of a
Dirac structure. A Dirac structure is a maximal isotropic
subspace of the direct sum of the vector space of flows and its
dual space of efforts. There exist several ways to represent a
Dirac structure. One approach that appears to be unknown to
the systems and control community is the pure spinor
approach based on the work of É. Cartan. It is the purpose of
this paper to explain the spinor formalism and to show how a
port-Hamiltonian system is reformulated in the language of
17.00 – 17.30
On Computing Normalized Coprime Factorizations of
Periodic Systems
András Varga
Abstract – A numerically reliable state space algorithm
is proposed for computing normalized coprime factorizations
of periodic descriptor systems. A preprocessing step is used in
the algorithm to convert the initial problem for possibly noncausal systems into a simpler problem for causal periodic
systems. The main computational ingredient here is the
computation of a coprime factorization with causal factors
which is addressed by computing right annihilators of an
appropriately extended
manipulation algorithms. The solution of the normalized
coprime factorization problem for a causal system involves the
solution of a generalized periodic Riccati equation. The
proposed two steps approach is completely general, being
applicable to periodic systems with time-varying dimensions.
Delay Systems
(Regular Session)
Room 11
Chair: Lars Naujok
15.30 – 16.00
Lyapunov-Razumikhin and Lyapunov-Krasovskii Theorems
for Interconnected ISS Time-Delay Systems
Sergey Dashkovskiy, Lars Naujok
Abstract – We consider an arbitrary number of
interconnected nonlinear systems with time-delays and
investigate them in view of input-to-state stability (ISS). The
useful tools for single time-delay systems, the ISS Lyapunov-
functions and
functionals are redefined and applied to interconnected
systems. By the help of a smallgain condition we prove that the
whole system with time-delays has the ISS property, if each
subsystem has an ISS Lyapunov- Razumikhin function or ISS
Lyapunov-Krasovskii functional. Furthermore we construct
the ISS Lyapunov-Razumikhin (- Krasovskii) function(al) and
the corresponding gains of the whole system.
16.00 – 16.30
Monodromy Operator Approach to Time-Delay Systems:
Fast-Lifting Based Treatment of Operator Lyapunov
Tomomichi Hagiwara
Abstract – This paper establishes a new fundamental
framework of an operator-theoretic approach to linear timeinvariant (LTI) time-delay systems (TDSs). The state transition
of TDSs is first viewed in discrete-time and described by a
bounded operator called the monodromy operator. A stability
condition in terms of the spectral radius of the monodromy
operator is given, which in turn is related to an operator
Lyapunov inequality about that operator. A special class of
operators, parametrized by two finite-dimensional (FD)
constant matrices and constructed via the fast-lifting
technique, is then introduced, which is ensured to contain a
nonempty subset of the solutions to the operator Lyapunov
inequality whenever the TDS is stable and the fast-lifting
parameter N is large enough. A fundamental framework for
asymptotically exact stability analysis is thus established. An
equivalent scaling treatment is also shown, and further
generalization of the arguments is carried out with the use of
Legendre polynomials in the construction of the
above operator class. These arguments proceed in a rather
linear algebraic way with many similarities to the stability
analysis of FDLTI discrete-time systems. The presented
framework could be said to be a “pseudo-discretization”
technique; it allows one to essentially reduce the arguments on
infinite-dimensional operators to those about two matrices,
with an increasing degree of freedom as N gets larger, but
without introducing any matrix approximation of infinitedimensional operators.
16.30 – 17.00
Monodromy Operator Approach to Time-Delay Systems:
Numerical Method for Solving Operator Lyapunov
Tomomichi Hagiwara, Takayuki Inui
Abstract–This paper extends the technique for
stability analysis of time-delay systems based on the operator
Lyapunov inequalities
operators by establishing a numerical method for solving the
operator Lyapunov inequalities. First, quasi-finite-rank
approximation is applied to the fast-lifted monodromy
operator, and an operator Lyapunov inequality is considered
with respect to the resulting approximated operator. Then, a
numerically tractable method is developed for finding a
solution to the approximated operator Lyapunov inequality out
of a special class of operators that is known to be
nonconservative (with respect to the original operator
Lyapunov inequality before quasi-finite-rank approximation)
as long as the fast-lifting parameter N is large enough. The
finitedimensional matrices, and thus solving the operator
Lyapunov inequality amounts to solving a finite-dimensional
LMI. In particular, due to the discrete-time viewpoint intrinsic
to the (fast-lifted) monodromy operator approach, the resulting
LMI is a discrete-time one, which is suitable for extension to
the case with discrete-time controllers. A method is also
provided to confirm that the solution to the approximated
operator Lyapunov inequality does solve the original operator
Lyapunov inequality. Furthermore, it is shown that the overall
procedure gives an asymptotically exact numerical method for
List of Abstracts
stability analysis of time-delay systems. A numerical example
illustrating the arguments of the paper is also given. A brief
sketch is also provided on the extension to the use of
generalized hold and sampling operators JHk and JSk based
on Legendre polynomials.
17.00 – 17.30
Well-Posedness of Problems Involving Time-Varying Delays
Erik I. Verriest
16.30 – 17.00
Introduction on Bit Memory Systems: approximation and
Koji Tsumura
Abstract – In this paper, we introduce recent results on
bit memory systems. The bit memory systems are operators
from analog inputs to discretized outputs and their memories
are elements of discrete numbers. Their dynamics is
represented by the time evolution of the bit memories and
output equations. In this paper, we consider an approximation
problem or a stabilization problem of ordinary linear time
invariant systems by the bit memory systems. Then, we
consider the minimization problem of the bit length of the
memories with which the bit memory systems attain a given
error bound for the approximation problem or the stability of
the closed loop systems for the stabilization problem. We show
several upper and lower bounds of the bit length of memories
in both of the problems.
17.00 – 17.30
Global Optimal Control of Quantized Systems
Lars Grüne, Florian Müller
Basic and Recent Results on Quantized Control
(Invited Session)
Room 12
Organizers: Toshiharu Sugie, Shun-ichi Azuma
Chair: Toshiharu Sugie
Abstract – We propose a set oriented approach to the
global infinite horizon optimal control of nonlinear systems
with quantized state measurement and quantized control
values. The algorithm relies on a dynamic programming
principle in which the quantization error is modelled as an
opponent in a min-max dynamic game formulation. For the
solution of the problem we propose a set oriented approach
followed by a graph theoretic optimization algorithm. We also
discuss a dynamic feedback extension and illustrate the
performance of the proposed approach by experimental
15.30 – 16.00
Coarseness in Quantization for Stabilization of Linear
Systems over Networks
Hideaki Ishii, Koji Tsumura, Tomohisa Hayakawa
Abstract – We present an overview on the quantized
control approach that was initiated by Elia and Mitter in the
context of networked control. In particular, two recent
extensions are presented where uncertainties in the channel
and the plant are taken into consideration. The general
problem setting is that the amount of communication is
measured by a notion of coarseness in quantization, and we
would like to find the coarsest quantizer for achieving
stabilization. Numerical examples are provided to illustrate the
differences among the quantized control schemes.
16.00 – 16.30
Stabilising Stochastic Linear Plants via Erroneous Channels
Girish N. Nair, Kartik Venkat
Abstract – In the field of networked control, a
powerful methodology for constructing quantised controllers
with rigorous stability bounds is the zooming strategy. This
approach is attractive for implementation purposes since it
yields finitedimensional schemes. However, most available
results somewhat unrealistically require no transmission errors
to occur in the channel. In this paper, a novel zooming-like
scheme is proposed for controlling a stochastic linear plant
over an erroneous digital channel, with neither transmitter nor
receiver informed when errors occur. Using a stochastic
pseudo-norm technique, it is shown that mean square stability
can be achieved, provided that the number of quantisation
points is sufficiently larger than the plant dynamical constant
and the probability of symbol error is sufficiently small.
Control of Distributed Stochastic Systems
(Invited Session)
Room 13
Organizers: Jan H. van Schuppen, Charalambos D.
Chair: Jan H. van Schuppen
15.30 – 16.00
Decentralized Detection with Signaling
Ashutosh Nayyar, Demosthenis Teneketzis
Abstract – We consider a sequential problem in
decentralized detection. Two observers can make repeated
noisy observations of a binary hypothesis on the state of the
environment. At any time, any of the two observers can stop
and send a final message to the other observer or it may
continue to take more measurements. After an observer has
sent its final message, it stops operating. The other observer is
then faced with a different stopping problem. At each time
instant, it can decide either to stop and declare a final decision
on the hypothesis or take another measurement. At each time,
the system incurs an operating cost depending on the number
of observers that are active at that time. A terminal cost that
measures the accuracy of the final decision is incurred at the
end. We show that, unlike in other sequential detection
problems, stopping rules characterized by two thresholds on an
observer’s posterior belief no longer guarantee optimality in
this problem. Thus the potential for signaling among observers
alters the nature of optimal policies. We obtain a new
parametric characterization of optimal policies for this
Tuesday, 6 July
16.00 – 16.30
Performance Evaluation of Multi-Agent Distributed
Collaborative Optimization under Random Communication
Ion Matei, John S. Baras
Abstract – We investigate collaborative optimization in
a multi-agent setting, when the agents execute in a
distributed manner using local information, while the
communication topology used to exchange messages and
information is modeled by a graph-valued random process,
independent of other time instances. Specifically, we study the
performance of the consensus-based multi-agent subgradient
method, for the case of a constant stepsize, as measured by two
metrics: rate of convergence and guaranteed region of
convergence, evaluated via their expected values. Under a
strong convexity type of assumption, we provide upper bounds
on the performance metrics, which explicitly depend on the
probability distribution of the random graph and on the
agents’ estimates of the optimal solution. This provides a guide
for tuning the parameters of the communication protocol such
that good performance of the multi-agent subgradient method
is ensured.
16.30 – 17.00
On the Optimal Reconstruction Kernel of Causal Rate
Distortion Function
Charalambos D. Charalambous, Christos K. Kourtellaris,
Photios A. Stavrou
Abstract – This paper considers source coding of
general sources with memory, when causal feedback is
available at the decoder. The rate distortion function defined as
the infimum of the directed information between source and
reconstruction sequences over causal data compression
channels, which satisfy a distortion fidelity constraint. The
form of the optimal causal data compression Kernel is derived.
17.00 – 17.30
Control of the Observation Matrix for Control Purposes
René K. Boel, Jan H. van Schuppen
Abstract – How to control the activiation of an expensive
observation channel of a stochastic system? The control
objective is to reduce the conditional error variance of state
estimation but a cost is to be paid for acquiring a reading of the
channel. The optimal control law depends only on the
conditional error variance and has to be determined
computationally. A second problem is to use the state estimate
for control of the conditional mean. The solution method is
stochastic control with partial observations.
Crypto and Applications
(Invited Session)
Room 14
Organizers: Joachim Rosenthal, Marcus Greferath
Chair: Marcus Greferath
15.30 – 16.00
Problems Related to Combinatorial Configurations with
Applications to P2P-User Private Information Retrieval
Maria Bras-Amorós, Klara Stokes, Marcus Greferath
Abstract – We
combinatorial configurations have to peer-to-peer user-private
information retrieval and we analyze some problems that arise
from these applications. In particular we deal with the
existence of combinatorial configurations, the characterization
of optimal configurations for peer-to-peer user-private
information retrieval and the existence of configurations
preventing collusion attacks of dishonest users.
16.00 – 16.30
On Binary Sequences Generated by Self-Clock Controlled
Michele Elia, Guglielmo Morgari, Maria Spicciola
Abstract – The paper considers some peculiar properties of
binary sequences generated by self-clocked linear feedback
shift registers of maximum length, and compares these
properties with those of truly random sequences. In particular
it examines their periods, their 0-1 distributions, and their
linear complexity profiles.
16.30 – 17.00
On the Number of Linear Feedback Shift Registers with a
Special Structure
Srinivasan Krishnaswamy, Harish K. Pillai
Abstract – Given a linear recurring relation whose
characteristic polynomial is primitive, we find out the number
of possible realisations using Linear Feedback Shift Registers
(LFSRs) with 2-input 2-output delay elements. We show the
equivalence between each realisation and a matrix having a
special structure. Further, the number of realisations is
computed by calculating the number of these structured
17.00 – 17.30
Codes as Ideals over Some Pointed Hopf Algebras
Juan Cuadra, Jose M. García-Rubira, Juan A. López-Ramos
Abstract – We give a Decomposition Theorem for a family of
Hopf algebras containing the well-know family of Taft Hopf
algebras. Therefore, those indecomposable codes over
this family of algebras (cf. [4]) is an indecomposable code over
the studied case. We use properties of Hopf algebras to show
that dual (in the module sense) of an ideal code is again an
ideal code.
Max-Plus, Tropical and Idempotent Methods in
Control – 2
(Invited Session)
Room 15
Organizers: John S. Baras, William M. McEneaney
Chair: William M. McEneaney
15.30 – 16.00
The Correspondence between Tropical Convexity and Mean
Payoff Games
Marianne Akian, Stéphane Gaubert, Alexander Guterman
Abstract – We show that several decision problems
originating from max-plus or tropical convexity are equivalent
to zero-sum, two player game problems. In particular, we set
up an equivalence between the external representation of
tropical convex sets and zero-sum stochastic games, in which
tropical polyhedra correspond to deterministic games with
finite action spaces. Then, we show that the winning initial
positions can be determined from the associated tropical
polyhedron. We obtain as a corollary a game theoretical proof
of the fact that the tropical rank of a matrix, defined as the
maximal size of a submatrix for which the optimal assignment
problem has a unique solution, coincides with the maximal
number of rows (or columns) of the matrix which are linearly
independent in the tropical sense. Our proofs rely on
techniques from non-linear Perron-Frobenius theory.
List of Abstracts
16.00 – 16.30
Wednesday, 7 July
A Max-Plus Approach to the Approximation of Transient
Bounds for Systems with Nonlinear L2-Gain
Huan Zhang, Peter M. Dower
Plenary Lecture
Room 1
Chair: Joachim Rosenthal
09.00 – 10.00
Codes, Trellis Representations and the Interplay of System
Theory and Coding Theory
Heide Glüsing-Lürssen
16.30 – 17.00
Idempotent Methods for Control of Diffusions
Ben G. Fitzpatrick, Li Liu
Abstract – In this paper we discuss the application of maxplus arithmetic to stochastic control problems. The
dynamic programming equation is not max-plus linear in the
stochastic case, but a max-plus distributivity property permits
efficient value function and control computation. We illustrate
the technique by controlling the van der Pol equation.
17.00 – 17.30
Curse-of-Dimensionality-Free Control Methods for the
Optimal Synthesis of Quantum Circuits
Srinivas Sridharan, William M. McEneaney, Matthew R.
Abstract – In this article we introduce the use of a
recently developed numerical method from optimal control
theory to the gate synthesis problem. This technique helps
avoid the curse-ofdimensionality (COD) in the spatial
dimension inherent in mesh based numerical approaches to
solving control problems which also arise in optimal gate
synthesis. In the proposed algorithm there is a however a
growth in the complexity, related to the cardinality of the
discretized control set, which is managed via a pruning
technique. Hence an exponential speed up in the solution to a
large class of control problems on spin systems is obtained. The
reduced complexity method is then applied to obtain the
approximate solution to an example problem on SU(4)- a 15
dimensional system.
Abstract – We will present state realizations (trellises)
for convolutional codes and block codes and discuss their
system theoretic properties. For convolutional codes we will
give an application of such realizations by presenting a
MacWilliams Identity which relates the weight distribution of a
code to the weight distribution of the dual code. For block
codes, state realizations are based on the interpretation of the
code (of length n, say) as a set of admissible trajectories
(codewords) on the time axis 0, . . . , n − 1. After discussing
conventional trellis realizations, we will turn to tail-biting
trellises in which the time axis is considered circular and time
index arithmetic is performed modulo n. In this case the future
and past of trajectories are intertwined. One particular
consequence is that minimal tail-biting trellises are not unique.
We will show how to obtain all minimal tail-biting trellises for
a given code and discuss further properties.
Semiplenary Lecture
Room 1
Chair: William M. McEneaney
10.30 – 11.30
Learning Algorithms for Risk-Sensitive Control
Vivek S. Borkar
Abstract – This is a survey of some reinforcement
learning algorithms for risk-sensitive control on infinite
horizon. Basics of the risk-sensitive control problem are
recalled, notably the corresponding dynamic programming
equation and the value and policy iteration methods for its
solution. Basics of stochastic approximation algorithms are also
sketched, in particular the ‘o.d.e.’ approach for its stability and
convergence, and implications of asynchrony. The learning
schemes give stochastic approximation versions of the
traditional iterative schemes for solving dynamic programs.
Two learning schemes, Q-learning and the actor-critic method,
are described along with their convergence analysis. As these
‘ideal’ schemes suffer from ‘curse of dimensionality’, one needs
to use function approximation as a means to beat down the
dimension to manageable levels. A function approximation
based scheme is described for the simpler problem of policy
evaluation. Some future research directions are pointed out.
Semiplenary Lecture
Room 14
Chair: Lorenzo Finesso
10.30 – 11.30
Challenges of Tracking Single Molecules in Live Cells
Raimund Ober
Abstract – For centuries microscopy has played an
important role in biological investigations. Despite this long
history, quite recently microscopy investigations of cells have
undergone revolutionary developments. This is due to two
independent but equally important developments. Modern
Thursday, 8 July
molecular biology, and in particular the use of the green
fluorescent protein, allow from the highly specific labeling of
proteins in live cells. On the engineering side it is the
development of highly sensitive imaging detectors, combined
with computer control that allows images to be recorded with
unprecedented sensitivity. This had as a result that, over the
last decade, the imaging of single molecules in a live cell has
become possible.
Thursday, 8 July
This new discipline of single molecule microscopy poses a
significant number of exciting problems for engineers and
mathematicians. For example, the classical notions of
resolution limits for microscopes need to be re-evaluated.
Current experiments in our lab and others show that distances
at least one order below those predicted by Rayleigh’s criterion
can be determined. Estimation algorithms for locations of
single molecules are the core of the analysis of most single
molecule imaging approaches. The question therefore arises
how well these locations can be determined. We will present an
information theoretic approach to this and other problems in
single molecule microscopy.
Robust Control, Multidimensional Systems and Multivariable
Function Theory: Commutative and Noncommutative
Joseph A. Ball, Sanne ter Horst
Semiplenary Lecture
Room 15
Chair: Hans Zwart
10.30 – 11.30
Coping with Time Delays in Networked Control Systems
Hitay Özbay
Abstract – In
controller and plant exchange information over a
communication network, performance of the feedback system
depends on certain properties of the communication channels.
For example, packet loss, network delay and delay jitter have
negative effect on networked system performance. Depending
on the communication infrastructure, different mechanisms
are implemented to reduce the packet loss rate and network
induced delay in communication networks. In this paper, one
of these mechanisms, namely, buffer/queue management is
studied. It will be shown that techniques from robust control of
uncertain time delay systems can be used effectively. Simpler
low order controllers (PI and PID) are also considered. The
effect of controller parameters on various performance metrics
are illustrated.
Room 1
11.30 – 12.30
The Mathematical Challenge of Large Networks
László Lovász
Room 1
Chair: William J. Helton
09.00 – 10.00
Abstract – In the classical 1-D case there is a
seamless connection between state-space (time-domain)
representations and transfer-function (frequency-domain)
representations for linear systems. In particular, the first
results on H∞-control were developed in the frequency-domain
leading to an active exchange of ideas between mathematicians
with backgrounds in function theory and engineers. Eventually
elegant computational algorithms for solving the standard H∞control problem were found in terms of state-space
coordinates, first in terms of a pair of coupled Riccati
equations, and then completely in terms of Linear Matrix
Inequalities. Here we discuss two kinds of extensions of these
ideas to the context of multidimensional systems and
multivariable function theory, namely: (1) the commutative
case, where the transfer-function is a function of several
complex variables, and (2) the noncommutative case, where the
transfer-function is a function of noncommuting operator (or
matrix) variables. Perhaps surprisingly, we shall see that the
noncommutative setting provides a much more complete
parallel with the classical case than the commutative setting.
Many of the ideas of the present report are taken from our
survey article [17].
New Developments in Stochastic System
(Invited Session)
Room 2
Organizer: Yoshito Ohta
Chair: Yoshito Ohta
10.30 – 11.00
A Prediction-Error Identification Framework for Linear
Parameter-Varying Systems
Roland Tóth, Peter S.C. Heuberger, Paul M.J. Van den Hof
Distinguished Lecturer
Chair: György Michaletzky
Plenary Lecture
Abstract – Identification of Linear Parameter-Varying
(LPV) models is often addressed in an Input-Output (IO)
setting using particular extensions of classical Linear TimeInvariant (LTI) prediction-error methods. However, due to the
lack of appropriate system-theoretic results, most of these
methods are applied without the understanding of their
statistical properties and the behavior of the considered noise
models. Using a recently developed series expansion
representation of LPV systems, the classical concepts of the
prediction-error framework are extended to the LPV case and
the statistical properties of estimation are analyzed in the LPV
context. In the introduced framework it can be shown that
under minor assumptions, the classical results on consistency,
convergence, bias and asymptotic variance can be extended for
LPV predictionerror models and the concept of noise models
can be clearly understood. Preliminary results on persistency
of excitation and identifiability can also established.
List of Abstracts
11.00 – 11.30
11.00 – 11.30
Continuous-Time Subspace Identification in Closed-Loop
Marco Bergamasco, Marco Lovera
LMI Conditions for the Stability of 2D State-Space Models
Djillali Bouagada, Paul van Dooren
Abstract – This paper deals with the problem of
continuoustime model identification and presents a subspacebased algorithm capable of dealing with data generated by
systems operating in closed-loop. The algorithm is developed
by reformulating the identification problem from the
continuous-time model to an equivalent one to which discretetime subspace identification techniques can be applied. More
precisely, the considered approach corresponds to the
projection of the input-output data onto an orthonormal basis,
defined in terms of Laguerre filters. In this framework, the
PBSID subspace identification algorithm, originally developed
in the case of discrete-time systems, can be reformulated for
the continuoustime case. Simulation results are used to
illustrate the achievable performance of the proposed approach
with respect to existing methods available in the literature.
Abstract – In this paper we consider the problem of
stability of two-dimensional linear systems. New sufficient
conditions for the asymptotic stability are derived in terms of
linear matrix inequalities.
11.30 – 12.00
Spectral Density Function under Observation Outliers
Hideyuki Tanaka, Jaafar ALMutawa, Yoshito Ohta
Abstract – This paper deals with a problem of
identifying purely stochastic linear systems from stationary
Gaussian processes with observation outliers. Reviewing [1],
the spectral density function of the process under observation
outliers is derived, and it is compared with the one free from
outliers. It is shown that the estimates of the spectral density
function is enormously affected by observation outliers on the
frequency where the spectral density function is small.
12.00 – 12.30
Bias-Compensated State Space Model Identification
Kenji Ikeda, Yoshio Mogami, Takao Shimomura
Abstract – A method of bias compensation in subspace
identification method is proposed. The noise is assumed to be
colored with 0 mean and is assumed to be uncorrelated with
the input. The covariance matrix of the noise is estimated
directly from the residuals instead of estimating the noise
model. The proposing method becomes an iterative algorithm
but it converges with order 2.
Linear Matrix Inequalities
(Regular Session)
Room 3
Chair: Christian Ebenbauer
10.30 – 11.00
11.30 – 12.00
Dynamic Quantizer Synthesis Based on Invariant Set
Analysis for SISO Systems with Discrete-Valued Input
Kenji Sawada, Seiichi Shin
Abstract – This paper proposes analysis and synthesis
methods of dynamic quantizers for linear feedback single
input single output (SISO) systems with discrete-valued input
in terms of invariant set analysis. First, this paper derives
the quantizer analysis and synthesis conditions that clarify
an optimal quantizer within the ellipsoidal invariant set
analysis framework. In the case of minimum phase feedback
systems, next, this paper presents that the structure of the
proposed quantizer is also optimal in the sense that the
quantizer gives an optimal output approximation property.
Finally, this paper points out that the proposed design method
can design a stable quantizer for non-minimum phase feedback
systems through a numerical example.
12.00 – 12.30
Fixed-Order Output-Feedback Control Design for LTI
Systems: a New Algorithm to Reduce Conservatism
Emile Simon, Pedro Rodriguez-Ayerbe, Cristina Stoica,
Didier Dumur, Vincent Wertz
Abstract – This work proposes an algorithm to reduce
the conservatism of fixed-order output-feedback control
design for Linear Time Invariant (LTI) systems with Linear
Matrix Inequalities (LMIs)-representable objectives. Using
Lyapunov theory and the Schur complement many objectives
can be written as Bilinear Matrix Inequalities (BMIs), which in
general are hard to solve and have a non-convex space of
solutions. The classical response to this is to use LMIs
reformulation of BMIs, therefore using convex subspaces of the
non-convex space of all solutions and thus introducing
conservatism in general. Here a new use of a change of
variables is proposed, so that consecutive convex subspaces are
considered iteratively. This algorithm explores further the nonconvex space of solutions, leading to improved objectives with
reduced conservatism.
Optimization Methods in Systems and Control
(Invited Session)
Linear Matrix Inequalities for Normalizing Matrices
Christian Ebenbauer
Abstract – A real square matrix is normal if it can
be diagonalized by an unitary matrix. In this paper novel
convex conditions are derived to decide if a matrix is similar to
a normal matrix and it is shown how to use these conditions
to normalize matrices. The conditions are expressed in terms
of linear matrix inequalities and hence they are efficiently
solvable using semidefinite programming. Since a matrix is
similar to a normal matrix if and only if the matrix is
diagonalizable, the present results also provide novel convex
conditions for deciding if a matrix is diagonalizable or not.
Some applications of these results are presented.
Room 4
Organizer: Lars Grüne
Chair: Lars Grüne
10.30 – 11.00
A Sparse Stability Test for Sparse Matrices
Anders Rantzer
Abstract – In the study of distributed control systems, it is of
fundamental interest to understand how specifications on local
interconnections influence properties of the global system. In
this paper, we consider linear continuous time systems
described by a sparse matrix. The property of interest is
stability. In particular, for matrices with sparsity
structure corresponding to a chordal graph, we show
that conditions for Hurwitz stability can be written with the
same sparsity structure.
Thursday, 8 July
11.00 – 11.30
11.00 – 11.30
Delay-Optimal Global Feedbacks for Quantized Networked
Event Systems
Stefan Jerg, Oliver Junge
Recent Progress on Obtaining Matrix Convexity when the
Variables are Matrices
William J. Helton
Abstract – We extend a recent optimization based
technique for the construction of globally stabilizing optimal
controllers for quantized nonlinear event systems in a digital
network. To this end, we assume that the plant and the
controller possess synchronized clocks and that at each event
the (global) time stamp is transmitted from the event generator
to the controller. The new construction explicitely incorporates
the delay information, rendering the controller more robust.
The method is illustrated by means of the inverted pendulum
in a digital loop.
Abstract – The talk will focus an aspect of noncommutative
analysis and be designed to mesh with a four lecture tutorial
surveying that topic. Especially challenging is to develop a
theory of change of noncommuting variables. In classical
control, eg H∞ control, reducing problems to convex ones
typically requires a change of matrix variables. Matrix convex
problems are now known to be equivalent to LMI problems.
The goal is to develop a theory which helps delineate some
possibilities for changing variables to produce an LMI. There
are results in special situations along these lines. Various
aspects of the work are done jointly with Jeremy Greene, Igor
Klep, Victor Vinnikov and Scott McCullough.
11.30 – 12.00
Efficient Model Predictive Control for Linear Periodic
Andreas Freuer, Marcus Reble, Christoph Böhm, Frank
Abstract – This
predictive control scheme for the stabilization of constrained
linear periodically time-varying systems. The results are based
on an existing Model Predictive Control scheme for uncertain
linear systems using linear matrix inequalities. A predetermined periodic feedback control law is used in
combination with superimposed free control moves as
additional degrees of freedom. Only the additional free control
moves are calculated online taking advantage of pre-computed
periodic invariant sets. Two simple algorithms are presented
for calculating offline ellipsoidal or polyhedral periodic
invariant sets. Since only a small number of free control moves
is calculated online by solving a convex optimization problem
after each time period, the computational cost can be reduced
significantly compared to existing schemes.
11.30 – 12.00
A Nullstellensatz for a Class of Two Sided Ideals of
Noncommutative Polynomials
Victor Vinnikov
12.00 – 12.30
Simple Homothetic Tube Model Predictive Control
Saša V. Raković, Basil Kouvaritakis, Rolf Findeisen, Mark
Abstract – This paper considers the robust model
predictive control synthesis problem for constrained linear
discrete time systems. The manuscript introduces a simple
homothetic tube model predictive control synthesis method.
The proposed method employs several novel features
including: a more general parameterization of the state and
control tubes based on homothety and invariance; a more
flexible form of the terminal constraint set; and a relaxation of
the dynamics of the sets that define the state and control tubes.
Under rather mild assumptions it is demonstrated that the
proposed method is computationally efficient while it induces
strong system theoretic properties.
Real Algebraic Geometry and Applications – 2
(Invited Session)
Room 6
Organizers: William J. Helton, Pablo Parrilo
Chair: William J. Helton
10.30 – 11.00
It is Convexity and Positivity of Polynomials
Pablo Parrilo
12.00 – 12.30
Relaxing LMI Domination Matricially
Igor Klep
List of Abstracts
12.00 – 12.30
Observers for DPS Back and Forth in Time
George Weiss, Karim Ramdani, Marius Tucsnak
This is a joint work with J. William Helton and Scott McCullough.
The Semigroup Approach to DPS
(Invited Session)
Room 7
Organizer: George Weiss
Chair: George Weiss
10.30 – 11.00
Analytic Properties of Matrix Riccati Equation Solutions
Ruth F. Curtain, Leiba Rodman
Abstract – For
platoontype systems and of systems arising from PDEs,
assuming the coefficients are analytic functions in a
suitable domain, the analyticity of the stabilizing solution
is proved under various hypotheses. In addition, general results
on the analytic behavior of stabilizing solutions are developed.
11.00 – 11.30
Structurally Damped Plate Equations with Random Point
Roland Schnaubelt
Riccati and Sylvester-Equations
(Regular Session)
Room 8
Chair: Timo Reis
10.30 – 11.00
Lur'e Equations and Singular Optimal Control
Timo Reis
Abstract – In this work we give an overview about
Lur’e matrix equations, linear-quadratic infinite time horizon
optimal control problems and their connections to the
eigenstructure of certain even matrix pencils. We characterize
the set of solutions in terms of deflating subspaces of even
matrix pencils. In particular, it is shown that these special
solutions can be constructed deflating subspaces of even matrix
11.00 – 11.30
Reduction of State Variables Based on Regulation and
Filtering Performances
Hidenori Shingin, Yoshito Ohta
11.30 – 12.00
A Spectral Approach to the Null-Controllability of Diffusion
Gérald Tenenbaum, Marius Tucsnak
Abstract – This paper provides a component analysis for the
state variables of stable linear disrete-time systems based on
the control and estimation performance criteria. In the optimal
regulation and filtering problems, the trade-offs between the
dimension of the control or estimation law and degree of
performance degenerations are invariants given as the
eigenvalues of the matrices depending on the solutions of both
Lyapunov and Riccati equations. This analysis shows the
dominant components of the state variables which have major
contribution to enhance the performances.
Thursday, 8 July
11.30 – 12.00
11.00 – 11.30
An Explicit Dynamic Programming Solution for a
Decentralized Two-Player Optimal Linear-Quadratic
John Swigart, Sanjay Lall
Guaranteed Cost for Control of Networked Control Systems
with Small Time-Varying Delays
Marc Jungers, Laurentiu Hetel, Jamal Daafouz
Abstract – We develop optimal controller synthesis
algorithms for decentralized control problems, in which
individual sub- systems are connected over a network. We
consider a simple information structure, consisting of two interconnected
optimal controller subject to a decentralization constraint via
a novel dynamic programming method. We provide ex- plicit
state-space formulae for the optimal controller, and show that
each player has to do more than simply esti- mate the states
that they cannot observe. In other words, the simplest
separation principle does not hold for this decentralized
control problem.
12.00 – 12.30
A State-Space Solution of Bilateral Diophantine Equations
over RH∞
Maxim Kristalny, Leonid Mirkin
Abstract – This paper studies a class of real-rational
matrix bilateral Diophantine equations (BDE) arising in
numerous control problems. A necessary and sufficient
solvability condition is derived in terms of state-space
realizations of rational matrices involved in the equation. This
condition is given in terms of a constrained matrix Sylvester
equation and is numerically tractable. An explicit state-space
parametrization of all solutions is also derived. This
parameterization effectively includes two parameters: one is a
“standard” RH∞ parameter and another one arises if the
Sylvester equation is non-uniquely solvable. A condition, in
terms of zeros of rational matrices involved in the BDE, is
found under which the Sylvester equation has a
unique solution and, hence, the parametrization is affine in a
single RH∞ parameter.
Network Stability
(Regular Session)
Room 9
Chair: Ulf T. Jönsson
10.30 – 11.00
Primal and Dual Criteria for Robust Stability
Ulf T. Jönsson
Abstract – Primal and dual formulations of stability
criteria based on integral quadratic constraints (IQC) are
discussed. The foundation for IQC based stability analysis is to
use a convex cone of multipliers to characterize the uncertainty
in a system. The primal and dual stability criteria are
formulated as convex feasibility tests involving the nominal
dynamics and multipliers from the cone and the polar cone,
respectively. The motivation for introducing the dual is that it
provides additional insight into the stability criterion and is
sometimes easier to use than the primal.
The case considered in this paper is when the
uncertainty represents the interconnection of a complex
network. The multipliers are used to describe characteristic
properties of the network such as the spectral location or the
structure of the underlying graph.
Abstract – This paper deals with the control design of
a networked control systems subject to small time-varying
delays, which takes into account the performance aspect.
The considered approach consists in using a limited Taylor
series expansion for the uncertain parameter exponential
matrix. An optimization problem involving LMI constraints is
proposed to obtain the minimal quadratic upper bound of the
guaranteed cost. An example illustrates our approach and is
the base for a extended discussion.
11.30 – 12.00
A Novel Discontinuous Lyapunov Functional Approach to
Networked-Based Stabilization
Kun Liu, Vladimir Suplin, Emilia Fridman
Abstract – This paper presents a new stability analysis
of Networked Control Systems (NCSs), where the sampling
and the constant network-induced delays are taken into
account. The new method is inspired by discontinuous
Lyapunov functionals that were recently introduced for
sampled-data systems in [1] (in the framework of impulsive
system representation) and in [2] (in the framework of input
delay approach). However, extensions of the above Lyapunov
constructions to NCSs lead to complicated conditions, which
become conservative if the network-induced delays is not small.
In the present paper a novel discontinuous Lyapunov
functional is introduced, which is based on the application of
the Wirtinger type inequality. This functional leads to efficient
stability conditions in terms of Linear Matrix Inequality
(LMIs). The new stability analysis is applied to sampled-data
stabilization by using artificial delay.
12.00 – 12.30
A Multichannel IOpS Small-Gain Theorem for Large Scale
Rudolf Sailer, Fabian Wirth
Abstract – This paper extends known results on the
stability analysis of interconnected systems. In particular, a
small-gain theorem for the interconnection of an arbitrary
number of systems via communication channels is presented.
Here the communication between the subsystems is over a
delayed, possibly lossy communication channel. To this end, a
notion of input-to-output stability for functional differential
equations is applied.
(Regular Session)
Room 10
Chair: Svyatoslav S. Pavlichkov
10.30 – 11.00
Further Remarks on Global Stabilization of Generalized
Triangular Systems
Sergey Dashkovskiy, Svyatoslav S. Pavlichkov
List of Abstracts
11.00 – 11.30
11.00 – 11.30
Invariance Results Involving Non-Monotonic Lyapunov
Functions with Applications
Anthony N. Michel, Ling Hou
Adaptive Control of Port-Hamiltonian Systems
Daniel A. Dirksz, Jacquelien M.A. Scherpen
Abstract – The invariance result for asymptotic stability
of autonomous systems involves positive definite Lyapunov
functions which are non-increasing along the systems’ motions.
We establish a result for discontinuous non-autonomous
systems with positive definite Lyapunov functions which are
nonincreasing only on an unbounded discrete set, resulting in
nonmonotonic Lyapunov functions. We show that the
invariance result for asymptotic stability reduces to the
invariance result reported herein. We apply this result in the
stabilization of conservative mechanical systems.
Abstract – In this paper an adaptive control scheme is
presented for general port-Hamiltonian systems. Adaptive
control is used to compensate for control errors that are caused
by unknown or uncertain parameter values of a system. The
adaptive control
transformation theory for port-Hamiltonian systems. This
allows for the adaptive control to be applied on a large class of
systems and for being included in the port-Hamiltonian
11.30 – 12.00
Model Reduction of Port-Hamiltonian Systems as Structured
Rostyslav V. Polyuga, Arjan J. van der Schaft
11.30 – 12.00
A New Characterisation of Exponential Stability
Elena Panteley, Antonio Loría
Abstract – We present a new characterization of
exponential stability for nonlinear systems in the form of
Lyapunov functions which may be upper and lower bounded
by monotonic functions satisfying a growth order relationship
rather than being polynomials of the state’s norm. In
particular, one may allow for Lyapunov functions with
arbitrary weakly homogeneous bounds.
12.00 – 12.30
Sufficient Conditions for Local Asymptotic Stability and
Stabilization for Discrete-Time Varying Systems
Angeliki Stamati, John Tsinias
Abstract – The purpose of this paper is to establish
sufficient conditions for local asymptotic stability and feedback
stabilization for discrete-time systems with time depended
dynamics. Our main results constitute generalizations of those
developed by same authors in a recent paper, for the case of
continuoustime systems.
Port-Hamiltonian Systems
(Regular Session)
Room 11
Chair: Arjan J. van der Schaft
10.30 – 11.00
Port-Hamiltonian Systems on Open Graphs
Arjan J. van der Schaft, Bernhard M. Maschke
Abstract – In this talk we discuss how to define in an
intrinsic manner port-Hamiltonian dynamics [3] on open
graphs. Open graphs are graphs where some of the vertices are
boundary vertices (terminals), which allow interconnection
with other systems. We show that a directed graph carries two
natural Dirac structures [3], called the Kirchhoff-Dirac
structure and the vertex-edge Dirac structure. The portHamiltonian dynamics corresponding to the Kirchhoff-Dirac
structure is exemplified by the dynamics of an RLC-circuit, see
also [5], [4]. The port-Hamiltonian dynamics corresponding to
the vertex-edge Dirac structure is illustrated by coordination
control, in which case there is dynamics associated to every
vertex and to every edge, and by standard consensus
algorithms where there is dynamics associated to every vertex
while every edge corresponds to a resistive relation.
Abstract – The goal of this work is to demonstrate that a
which provides an H2 error bound, turns out to be applicable
to port- Hamiltonian systems, preserving the port-Hamiltonian
structure for the reduced order model, and, as a consequence,
12.00 – 12.30
Modeling for Control of an Inflatable Space Reflector, the
Linear 2-D Case
Thomas Voss, Jacquelien M.A. Scherpen
Abstract – In this paper we develop a mathematical
model for the dynamics of a linear plate with piezoelectric
actuation. This model can then be used to design controllers
with the goal of achieving a desired shape of the plate. This
control scheme can be used for several applications, e.g.,
vibration control in structures or shape control for high
precision structures like inflatable space reflectors. The
starting point of the control design is modeling for control. We
will do this in the framework of port-Hamiltonian (pH)
modeling, since the pH modeling framework has very nice
properties which can be exploited if one wants to design a
controller for a specific task. One property for example is that
it facilitates modeling multi physics systems or systems which
consist of several systems by first modeling all parts separate
and then interconnecting them. This is possible because any
interconnection of pH systems yields again a pH system. Hence,
the pH framework is useful for our multi-domain modeling
Stochastic Control – 1
(Invited Session)
Room 13
Organizers: Bozenna Pasik-Duncan, Tyrone E. Duncan
Chair: Bozenna Pasik-Duncan
10.30 – 11.00
Social Certainty Equivalence in Mean Field LQG Control:
Social, Nash and Centralized Strategies
Minyi Huang, Peter E. Caines, Roland P. Malhamé
Abstract – We study social decision problems and
Nash games for a class of linear-quadratic-Gaussian (LQG)
models with N decision makers possessing different dynamics.
For the social decision case, the basic objective is to minimize a
social cost as the sum of N individual costs containing mean
requires centralized information. Continuing from the previous
work (Huang, Caines, and Malhamé, 2009 Allerton
we develop
Thursday, 8 July
optimization so that each agent only uses its own state and a
function which can be computed off-line.We prove asymptotic
social optimality results with general vector individual states
and continuum dynamic parameters. In finding the asymptotic
social optimum, a key step is to let each agent optimize a new
cost as the sum of its own cost and another component
capturing its social impact on all other agents. We also discuss
the relationship between the socially optimal solution and the
so-called Nash Certainty Equivalence (NCE) based solution
presented in previous work on mean field LQG games, and for
the NCE case we illustrate a cost blow-up effect due to the
strength of interaction exceeding a certain threshold.
11.00 – 11.30
Idempotent Algorithms for Continuous-Time Stochastic
Hidehiro Kaise, William M. McEneaney
Abstract – Previously, idempotent methods have been found
to be extremely fast for solution of dynamic
programming equations
control problems. The original methods exploited the
idempotent (e.g., max-plus) linearity of the associated
semigroup operator. However, it is now known that the curseofdimensionality-free idempotent methods do not require this
linearity, and may be used to solve some classes of stochastic
control problems. The key is the use of the idempotent
distributive property. This was previously demonstrated for a
class of discrete-time stochastic control problems. Here, we
extend this approach to a class of continuous-time stochastic
control problems.
11.30 – 12.00
Tuning the TCP Timeout Mechanism in Wireless Networks to
Maximize Throughput via Stochastic Stopping Time Methods
George Papageorgiou, John S. Baras
Abstract – We present an optimization problem that aims
to maximize the throughput of a Transmission Control
Protocol (TCP) connection between two nodes in a wireless adhoc network. More specifically, a persistent TCP connection
is established between two nodes that are one hop away in
a wireless unslotted Aloha network. The optimization is over
the TCP timeout period, i.e. the problem is to find the
optimal waiting period before the TCP sender declares a
timeout event in the absence of a received acknowledgment for
a transmitted packet. The problem is formulated as an optimal
stopping problem. In the absence of a tractable analytical
solution to the problem, a numerical method is proposed to
achieve performance improvement of the system.
12.00 – 12.30
Control of Linear Systems with Fractional Brownian Motions
Tyrone E. Duncan, Bozenna Pasik-Duncan
Abstract – In this paper a control problem for
a multidimensional linear stochastic system with a vector of
fractional Brownian motions and a cost functional that
is quadratic in the state and the control is solved. An
optimal control is given explicitly using the (RiemannLiouville) fractional calculus and the control is shown to be the
sum of a prediction of the optimal system response to the
future fractional Brownian motion and the well known
linear feedback control for the deterministic linear-quadratic
control problem. It is noted that the methods to obtain an
optimal control extend to other noise processes with continuous
sample paths and finite second moments.
Algebraic Systems Theory, Behaviors, and
Codes: Recent Approaches to New System
(Invited Session)
Room 14
Organizers: Eva Zerz, Heide Glüsing-Lürssen
Chair: Eva Zerz
10.30 – 11.00
Elimination, Fundamental Principle and Duality for Analytic
Linear Systems of Partial Differential-Difference Equations
with Constant Coefficients
Henri Bourlès, Ulrich Oberst
Abstract – Partial differential-difference equations are
the multidimensional generalization of ordinary delaydifferential equations. We investigate behaviors of analytic
signals governed by equations of this type, i.e., solution
modules of linear systems with constant coefficients of such
equations, and especially the problems of elimination and
duality. The first concerns the question whether the images of
behaviors are again behaviors and in particular the existence of
solutions of inhomogeneous linear systems which satisfy the
obvious necessary compatibility conditions. Duality refers to
the determination of the module of equations by the behavior.
Our theory is presently restricted to analytic signals because
the proofs make substantial use of the Stein algebra of
multivariate entire functions and of Stein modules over it, but
the extension to smooth or distributional signals is of course an
important task for the future. We especially prove the validity
equations with
incommensurate delays and thus solve, for analytic signals, an
open problem stated by Glüsing-Lürssen, Vettori
and Zampieri. Duality is expressed and derived by means of
the polar theorem for locally convex spaces in duality.
Gluesing-Luerssen’s rather complete and seminal behavioral
theory of delay-differential equations with commensurate
delays relies on the fact that the appropriate ring of operators
is a Bezout domain and especially coherent. Coherence of the
relevant rings of operators in the more general situations is
important, but has not yet been proven. Further contributors
to the module theoretic or behavioral approach to delaydifferential equations are Fliess, Habets, Mounier, Rocha,
Willems et al.
11.00 – 11.30
Symmetries, Parametrizations and Potentials of
Multidimensional Linear Systems
Thomas Cluzeau, Alban Quadrat
Abstract – Within the algebraic analysis approach to
linear systems theory, the purpose of this paper is to study how
left D- homomorphisms between two finitely presented left Dmodules associated with two linear systems induce natural
transformations on the autonomous elements of the two
systems and on the potentials of the parametrizations of the
parametrizable subsystems. Extension of these results are also
considered for linear systems inducing a chain of successive
List of Abstracts
11.30 – 12.00
Semiplenary Lecture
Degree Structures of Polynomial Vector Modules with
Applications to Systems over Fields and Rings
Margreta Kuijper
Room 1
Chair: Michael A. Demetriou
14.00 – 15.00
Dynamic "Magic" Graphs in Cooperative Networked Systems
John S. Baras
Semiplenary Lecture
Room 14
Chair: Tryphon T. Georgiou
14.00 – 15.00
12.00 – 12.30
On Periodic Solutions of Linear Difference Equations
Eva Zerz
Abstract – We study systems of linear difference
equations with constant coefficients in a commutative quasiFrobenius ring F, that is, F is Noetherian and self-injective. For
instance, F could be a field or a residue class ring of the
integers. Given a positive integer p, we first answer the
following basic questions: Does there exist a p-periodic
solution? When are all solutions p-periodic? Then we address
the more interesting question of how to determine candidates
for the period p. We characterize strong autonomy (i.e., finitely
many initial data) and weak autonomy (i.e., no free variables),
which are nonequivalent concepts, in general. If F is finite, all
trajectories of a strongly autonomous system eventually
become periodic, and we characterize the case where they are
purely periodic (i.e., no pre-period), as well as the minimal
period in this case. These methods can be applied to
systems as
Kuijper/Willems and Aleixo/Polderman/Rocha, and the
question whether a p-periodic system admits (only) p-periodic
solutions can be tackled using the known lifting technique to
shiftinvariant system.
Orthogonal Rational Functions and Non-Stationary
Stochastic Processes: a Szegö Theory
Laurent Baratchart
(based on joint work with L. Golinskii, S. Kupin, M. Olivi and
V. Lunot)
Abstract – We present a generalization of Szegö theory
of orthogonal polynomials on the unit circle to orthogonal
rational functions. Unlike previous results, the poles of the
rational functions may tend to the unit circle under smoothness
assumptions on the density of the measure. Just like the
Kolmogorov-Krein- Szegö theorem may be interpreted as an
asymptotic estimate of the prediction error for stationary
stochastic processes, the present theory yields an asymptotic
estimate of the prediction error for certain, possibly
nonstationary, stochastic processes. The latter admit a spectral
calculus where the time-shift corresponds to multiplication by
elementary Blaschke products of degree 1 (that reduce to
multiplication by the independent variable in the stationnary
case). When the poles of the best predictor tend to a point on
the unit circle where the spectral density is nonzero, the
prediction error goes to zero, i.e. the process is asymptotically
Semiplenary Lecture
Room 15
Chair: Andrea Gombani
14.00 – 15.00
Geometric Control Theory for Linear Systems–1
Mathematical Finance with Heavy-Tailed Distributions
Mathukumalli Vidyasagar
Room 15
Organizers: Giovanni Marro, Fabio Morbidi, Lorenzo
Ntogramatzidis, Domenico Prattichizzo
Stochastic Adaptive Systems
(Regular Session)
Room 2
10.30 – 12.30
Geometric Control Theory for Linear Systems
Giovanni Marro, Fabio Morbidi, Lorenzo Ntogramatzidis,
Domenico Prattichizzo
Abstract – This paper reviews in a condensed form the
main tools and results of the geometric approach developed in
the last forty years. Because of the vastness of the subject,
this tutorial does not pretend to be exhaustive, and more
emphasis will be given to selected topics and to the related
computational tools. The authors hope their effort to provide a
unified view of geometric control theory may be profitable to
awake renewed interest in this research field.
Chair: Mario di Bernardo
15.30 – 16.00
Synchronization and Pinning Control of Networks via
Adaptation and Edge Snapping
Pietro DeLellis, Mario di Bernardo, Maurizio Porfiri
Abstract – In this paper, we propose novel adaptive
pinning control strategies for synchronization of complex
networks. The novelty of theses approaches is the adaptive
selection of pinned nodes along with the fully decentralized
adaptation of the coupling and control gains. The effectiveness
of the proposed strategies is validated with numerical
simulations on a testbed example.
Thursday, 8 July
16.00 – 16.30
used to derive efficient eigenvalue methods for linear firstorder statespace systems.
Networked Adaptive Model Predictive Control
João M. Lemos
16.00 – 16.30
Abstract – This paper presents a neighborhood based
networked predictive adaptive controller that is suitable for
control of multivariable systems made of subsystems that
interact sequentially and on its demonstration using
simulations in a detailed nonlinear model of a water delivery
canal. Furthermore some convergence properties of the
adaptive algorithm are proved.
According to the approach followed, the overall system is
decomposed in local systems. To each local system, a local
adaptive predictive controller that manipulates its input is
associated. In order to improve the overall performance,
communication between local controllers is provided through
feedforward terms from adjacent local systems.
16.30 – 17.00
Approximate Solution to Nonlinear Optimal Regulator
Problem Using Quantum and Stochastic Theories
Yuki Nishimura, Yuji Wakasa, Kanya Tanaka
Abstract – This paper proposes a new constraction scheme
of nonlinear optimal regulator by using two methods.
Nishimura and Yamashita has introduced the method of
obtaining approximate Lyapunov functions for deterministic
and stochastic systems, which is based on difference
apporximation with directions and quantization of Markov
processes. On the other hand, Itami has proposed the method
replacing the problem of obtaining nonlinear optimal
regulators by the issue of solving Schrödinger equations. We
combine above two schemes for constructing our new method.
17.00 – 17.30
A New Unscented Kalman Filter with Higher Order MomentMatching
Ksenia Ponomareva, Paresh Date, Zidong Wang
Abstract – This paper is concerned with filtering
nonlinear multivariate time series. A new approximate
Bayesian algorithm is proposed which generates sample points
and corresponding probability weights that match exactly the
predicted values of average marginal skewness and average
marginal kurtosis of the unobserved state variables, in addition
to matching their mean and the covariance matrix. The
performance of the algorithm is illustrated by an empirical
example of yield curve modelling with real financial market
data. Results show an improvement in accuracy in comparison
with extended Kalman filter (EKF) and traditional unscented
Kalman filter (UKF).
Algebraic Properties of Riccati Equations
Ruth F. Curtain
16.30 – 17.00
Parametrization of Stabilizing Controllers with Fixed
Kazuyoshi Mori
Abstract – In the framework of the factorization
approach, we give a parameterization of a class of stabilizing
controllers. This class is characterized by some fixed strictly
causal precompensators. As applications, we present the
parameterization of all causal stabilizing controllers including
the some fixed number or more integrators, and the
parameterization of all strictly causal stabilizing controllers
which has the some fixed number or more delay operators.
17.00 – 17.30
Output Regulation of Distributed Parameter Systems with
Time-Periodic Exosystems
Lassi Paunonen, Seppo Pohjolainen
Abstract – In this paper the output regulation of a
linear distributed parameter system with a nonautonomous
periodic exosystem is considered. It is shown that the
solvability of the output regulation problem can be
characterized by an infinite- dimensional Sylvester differential
equation. Conditions are given for the existence of a controller
solving the regulation problem along with a method for its
New Mathematical Methods in Multidimensional
Systems Theory – 3
(Invited Session)
Room 4
Organizers: Alban Quadrat, Thomas Cluzeau
Chair: Alban Quadrat
Algebraic Systems
(Regular Session)
15.30 – 16.00
Room 3
Chair: Tobias Brüll
15.30 – 16.00
Linear Quadratic Optimal Control, Dissipativity, and ParaHermitian Matrix Polynomials
Tobias Brüll
Abstract – In this paper we will look at two results in which a
special para-Hermitian matrix polynomial appears in
linear quadratic systems theory. The first result constitutes the
first step in a dissipativity check. The second result shows
that dissipativity is equivalent to the solvability of the
infinitehorizon linear quadratic optimal control problem and
that its solutions are given by the behavior specified by the
special para-Hermitian matrix polynomial. The results can be
Conley Index Theory
Mohamed Barakat, Stanislaus Maier-Paape
Abstract – Conley’s index theory provides powerful tools
to prove either the existence or the nonexistence of
connecting orbits between equilibria of the dynamical systems
under consideration. Conley’s idea was to relate the local and
the global topological properties of the dynamical system by an
algebraic object called the connection matrix. The structure of
this matrix imposes serious restrictions on the possible
configurations of local and global topological data. These
restrictions can now be utilized to derive unknown properties
of the system out of known ones.
List of Abstracts
16.00 – 16.30
The Significance of Gabriel Localization for Stability and
Stabilization of Multidimensional Input/Output Behaviors
Ulrich Oberst
Abstract–Serre (1953) was the first who considered
categories of groups up to negligible ones, negligibility
being determined by the considered context. Gabriel in his
thesis (1962) developed this idea into a complete theory of
quotient categories, rings and modules which is called
Gabriel localization. This theory was nicely exposed by
Stenström (1975). The author has recently observed that this
theory is a valuable tool for stability and stabilization of
multidimensional behaviors where a finitely generated
multivariate polynomial torsion module is considered
negligible if its characteristic variety has points in a preselected
stability region only or, equivalently, if its associated
autonomous behavior is stable, i.e., has polynomial-exponential
solutions with frequencies in this stability region only. Via the
Integral Representation Formula of Ehrenpreis/Palamodov
corresponding properties hold for all trajectories of the
behavior. In the one-dimensional standard cases stability
signifies asymptotic stability. In our approach to output
feedback stabilization of multidimensional systems almost
direct decompositions, i.e., direct sum decompositions up to
negligible modules, are essential. Quadrat had observed the
stabilization theory. These decompositions are usually hidden
in coprime factorizations of transfer matrices which, however,
do not always exist. Bisiacco, Valcher and Napp Avelli studied
almost direct decompositions for two-dimensional polynomial
modules and behaviors, but without using localization theory.
This paper explains Gabriel’s localization and quotient
modules and their use in multidimensional stabilization theory.
It also contains a new algorithm for the computation of the
(Gabriel) quotient module of a finitely generated torsionfree
module over a multivariate polynomial ring. This algorithm
can also be used for the computation of Willems closures of
such modules and thus generalizes work of Shankar, Sasane,
Napp Avelli, van der Put et al.. It is also useful for the
computation of the purity filtration of a finitely generated
polynomial torsion module which is the subject of Barakat’s
talk at this conference where also the history of this filtration is
discussed. The author gratefully acknowledges financial
support of the Austrian FWF.
16.30 – 17.00
Purity Filtration and the Fine Structure of Autonomy
Mohamed Barakat
17.00 – 17.30
Controlled Invariant Varieties of Polynomial Control
Eva Zerz, Sebastian Walcher, Fadime Güçlü
Abstract – We study input-affine control systems with
polynomial nonlinearity. A variety V is said to be controlled
invariant if there exists a feedback law of polynomial type that
causes the closed loop system to have V as an invariant variety,
which means that any trajectory starting in V will remain
there for all times. Using the theory of Gröbner bases, we show
(under certain conditions on the given representation of V)
how to constructively decide whether a given variety is
controlled invariant for a given system, and if so, how to
determine all feedback laws achieving the task.
Real Algebraic Geometry and Applications – 3
(Invited Session)
Room 6
Organizers: William J. Helton, Pablo Parrilo
Chair: William J. Helton
15.30 – 16.00
A Class of Forms which are PSD iff are SOS
Carla Fidalgo, Alexander Kovačec
16.00 – 16.30
Positivity in Power Series Rings
Salma Kuhlmann
Abstract – We extend and generalize the results of
Scheiderer (2006) on the representation of polynomials
nonnegative on two-dimensional basic closed semialgebraic
sets. Our extension covers some situations where the defining
polynomials do not satisfy the transversality condition. Such
situations arise naturally when we consider semialgebraic sets
invariant under finite group actions.
16.30 – 17.00
Relationship between Nonnegative Forms, Convex Forms
and Sums of Squares: Faces and Volumes
Grigoriy Blekherman
Abstract – Sums of Squares (SOS) relaxations are often
used as a substitute for testing whether a polynomial is
nonnegative. Testing whether a polynomial is SOS is a Semidefinite Programming problem and thus computationally
tractable. We use techniques from convex geometry to study
the relationship between sums of squares and non-negative
polynomials. We will present results about the boundary
structure and quantitative relationship between SOS and nonnegative polynomials, which allow us to understand the quality
of SOS relaxations. We use similar techniques to study the
relationship between convex and non-negative forms. Although
we show that there exist convex forms that are not sums of
squares, there are currently no explicit examples.
17.00 – 17.30
Semidefinite Representation of Convex Hulls of Rational
Didier Henrion
Abstract – Using
positive semidefinite moment matrices and polynomial sum-ofsquares decompositions, we prove that the convex hull of
rationally parameterized algebraic varieties is semidefinite
representable (that is, it can be represented as a projection of
an affine section of the cone of positive semidefinite matrices)
in the case of (a) curves; (b) hypersurfaces parameterized by
quadratics; and (c) hypersurfaces parameterized by bivariate
quartics; all in an ambient space of arbitrary dimension.
Thursday, 8 July
Operator Theoretic Approaches to DPS
(Invited Session)
Room 7
Organizer: George Weiss
Chair: George Weiss
15.30 – 16.00
Operator Splitting and Evolution Equations
Andras Batkai, Petra Csomos, Balint Farkas, Klaus-Jochen
Engel, Gregor Nickel
16.00 – 16.30
The Weiss Conjecture and Katos Method for the NavierStokes Equations
Bernhard H. Haak, Peer Chr. Kunstmann
Abstract – We investigate Kato’s method for parabolic
equations with a quadratic non-linearity in an abstract form.
We extract several properties known from linear systems
theory which turn out to be the essential ingredients for the
method. We give necessary and sufficient conditions for these
conditions and provide new and more general proofs, based on
real interpolation. In application to the Navier-Stokes
equations, our approach unifies several results known in the
literature, partly with different proofs.
16.30 – 17.00
The Reciprocal Symmetry in State/Signal Systems in
Continuous Time
Olof J. Staffans
Abstract – The notion of reciprocity is well-known in
circuit theory: if a linear passive time-invariant circuit does
not contain any gyrators, then it is reciprocal in the
standard input/state/output sense, i.e., the impedance and
conductance transfer functions are congruent to their adjoints.
Here we extend this notion to the class of all (possibly infinitedimensional) state/signal systems in continuous time.
17.00 – 17.30
An H∞ Calculus of Admissible Operators
Hans Zwart
Treatable H² Optimization for InfiniteDimensional Systems
(Invited Session)
Room 8
Organizers: Yoshito Ohta, Leonid Mirkin
Chair: Yoshito Ohta
15.30 – 16.00
H² Performance on Multiple Preview Compensation and
Internal State Setting
Akira Kojima, Kazuki Yokoyama
Abstract – For a generalized H² preview control
problem which includes the command of partial state setting, a
design method of the preview control law is clarified. The
result stated here covers H² multiple preview control problems
and enables us to derive a compensation law for typical control
systems. The feature of the preview compensation law is
illustrated with a design example of preview servo-mechanism.
16.00 – 16.30
On the H² Two-Side Model Matching Problem with Preview
Maxim Kristalny, Leonid Mirkin
Abstract – The H² optimization problem with preview
and asymptotic behavior constraints is considered in a general
twoside model matching setting. The solution is obtained in
terms of two constrained Sylvester equations, associated with
asymptotic behavior, and stabilizing solutions of two algebraic
Riccati equations. The Riccati equations do not depend on the
preview length, yet are affected by asymptotic behavior
constraints and are thus different from the standard H² Riccati
equations arising in problems with no steady-state
requirements or in one-side problems.
16.30 – 17.00
Optimal Signal Reconstruction from a Series of Recurring
Delayed Measurements
Gjerrit Meinsma, Leonid Mirkin
Abstract – The modern sampled-data approach provides
a general methodology for signal reconstruction. This
paper discusses some implications for optimal signal
reconstruction when a series of recurring measurements, some
delayed, are available for the reconstruction.
17.00 – 17.30
A Unified Solution of a Class of Continuous/Discrete-Time
H² Control
Kotaro Hashikura, Yoshito Ohta
Abstract – Explicit solutions for the H² control problem for
systems with non-minimum phase property at the input such as
input time-delay systems attract the interest of
many researchers. The existing solution method cannot be seen
as a natural generalization of closed-loop reduction of the
finitedimensional standard H² problem. We give the method
of solving two-block problems twice based on analysis of a
statespace representation. The implication of the method is
that the optimal controller of Smith form is obtained directly
in both continuous- and discrete-time systems. The
truncation operator introduced in the literature is utilized in
solving the problem. We propose an alternative definition that
is consistent for both continuous- and discrete-time cases.
Furthermore, we show that the new definition is more
appropriate to characterize the optimal control cost.
List of Abstracts
Systems on Graphs
(Regular Session)
Room 9
Chair: Giacomo Como
15.30 – 16.00
Persistence of Disagreement in Social Networks
Daron Acemoglu, Giacomo Como, Fabio Fagnani, Asuman
Abstract – Disagreement among individuals in a society,
even on central questions that have been debated for centuries,
is the rule; agreement is the rare exception. How can
disagreement of this sort persist for so long? Existing models of
communication and learning, based on Bayesian or nonBayesian updating mechanisms, typically lead to consensus
provided that communication takes place over a strongly
connected network.
We analyze a stochastic model of communication
combined with the assumption that there are some “stubborn”
agents in the economy who never change their opinions. We
show that the presence of these stubborn agents leads to
persistent disagreements among the rest of the society—
because different individuals are within the “sphere of
influence” of distinct stubborn agents and are influenced to
varying degrees. Under general conditions, there is no
convergence to a consensus. Instead, the expected crosssectional distribution of beliefs in society converges (in
distribution), and generally, the opinion of a single individual,
and in fact that of the whole society, potentially fluctuates
forever. This model provides a new approach to understanding
persistent disagreements, and in the process, introduces new
tools for the analysis of opinion formation and consensus
16.00 – 16.30
Efficient Communication Infrastructures for Distributed
Control and Decision Making in Networked Stochastic
John S. Baras, Pedram Hovareshti
Abstract–In networked systems, groups of agents
achieve certain objectives via interaction at local levels in a
decentralized manner. The performance of such systems is
determined by the communication infrastructure of the
network as well as the system dynamics. The interdependence
of agents in a networked system is often modeled by graphs.
We study the interdependence of communication and
collaboration graphs in a networked system in the context of a
coordination control and decision making problems. We model
the decision on whether to cooperate or not in a group effort as
a result of a series of two-person games between agents and
their neighbors. The payoff of each agent is computed as the
sum of the agent’s payoffs from each of these games. Since
coordination games have more than one equilibrium point, the
problem is then which equilibrium point will the agents choose
and whether they will settle on a Pareto-optimal equilibrium
point. We consider a behavior learning algorithm and study its
effect on the emergence of a collaboration graph. We also study
the effect of the communication network topology on the
convergence speed of the scheme.
16.30 – 17.00
Spread of Epidemics in Time-Dependent Networks
Vahid S. Bokharaie, Oliver Mason, Fabian Wirth
Abstract – We consider SIS models for the spread of
epidemics. In particular we consider the so called
nonhomogeneous case, in which the probability of infection
and recovery are not uniform but depend on a neighborhood
graph which describes the possibility of infection between
individuals. In addition it is assumed, that infection, recovery
probabilities as well as the interconnection structure may
change with time. Using the concept of the joint spectral radius
of a family of matrices conditions are provided that guarantee
robust extinction of the epidemics.
17.00 – 17.30
On Robust Stability of the Belief Propagation Algorithm for
LDPC Decoding
Björn S. Rüffer, Peter M. Dower, Christopher M. Kellett,
Steven R. Weller
Abstract – The exact nonlinear loop gain of the belief
propagation algorithm (BPA) in its log-likelihood ratio
(LLR) formulation is computed. The nonlinear gains for
regular lowdensity parity-check (LDPC) error correcting codes
can be computed exactly using a simple formula. It is shown
that in some neighborhood of the origin this gain is actually
much smaller than the identity. Using a small-gain argument,
this implies that the BPA is in fact locally input-to-state
stable and produces bounded outputs for small-in-norm input
LLR vectors. In a larger domain the algorithm produces at
least bounded trajectories. Further it is shown that, as the
block length increases, these regions exponentially shrink.
17.30 – 18.00
The PABTEC Algorithm for Passivity-Preserving Model
Reduction of Circuit Equations
Tatjana Stykel, Timo Reis
Abstract – We present a passivity-preserving balanced
truncation model reduction method for circuit equations
(PABTEC). This method is based on balancing the solutions of
the projected Lur’e equations and admit computable error
bounds. We show how the topological structure of circuit
equations can be exploited to reduce the computational
complexity of the presented model reduction method.
Applications of Differential Geometry
(Regular Session)
Room 10
Chair: Markus Schöberl
15.30 – 16.00
System Parametrization Using Affine Derivative Systems
Markus Schöberl, Karl Rieger, Kurt Schlacher
Abstract – In this paper we discuss the constructive
calculation of a flat system parametrization for nonlinear
implicit control systems which are quasilinear in the derivative
coordinates. The proposed scheme is based on the successive
reduction of derivative variables and the elimination of nonderivative variables. A key challenge is the derivation of a
procedure that maintains the quasi-linearity also in the
elimination steps, since this is beneficial for the subsequent
reductions. Two examples demonstrate the applicability of the
suggested methods.
16.00 – 16.30
Impulsive Transfer of Elliptical Orbits for Consensus of Two
Spacecraft with Conservation of System Areal Velocity
Kazunori Furusawa, Tomohisa Hayakawa
Abstract – A consensus control framework for two
spacecraft traveling around the earth on elliptical orbits is
developed. The control inputs are assumed to be impulsive and
achieves a consensus state of the two spacecraft with control
instants at most twice under the condition that the total
angular momentum is conserved. We show an illustrative
numerical example to demonstrate efficacy of the proposed
Thursday, 8 July
16.30 – 17.00
Integrability and Optimal Control
Claudiu C. Remsing
Abstract – This paper considers left-invariant control
affine systems evolving on matrix Lie groups. Any leftinvariant optimal control problem (with quadratic cost) can be
lifted, via the celebrated Maximum Principle, to a Hamiltonian
system on the dual of the Lie algebra of the underlying state
space G. The (minus) Lie-Poisson structure on the dual space
g* is used to describe the (normal) extremal curves. Complete
integrability of (reduced) Hamiltonian dynamical systems is
discussed briefly. Some observations concerning Casimir
functions and the case of semisimple (matrix) Lie groups are
made. As an application, a drift-free left-invariant optimal
control problem on the rotation group SO (3) is investigated.
The reduced Hamilton equations associated with an extremal
curve are derived in a simple and elegant manner. Finally,
these equations are explicitly integrated by Jacobi elliptic
17.00 – 17.30
Accessibility and Observability for a Class of First-Order PDE
Systems with Boundary Control and Observation
Karl Rieger, Kurt Schlacher
temporal-spatial distribution of key biochemical cues (e.g.,
nutrients, growth factors). The distribution of nutrients
depends on a vascular network mimicking the structure of the
actual brain white matter. Computational bottlenecks
associated with the large number of cells and the solution of
PDEs in a 3-D domain are addressed by using an in situ
adaptive tabulation (ISAT) technique and a multi-grid method,
respectively. The model is used to simulate the early stages of
tumor development, before the onset of tumor-induced
16.30 – 17.00
Global Analysis of Firing Maps
Alexandre Mauroy, Julien M. Hendrickx, Alexandre
Megretski, Rodolphe Sepulchre
Abstract – In this paper, we study the behavior of
pulsecoupled integrate-and-fire oscillators. Each oscillator is
characterized by a state evolving between two threshold values.
As the state reaches the upper threshold, it is reset to the
lower threshold and emits a pulse which increments by a
constant value the state of every other oscillator.
The behavior of the system is described by the socalled firing map: depending on the stability of the firing
map, an important dichotomy characterizes the behavior of
the oscillators (synchronization or clustering). The firing map
is the composition of a linear map with a scalar nonlinearity.
Abstract – A group-theoretical approach is used to
tackle both the problem of local accessibility and observability
along a trajectory for a class of first-order PDE systems with
boundary control and observation. Based on an intrinsic
formulation including boundary terms (local) criteria are
derived in form of equivalence problems, where existence
and/or non-existence of (pointwise) transformation groups and
their invariants is related to (non-)observability and/or (non-)
accessibility of PDE systems, respectively. Examples
demonstrate the theory and results.
17.00 – 17.30
Structured Non-Linear Systems
(Regular Session)
Intraoperative Prediction of Tumor Cell Concentration from
Mass Spectrometry Imaging
Vandana Mohan, Ivan Kolesov, Ferenc A. Jolesz, Nathalie
Y.R. Agar, Allen R. Tannenbaum
Room 11
Chair: László Gerencsér
15.30 – 16.00
Change Detection for Hidden Markov Models
László Gerencsér, Cecilia Prosdocimi
Abstract – Hidden Markov Models (HMM-s) are widely
used in a number of application areas. In this paper we
consider the problem of detecting changes in the statistical
pattern of a hidden Markov process. We adapt the so-called
Hinkley-detector that was first introduced for independent
observations. Assuming that the dynamics before and after the
change is known, we are lead to the problem of analyzing the
Hinkley-detector with an L-mixing input. It is shown that,
under suitable technical conditions, the output process is also
L- mixing. The result yields a rigorous upper bound for the
falsealarm frequency. The limitations and potentials of the
result will also be discussed.
16.00 – 16.30
Three Dimensional Multi-Scale Modeling of Brain Tumor
Francisco Gamaliel Vital-Lopez, Costas D. Maranas,
Antonios Armaou
Abstract – We present a mathematical agent-based
model that describes tumor progression as the outcome of the
collective behavior of individual tumor cells, which is
determined by capturing the interplay between the
intracellular signaling pathways (e.g., MAPK pathway) and the
After briefly discussing the case of the scalar firing
map (corresponding to two oscillators), the stability analysis
is extended to the general n-dimensional firing map (for
n+1 oscillators). Different models are considered (leaky
oscillators, quadratic oscillators,...), with a particular emphasis
on the persistence of the dichotomy in higher dimensions.
Abstract – This work is motivated by the problem of
accurately locating tumor boundaries during brain tumor
surgery. Currently, such boundary is typically localized using
preoperative images and neuronavigation tools. While
improved prognosis is associated with minimal residual tumor,
an added challenge arises in surgical-decision making to
completely excise the tumor and preserve eloquent cortex. We
propose that an objective assessment of patterns of tumor cell
concentration will help in performing this boundary location
by identification of local minima of the tumor cell
concentration as tumor boundaries. In this work, we aim to
relate the mass spectrometry data - acquired from tissue
sections by the Desorption Electrospray Ionization (DESI)
approach - to histopathological scores of tumor cell
concentration (as evaluated by the neuropathology expert),
towards demonstrating that a system can be trained apriori on
available tissue samples with known scores, and can be used
intraoperatively as an integrated DESI probe to predict the
score of the tissue under analysis. We apply the
Relevance Vector Machine technique towards learning a
”model” that allows us to estimate the tumor cell concentration
given the mass spectra. We quantify the performance of this
model by testing the framework on real mass spectrometry
data acquired from brain tumors (gliomas) of different grades
and subtypes with promising results in prediction, and further
motivate its intraoperative application.
List of Abstracts
17.00 – 17.30
Stochastic Control – 2
(Invited Session)
Room 13
Organizers: Bozenna Pasik-Duncan, Tyrone E. Duncan
Chair: Tyrone E. Duncan
15.30 – 16.00
Convergence Rates of Markov Chain Approximation Methods
for Controlled Regime-Switching Diffusions with Stopping
Qingshuo Song, George Yin
Abstract–This work summarizes our recent work on rates of
convergence of Markov chain approximation methods
for controlled switching diffusions, in which both continuous
dynamics and discrete events coexist. The discrete events
are formulated by continuous-time Markov chains to
delineate random environment and other random factors that
cannot be represented by diffusion processes. The cost function
is over an infinite horizon with stopping times and without
discount. The paper demonstrates how to use a probabilistic
approach for studying rates of convergence. Although there
have been significant developments in the literature using PDE
(partial differential equation) methods to approximate
controlled diffusions, there appear to be yet any PDE results to
date for rates of convergence of numerical solutions for
controlled switching diffusions to the best of our knowledge.
Moreover, in the literature, to prove the convergence using
Markov chain approximation methods for control problems
involving cost functions with stopping (even for uncontrolled
diffusion without switching), an assumption was used to avoid
the socalled tangency problem. By modifying the value
function, we demonstrate that the anticipated tangency
problem will not arise in the sense of convergence in
probability and convergence in L1.
Calculus of Expected Present Value Operators and
Comparative Statics in Real Option Theory
Svetlana Boyarchenko, Sergei Levendorskii
Abstract – The paper provides a general framework for
study of impact of various policy interventions on investment
decisions of firms in a regime-switching environment. The
underlying uncertainty is modeled as a Markov-modulated
Brownian motion with embedded jumps or more general
process with i.i.d. increments. Not only characteristics and type
of a process depend on a state but the riskless rate and profit
flow as well. In each state, the profit flow is an arbitrary nondecreasing function of the underlying process, which allows for
profit flows with caps and floors, different level of taxation at
different levels of profits, etc. Different levels of the riskless
rate can be used to model stochastic interest rates. The
formulas for the value functions and investment thresholds in
different states allow for comparative statics. As an example,
we study how the investment thresholds, firm’s values and
output levels depend on the floor for the output price. We
design an efficient numerical procedure for calculation of the
investment thresholds and value functions and their derivatives
of order 1. The procedure does not involve numerical
integration, and the derivatives are calculated as accurately as
the thresholds and value functions themselves.We use
numerical examples to study the dependence of investment
thresholds on the floor, transition rates and interest rate
Finite Geometry and Network Codes
(Invited Session)
16.00 – 16.30
Existence of Strict Optimal Controls for Long-Term Average
Stochastic Control Problems
Francois Dufour, Richard H. Stockbridge
Abstract – Convexity conditions are identified under
which optimal controls in the class of strict controls exist for a
large class of stochastic processes under a long-term average
criterion in the presence of hard and/or soft constraints. The
result adapts a similar result obtained by Haussmann and
Lepeltier (1990) for a controlled diffusion under a mixed
optimalstopping/ finite-horizon/first-exit
approach taken in this paper is to utilize an equivalent linear
programming formulation of the control problem. These
results apply to controlled processes such as diffusions,
Markov chains, simple Markov jump processes, diffusions with
jumps, regimeswitching diffusions and solutions to Lévy
stochastic differential equations.
16.30 – 17.00
The Beneš-Problem and Related Problems Revisited
Kurt L. Helmes, Richard H. Stockbridge
Abstract – We show how the Beneš-Problem, i. e. the
problem of how to choose a nonanticapting control process u
whose absolute value is bounded by 1 such that the second
moment at time T of the controlled diffusion process X with
drift process u is as small as possible, can be solved by
analyzing a special entry-and-exit problem. A characterization
of the optimal strategy of general entry-and-exit problems can
be phrased in terms of a finite-dimensional nonlinear
optimization problem. This nonlinear optimization problem
can be solved explicitly for the case of switching controls of
Brownian motion with a quadratic cost function of the state.
The explicit solution is an essential ingredient of a new proof of
the Beneš-Problem as well as related problems.
Room 14
Organizers: Joachim Rosenthal, Marcus Greferath
Chair: Joachim Rosenthal
15.30 – 16.00
Two-Intersection Sets in Projective Hjelmslev Spaces
Thomas Honold
Abstract – A set S of points in a finite incidence structure
is said to be a two-intersection set if there are integers a < b
such that S meets every block in either a or b points (and both
a, b actually occur as intersection numbers). For pointhyperplane designs of the classical geometries PG(k, q) such
sets have been studied extensively and related to other
combinatorial objects (maximal arcs, two-weight codes,
strongly-regular graphs, partial difference sets). In this paper
two-intersection sets in the coordinate projective Hjelmslev
geometries PHG(k,R) over finite chain rings R of length 2 are
investigated along similar lines.
16.00 – 16.30
Construction of Codes for Network Coding
Andreas-Stephan Elsenhans, Axel Kohnert, Alfred
Abstract – Based on ideas of Kötter and Kschischang [6] we
use constant dimension subspaces as codewords in a network.
We show a connection to the theory of q-analogues of a
combinatorial designs, which has been studied in [1] as a
purely combinatorial object. For the construction of network
codes we successfully modified methods (construction with
prescribed automorphisms) originally developed for the qanalogues of a combinatorial designs. We then give a special
case of that method which allows the construction of network
codes with a very large ambient space and we also show how to
decode such codes with a very small number of operations.
Friday, 9 July
16.30 – 17.00
Friday, 9 July
Linear Codes from Projective Spaces
Michel Lavrauw, Leo Storme, Geertrui van de Voorde
Plenary Lecture
Room 1
Chair: Jan H. van Schuppen
09.00 – 10.00
From Qualitative to Quantitative Models of Gene Regulatory
Networks in Bacteria
Hidde de Jong
Distributed Parameter Systems III: Optimal
(Invited Session)
Room 1
17.00 – 17.30
On the Normality of (Non-Mixed and Mixed) Optimal
Covering Codes
Gerzson Kéri
Abstract – According to the experiences, it is a known fact
that some essential features of normal codes are quite different
for binary and for non-binary codes. After giving some
explanation concerning this observation by referring to an old
conjecture with its partial proof, its possible extensions and
restrictions, and by giving some counterexamples in Section 3,
subsequently an interesting inequality between the minimum
distance and covering radius of normal codes is expounded in
Section 4, which provides a lucid explanation for the observed
dissimilarity between the behavior of binary and non-binary
Geometric Control Theory for Linear Systems–2
Room 15
Organizers: Giovanni Marro, Fabio Morbidi, Lorenzo
Ntogramatzidis, Domenico Prattichizzo
15.30 – 17.30
Geometric Control Theory for Linear Systems
Giovanni Marro, Fabio Morbidi, Lorenzo Ntogramatzidis,
Domenico Prattichizzo
Organizers: Birgit Jacob, Michael A. Demetriou, Miroslav
Krstic, Kirsten Morris, Hans Zwart
Chair: Hans Zwart
10.30 – 11.00
Representation of Solutions of Riccati Equations for WellPosed Systems
Mark R. Opmeer, Orest V. Iftime
Abstract – We give a representation of self-adjoint
solutions of the control Riccati equation of a well-posed linear
system. At this level of generality the appropriate Riccati
equation is an integral Riccati equation. We assume that the
Riccati equation has a strongly stabilizing and a strongly antistabilizing solution, and that the difference of these two
solutions is coercive. We further assume that the uncontrolled
dynamics are given by a strongly continuous group. Our
representation is in terms of invariant subspaces of the
stabilizing closed-loop semigroup.
11.00 – 11.30
LQ-Optimal Control for a Class of Time-Varying Coupled
PDEs-ODEs System
Amir Alizadeh Moghadam, Ilyasse Aksikas, J. Fraser Forbes,
Stevan Dubljevic
Abstract – This contribution addresses the development of
a Linear Quadratic Regulator (LQR) for a set of timevarying hyperbolic PDEs coupled with a set of time-varying
ODEs through the boundary. The approach is based on an
infinitedimensional Hilbert state-space realization of the
system and operator Riccati equation (ORE). In order to solve
the optimal control problem, the ORE is converted to a set of
differential and algebraic matrix Riccati equations. The
feedback gain can then be found by solving the resulting
matrix Riccati equations. The control policy is applied to a
system of continuous stirred tank reactor (CSTR) and a plug
flow reactor (PFR) in series and the controller performance is
evaluated by numerical simulation.
11.30 – 12.00
Design of Optimal Deterministic Output Estimators for
Distributed Parameter Systems
Jochem Vissers, Siep Weiland
Abstract – This paper considers the optimal H2
estimation problem for infinite dimensional systems with finite
dimensional outputs. It is shown that this problem is equivalent
to a dual problem that allows an interpretation as a standard
Linear Quadratic optimization problem for an infinite
dimensional system. A solution to the latter problem is derived
which, in turn solves the optimal estimator problem.
List of Abstracts
12.00 – 12.30
12.00 – 12.30
Linear-Quadratic Differential Games Revisited
Michel C. Delfour
Weakly Operator Harmonizable Processes in Complete
Correlated Actions
Dan Popovici
Abstract – This paper revisits the pioneering work of
P. Bernhard [2] on two-person zero-sum linear quadratic
differential games and generalize it to utility functions
without positivity assumptions on the matrices acting on the
state variable and to linear dynamics with bounded measurable
data matrices. The paper specializes to state feedback via
Lebesgue measurable affine closed loop strategies with possible
non L²- integrable singularities. It first deals with L²-integrable
closed loop strategies and then with the larger family of
strategies that may have non L²-integrable singularities.
Realization and Information
(Regular Session)
Room 2
Abstract – We prove that a stochastic process in a
correlated action is weakly operator harmonizable (w.o.h.) if
and only if it has a stationary dilation. We identify, by analogy
with the stationary case, the shift operator of a w.o.h. process
which is, in our context, a linear contraction. Other conditions
which are equivalent to the notion of weak operator
harmonizability follow the solution of an operator moment
problem proposed by Z. Sebestyén and solved by Z. Sebestyén
and D. Popovici. The main result of the paper extends, to the
case of w.o.h. processes in complete correlated actions, the
classical decomposition of H. Cramér.
Analysis of Physical Systems
(Regular Session)
Room 3
Chair: Charalambos D. Charalambous
10.30 – 11.00
Chair: Hans Zwart
Rate Distortion Function for a Class of Relative Entropy
Farzad Rezaei, Charalambos D. Charalambous, Photios A.
10.30 – 11.00
Abstract – This paper deals with rate distortion or
source coding with fidelity criterion, in measure spaces, for a
class of source distributions. The class of source distributions
is described by a relative entropy constraint set between the
true and a nominal distribution. The rate distortion problem
for the class is thus formulated and solved using minimax
strategies, which result in robust source coding with fidelity
criterion. It is shown that minimax and maxmin strategies can
be computed explicitly, and they are generalizations of the
classical solution. Finally, for discrete memoryless uncertain
sources, the rate distortion theorem is stated for the class
omitting the derivations while the converse is derived.
11.00 – 11.30
Stochastic Realization of Binary Exchangeable Processes
Lorenzo Finesso, Cecilia Prosdocimi
Abstract – A discrete time stochastic process is called
exchangeable if its n-dimensional distributions are, for all
n, invariant under permutation. By de Finetti theorem any
exchangeable process is representable through a unique,
generally infinite, mixture of i.i.d. processes. We formulate, as
a stochastic realization problem, the question of characterizing
the binary exchangeable processes which are finite mixtures of
i.i.d. processes. The realizability conditions and an exact
realization algorithm are given in terms of the Hankel matrix
of the process. We establish a connection with the realization
problem of deterministic positive linear systems of the
relaxation type.
11.30 – 12.00
Stochastic Input-Output Realization of Bilinear Systems
György Terdik, József Bokor
Abstract – A bilinear stochastic system given in state space
form is studied when both the input and the output are
measured. The Hankel matrix of the system is built up in terms
of the Fliess-series representation of the process. The Fliesscoe¢ cients are calculated by the cumulants between the output
and the polynomials of the input.
Analysis of the Three Dimensional Heat Conduction in Nanoor Microscale
Hanif Heidari, Hans Zwart, Alaeddin Malek
Abstract–The Dual-Phase-Lagging (DPL) equation is
formulated as an abstract differential equation. In the
absence of a heat source term the DPL equation with
homogeneous boundary conditions generates a contraction
semigroup. The exact expression of the semigroup is achieved.
It is proved that the associated eigenfunctions form a Riesz
basis. The stability of semigroup is proved. Moreover, it is also
shown that the spectrum of DPL equation contains an interval.
This implies that the infinitesimal generator associated to the
DPL equation is not a Riesz spectral operator. Therefore, the
known test for approximate controllability cannot be used.
Several controllability properties are investigated.
11.00 – 11.30
Well-Posedness, Regularity and Exact Controllability for the
Problem of Transmission of the Schrödinger Equation
Salah-Eddine Rebiai
11.30 – 12.00
Hamiltonian Evolution Equations of Inductionless
Andreas Siuka, Markus Schöberl, Kurt Schlacher
Abstract – The objective of this contribution is to find a
coordinate independent Hamiltonian representation of
the governing
Magnetohydrodynamics, where we are interested in analysing
the relevant energy flows in a purely geometric fashion also
taking dissipative effects into account. We especially treat the
boundary conditions in an extraordinary manner and we
define control inputs which may act on the system boundary.
Port-Controlled Hamiltonian
representation, well-known in the lumped parameter case, is
also reflected in the infinite dimensional case which is crucial
particularly with regard to control theoretic aspects.
Friday, 9 July
lossless mechanical systems with given transfer functions using
springs and masses.
12.00 – 12.30
Instantaneous Control of the Linear Wave Equation
Nils Altmüller, Lars Grüne, Karl Worthmann
11.00 – 11.30
Abstract – We are concerned with the one dimensional
linear wave equation with Dirichlet boundary condition and
Neumann boundary control. It has been shown numerically
that this hyperbolic partial differential equation can be
stabilized by instantaneous control, i.e. model predictive
control with the shortest feasible prediction and optimization
horizon. Our contribution is the complete theoretical analysis.
12.30 – 13.00
Hybrid Modeling, Control and Estimation in ABS Applications
Based on In-Wheel Electric Motors
Ricardo de Castro, João de Sousa, Rui Esteves Araújo,
Fernando Pereira, Diamantino Freitas
Abstract – The problems of tire slip control and peak
friction estimation for Anti-lock Braking Systems are
formulated and solved in the framework of hybrid control
techniques. The hybrid systems model of the vehicle dynamics
arises from approximating the nonlinear behavior of the tireroad friction coefficient with a Piecewise Linear Function
(PLF). The problem of tyre slip control is formulated as an
invariance problem. The invariance control problem is solved
with the help of a simple hysteretic controller which takes
advantage of the fast torque response from the in-wheel electric
motor. The controller’s properties, such as necessary
conditions to ensure invariance, robustness to disturbances,
convergence and limit cycle period are analyzed in detail. The
problem of the estimating the peak friction for the tire-road
interface is solved using properties of the limit cycles arising in
the hybrid model. Monitoring the duty cycle, imposed by
hysteretic controller, was found to be sufficient to extract
information regarding the conditions of adhesion in the road.
The controller and estimator were evaluated under variable
grip levels, in a vehicle dynamics simulation software,
obtaining a good performance for both the tire slip regulation
and for the peak friction estimationd.
Behavioral Systems and Control Theory
(Invited Session)
Room 4
Organizer: Paolo Rapisarda
Chair: Paolo Rapisarda
10.30 – 11.00
A Polynomial Approach to the Realization of J-Lossless
Shodhan Rao, Paolo Rapisarda, Lewis Moody
Abstract–In this paper, a class of behaviours known as Jlossless behaviours is introduced, where J is a symmetric
twovariable polynomial matrix. For a certain J, it is shown
that the resulting set of J-lossless behaviours are SISO
behaviours such that for each of such behaviours, there exists a
quadratic differential form which is positive for nonzero
trajectories of the behaviour and whose derivative is equal to
the product of the input variable and the derivative of the
output variable. Earlier, Van der Schaft and Oeloff had
considered a specific form of realization for such behaviours
that plays an important role in their model reduction
procedure. In our paper, we give a method of computation of a
state space realization from a transfer function of such a
behaviour in the same form as considered by Van der Schaft
and Oeloff, using polynomial algebraic methods. Apart from
being useful in enlarging the scope of the model reduction
procedure of Van der Schaft and Oeloff, we show that our
method of realization also has application in the synthesis of
Pseudorational Behaviors and Bezoutians
Yutaka Yamamoto, Jan C. Willems, Masaki Ogura
Abstract – Behavioral system theory has been successful in
providing a viewpoint that does not depend on a priori notions
of inputs/outputs. While there are some attempts to extend this
theory to infinite-dimensional systems, for example, delay
systems, the overall picture seems to remain still incomplete.
The first author has studied a class of infinite-dimensional
systems called pseudorational. This class allows a compact
fractional representation for systems having bounded-time
memory. It is particularly appropriate for extending the
behavioral framework to infinite-dimensional context.
We have recently studied several attempts to extend this
framework to a behavioral context. Among them are
characterizations of behavioral controllability, particularly
involving a coprimeness condition over an algebra of
distributions, and some stability tests involving Lyapunov
functions derived from Bézoutians.
This article gives a brief overview of pseudorational transfer
functions, controllability issues and related criteria, path
integrals, and finally the connection with Lyapunov functions
derived from Bézoutians.
11.30 – 12.00
Tracking and Regulation in the Behavioral Framework
Shaik Fiaz, Kiyotsugu Takaba, Harry Trentelman
Abstract – This paper considers the problem of tracking and
regulation for the class of linear differential systems in the
behavioral framework. Given a plant, together with an
reference signals, the problem of tracking and regulation is to
find a controller such that the plant variable tracks the
reference signal regardless of the disturbance acting on the
system. A controller which achieves this design objective is
called a regulator for the plant with respect to the exosystem.
In this paper we formulate the tracking and regulation
problem in the behavioral framework, with control as
interconnection. We obtain necessary and sufficient conditions
for the existence of a controller which acts like a regulator for
the plant with respect to the exosystem. The problem
formulation and its resolution are completely representation
free, and specified only in terms of the plant and the exosystem
12.00 – 12.30
Ports and Terminals
Jan C. Willems
Abstract – We examine what is meant by the power and the
energy which a physical system exchanges with its
environment. The systems which we consider interact through
terminals, as wires for electrical circuits, and pins for
mechanical devices. Associated with each terminal, there are
variables through which the system interacts with its
environment. For circuits, these variables are current and
potential, and for (one-dimensional) mechanical systems,
position and force. Systems are interconnected by sharing
variables at the interconnected terminals.
We define a port as a set of terminals that satisfy certain
conditions, which we call the port-Kirchhoff laws. For ports,
and only for ports, we define the power and the energy which
flows into a system. Since a port involves more than one
terminal, power and energy are not `local', but involve `action
List of Abstracts
at a distance'. Moreover, we cannot speak about the power and
the energy flow along any set of terminals.
We discuss the nature of ports for electrical and mechanical
systems, and derive an expression for energy that is not frame
dependent. We prove that a connected RLC circuit forms a 1port. This implies that in open systems energy rarely flows
between subsystems along the interconnected interface.
For mechanical systems the definition of a port implies that
springs, dampers, and inerters form port, but a mass does not.
This leads to the problem how one should define the motion
energy of a group of moving masses. We derive an expression
for the motion energy, which is different from the classical
expression of kinetic energy. Our formula implies that motion
energy is not an extensive quantity.
data functions and admissible controllers leading to a
description of the solutions in terms of solutions of
coupled linear
connection between the interpolation and the state-space
approach relies on the seamless equivalence of frequencydomain and state-space representation. In the cases of
multivariable interpolation and systems with structured
uncertainty similar reductions and solution criteria exist (after
some compromises with respect to the solution criterions). The
connection between the results however is not clear due to the
failure of the state-space similarity theorem and Kalman
decomposition in these settings, and as a result research on the
two topics has diverted. In this talk, which is based on the
paper [1], we discuss these developments and make some
explicit connections.
12.00 – 12.30
Shadows of Multidimensionality:
Multidimensional Systems with Applications to
1-D Systems – 1
(Invited Session)
Room 6
Organizers: Joseph A. Ball, Victor Vinnikov
Chair: Victor Vinnikov
10.30 – 11.00
Structured Noncommutative Multidimensional Linear
Systems and Scale-Recursive Modeling
Tanit Malakorn, Joseph A. Ball
Abstract – Recently, the multiscale signal and image
processing community has recognized that a suitable model for
multiresolution processes is a model with time-like variable
indexed by the nodes on a homogeneous tree with different
depths in the tree corresponding to different spatial scales
associated with the signal or image. It turns out that these
system models are close relatives of the Structured
Noncommutative Multidimensional Linear Systems (SNMLSs)
introduced by Ball-Groenewald- Malakorn [5], but with system
operators dependent on the node of the tree at which the state
update occurs. This provides engineering motivation for the
introduction of a “parametervarying” version of SNMLS.
11.00 – 11.30
Controllability of Autonomous Behaviors and Livšic
Overdetemined Systems as 2D Behaviors with Pure
Autonomy Degree One
Grant Boquet, Joseph A. Ball
Abstract – In [8] discrete time Livšic systems are related to
necessary component for transitioning from a behavior to a
Livšic system is a “controllability theory” for autonomous
behaviors. It turns out that controllability for behaviors is a
special case of the presented j-controllability. First we present a
brisk overview of the ingredients that go into (algebraic) Drcontrollability and (trajectory) j-controllability. We conclude
with results demonstrating Livšic controllability implies 1controllability and then Dr-controllability by a series of
reduction steps.
11.30 – 12.00
The H∞-Problem in Multidimensional Control Theory: StateSpace versus Frequency-Domain Formulation
Joseph A. Ball, Sanne ter Horst
Abstract – Two by now standard approaches to the
classical H∞-problem
factorizations reducing it to a metric constrained interpolation
problem, or via state-space realizations of the given
Linear State Space Theory in the White Noise Space Setting
Daniel Alpay, David Levanony, Ariel Pinhas
Abstract – We study state space equations within the white
noise space setting. A commutative ring of power series in a
countable number of variables plays an important role.
Transfer functions are rational functions with coefficients in
this commutative ring, and are characterized in a number of
ways. A major feature in our approach is the observation that
key characteristics of a linear, time invariant, stochastic system
are determined by the corresponding characteristics associated
with the deterministic part of the system, namely its average
Systems on Graphs – Consensus
(Regular Session)
Room 7
Chair: Enrico Lovisari
10.30 – 11.00
A Resistance-Based Approach to Performance Analysis of
the Consensus Algorithm
Enrico Lovisari, Federica Garin, Sandro Zampieri
Abstract – We
consensus algorithm by means of a LQ-type performance cost.
We want
topology influences this algorithm. In order to do this, we
recall the analogy between Markov Chains and electrical
resistive networks. By exploiting this analogy, we are able to
rewrite the performance cost as the average effective resistance
on a suitable network. We use this result to show that if
the communication graph fulfills some local properties, then its
behavior can be approximated with that of a suitable grid, over
which the behavior of the cost is known.
11.00 – 11.30
Notes on the Deficiency One Theorem: Single Linkage Class
Balázs Boros
Abstract – The Deficiency One Theorem tells us about
certain chemical reaction systems that they cannot admit
multiple interior equilibria. The theorem was proven by
Feinberg. In this paper we provide a relatively short proof of
that theorem for the special case of one linkage class. We also
extend that result by giving an equivalent condition to the fact
that the set of interior equilibria is nonempty for a chemical
reaction system with one linkage class considered in the
Deficiency One Theorem.
Friday, 9 July
11.30 – 12.00
Optimal Finite-Time Distributed Linear Averaging
Qing Hui
Abstract – A new optimal finite-time distributed linear
averaging (OFTDLA) problem is presented in this paper.
This problem is motivated from the distributed averaging
problem which arises in the context of distributed algorithms
in computer science and coordination of groups of
autonomous agents in engineering. The aim of the OFTDLA
problem is to compute the average of the initial values in finitetime steps at nodes of a graph through an optimal distributed
algorithm in which the nodes in the graph can only
communicate with their neighbors. Optimality is given by a
minimization problem of a quadratic cost functional under
finite-time horizon. We show that this problem has a very close
relationship with the notion of semistability. By developing new
necessary and sufficient conditions for semistability of discretetime linear systems, we convert the original OFTDLA problem
into two equivalent optimization problems. One of them is a
convex optimization problem and can be solved by using
semidefinite programming methods.
12.00 – 12.30
truncation is one of the most useful model order reduction
methods. In general, however, the stability of the feedback
system is not maintained when the order of the controller is
reduced by balanced truncation. This paper proposes a novel
method of state feedback controller reduction by which we can
preserve the stability of the resulting reduced order state
feedback system. A numerical example demonstrates the
effectiveness of the proposed method.
11.30 – 12.00
Numerical Algorithm for Structured Low Rank
Approximation Problem
Swanand R. Khare, Harish K. Pillai, Madhu N. Belur
Abstract – In this paper we discuss an important problem of
Structured Low
Rank Approximation (SLRA) of
linearly structured matrices. This is a very important problem
having many applications like computation of approximate
GCD, model order reduction to name a few. In this paper
we formulate SLRA problem as an unconstrained
optimization problem on a smooth matrix manifold. We use
Armijo line search algorithm on the matrix manifold to
compute the nearest SLRA of the given matrix.
12.00 – 12.30
Dilatability of Linear Cellular Automata
Adriana Popovici, Dan Popovici
Abstract – We introduce a notion of dilatability between
two LCAs and relate it with the notion of (power)
dilatability between the corresponding global transition
functions. We prove that a partial isometric LCA can be
dilated to a quantum LCA which is reversible. In particular,
any isometric LCA A can be dilated to a quantum LCA B such
that the global rule of B extends the global rule of A.
Finite Time System Operator and Balancing for Model
Reduction and Decoupling
Erik I. Verriest
Abstract – In this paper we explore the operator mapping
a finite time segment of the input signal to the output over
the same interval. The properties of this operator are
compared to the finite time Hankel operator that was useful in
sliding interval balancing (SIB). Potential applications for
model reduction and decoupling of systems are discussed.
12.30 – 13.00
Model Reduction
(Regular Session)
Model Order Reduction of Nonlinear Circuits
Andreas Steinbrecher
Room 8
Chair: André Schneider
10.30 – 11.00
Balanced Truncation Model Order Reduction for LTI Systems
with many Inputs or Outputs
Peter Benner, André Schneider
Abstract – We discuss balanced truncation (BT) based
methods for model order reduction (MOR) of linear time
invariant (LTI) systems with many input or many output
terminals. Applying BT methods makes it necessary to balance
the system, which is equivalent to finding the controllability
and observability Gramian of the system in a special
diagonal form. The Cholesky factors of these Gramians are
efficiently computable as solutions of dual Lyapunov equations
for systems with only few inputs and outputs. After a brief
introduction and a short recollection of basic knowledge of BT,
we show a method to get the Gramians’ factors also for systems
with many inputs and outputs with the help of the GaussKronrod quadrature formula. We show some numerical results
using this quadrature rule and explain how to get the BT
reduced order model out of these results.
11.00 – 11.30
Balanced Truncation for Linear Interconnected Systems: the
State Feedback Case
Kenji Fujimoto, Sayaka Ono, Yoshikazu Hayakawa
Abstract – Model order reduction is an important tool
in control systems theory. In particular, it is useful for
controller design since the dimension of the controller becomes
very high when we use advanced control theory. Balanced
Abstract – In this talk we develop a model order
reduction for the model equations of nonlinear circuits with a
small number of nonlinear elements. The presented model
reduction technique is based on the decoupling of the linear
and nonlinear subcircuits of the electrical circuit in a suitable
way. Afterwards, a model reduction of the remained linear
part will be performed using passivity-preserving balanced
truncation followed by an adequate recoupling of the
unchanged nonlinear subcircuit and the reduced linear
subcircuit to obtain a nonlinear reducedorder model. The
efficency and applicability of the proposed model reduction
approach is demonstrated on a numerical examples.
Networked Control – 1
(Regular Session)
Room 9
Chair: Karl Mårtensson
10.30 – 11.00
Sub-Optimality Bound on a Gradient Method for Iterative
Distributed Control Synthesis
Karl Mårtensson, Anders Rantzer
Abstract – A previous paper introduced an online
gradient method to iteratively update local controllers for
improved performance. In this paper we modify that method to
get an offline method for distributed control synthesis. The
complexity of the method is linear in the number of neighbors
to each agent.
List of Abstracts
Since the controllers are constructed to be distributed
and the method is an iterative scheme, the controllers will
always be sub-optimal compared to a centralized controller.
We describe a method to calculate bounds of the sub-optimality
of the controllers, using the same variables that take part in the
update scheme.
Analytical Methods
(Regular Session)
Room 10
Chair: Tamás Kalmár-Nagy
10.30 – 11.00
11.00 – 11.30
The Logarithmic Quantiser is Not Optimal for LQ Control
Jean-Charles Delvenne
Abstract – We seek to stabilise a scalar linear system
through a finite-capacity communication channel, while
minimising a quadratic cost. We show that the logarithmic
quantiser strategy is not optimal for the quadratic cost in the
limit of low capacities.
Random Walk on a Rooted, Directed Husimi Cactus
Tamás Kalmár-Nagy
Abstract – The objective of this paper is to further
explore the connection between the random Fibonacci series
and a random walk on the so-called triangular Husimi
cactus. Various statistical properties of this random walk are
11.00 – 11.30
11.30 – 12.00
Distributed Inequality Constrained Kalman Smoother
Simone Del Favero, Gianluigi Pillonetto, Bradley M. Bell
Abstract – This paper, which is the sequel of [1],
considers smoothing of Gauss-Markov linear systems via
distributed optimization. As an application we consider the
distributed estimation problem over sensor networks and
assume that each node has access to noisy measurements of
different but correlated states. Then, the aim is to reconstruct
the overall state sequence in a cooperative way, by taking
advantage of all the data collected by the network.
In this paper, the convergence analysis in [1] is
deepened, pointing out the importance in the algorithm design,
of finding the right trade off between parallelism and
convergence rate. Moreover an extension of the algorithm to
the case of state sequence subject to inequality constraints is
also provided. In particular, we show that the same algorithmic
architecture and communication protocol used in the
unconstrained case can be exploited in the constrained
scenario. Hence, the network can efficiently include in the
estimation process relevant a priori information on the state,
such as nonnegativity. Numerical experiments regarding the
distributed reconstruction of a function via spline regression is
used to test the new approach.
12.00 – 12.30
MIMO Encoder and Decoder Design for Signal Estimation
Erik Johannesson, Andrey Ghulchak, Anders Rantzer, Bo
Abstract – We study the joint design of optimal linear
MIMO encoders and decoders for filtering and transmission of
a vectorvalued signal over parallel Gaussian channels subject
to a realtime constraint. The objective is to minimize the sum
of the estimation error variances at the receiving end. The
design problem is nonconvex, but it is shown that a global
optimum can be found by solving a related two-stage problem.
The first stage consists of a mixed norm minimization
problem, where the 2-norm corresponds to the error variance
in a corresponding Wiener-Kolmogorov filtering problem and
the 1-norm is induced by the channel noise. The second
stage consists of a matrix spectral factorization.
Scalable Decentralized Control and the Davis-Wielandt Shell
Ioannis Lestas
Abstract – We consider a large scale network comprised
of heterogeneous dynamical agents. We derive scalable
stability certificates that involve the input/output properties of
individual subsystems and corresponding properties of the
interconnection matrix. The stability conditions presented are
based on the Davis-Wielandt shell, a higher dimensional
version of the numerical range, which allows to relax normality
or symmetry assumptions on the interconnection matrix. The
conditions derived include small gain and passivity approaches
as special cases, and generalize many results within the areas of
consensus protocols and Internet congestion control.
11.30 – 12.00
Characterization of Shift Invariant Subspace of Matrixvalued Hardy Space
Yohei Kuroiwa
Abstract – The characterization of the shift invariant
subspace of the matrix-valued Hardy space is given. It is
a matrix-valued generalization of the Beurling-Lax theorem.
The Beurling-Lax
representation of a shift invariant subspace by a unique inner
function. Our characterization of the shift invariant subspace
of the matrixvalued Hardy space is given by a two-sided
representation of a shift invariant space by inner and co-inner
Quantum Systems
(Regular Session)
Room 11
Chair: Francesca Albertini
10.30 – 11.00
Methods of Control Theory for the Analysis of Quantum
Walks on Graphs
Francesca Albertini, Domenico D’Alessandro
Abstract – The goal of this paper is to summarize
recent results on the analysis of quantum walks on graphs.
These systems are used in quantum information theory as
protocols to design quantum algorithms. By taking into
account that some model variables can be changed with time,
quantum walks can be looked at as control systems and several
questions can be posed in control theoretic terms. In particular
the set of states that can be reached for these systems can be
characterized via controllability analysis. After setting up the
model, the main results of the paper characterize the
controllability of quantum walks both in algebraic and in
combinatorial terms. Several examples are also discussed.
Friday, 9 July
11.00 – 11.30
Parameter Estimation of Quantum Processes Using Convex
Gábor Balló, Attila Magyar, Katalin M. Hangos
Abstract – A convex optimization based method is
proposed for quantum process tomography, in the case of
known channel model structure, but unknown channel
parameters. The main idea is to select an affine
parametrization of the Choi matrix as a set of optimization
variables, and formulate a semidefinite programming problem
with a least squares objective function. Possible convex
relations between the optimization variables are also taken into
account to improve the estimation.
Simulation case studies show, that the proposed method
can significantly increase the accuracy of the parameter
estimation, if the channel model structure is known. Beside the
convex part, the determination of the channel parameters from
the optimization variables is a nonconvex step in general. In
the case of Pauli channels however, the method reduces to a
purely convex optimization problem, allowing to obtain a
globally optimal solution.
11.30 – 12.00
Martingale Approach in Quantum State Estimation Using
Indirect Measurements
László Ruppert, Katalin M. Hangos
Abstract – The aim of this work is to propose
mathematically well grounded statistical methods for state
estimation in the indirect measurement settings by using
martingale theory, and to compare their efficiency to the usual
direct approaches. The measurement scheme considered is the
simplest possible discrete time case, where both the unknown
and the measurement quantum systems are quantum bits. The
repeated measurements performed on the measurement
subsystem of the composite system enables us to construct an
estimator for the initial state of the unknown system. An initial
state relative method of detecting a stopping time is proposed
where the final states are defined using a given distance from
the unknown initial state. A simple estimator is proposed and
used as an excellent initial point to build more complex and
better estimation methods. The efficiency of the proposed
procedure is investigated both analytically and experimentally
using simulation in different settings of parameters. The
possible generalizations of the proposed estimation methods
are also outlined.
12.00 – 12.30
The Energy Minimization Problem for Two-Level Dissipative
Quantum Systems
Dominique Sugny, Bernard Bonnard
Abstract – The objective of this article is to present
developments of geometric optimal control to analyze the
energy minimization problem of dissipative two-level quantum
systems whose dynamics is governed by Kossakowski-Lindblad
equations. This analysis completed by numerical simulations
based on adapted algorithms allows a computation of the
optimal control law whose robustness with respect to initial
conditions and dissipative parameters is also detailed.
12.30 – 13.00
On a Canonical QR Decomposition and Feedback Control of
Discrete-Time Quantum Dynamics
Francesco Ticozzi, Saverio Bolognani
Abstract – We study feedback-controlled, discrete-time
quantum Markovian dynamics focusing on pure-state
stabilization problem. Assuming that the system is unitarily
controllable, and accessible via a given quantum measurement,
we explicitly construct a choice of control actions conditioned
on the measurement outcome that globally stabilizes the target
state for the averaged dynamics. A key step in deriving this
result is the definition of a canonical QR decomposition for
complex matrices.
Applications in Medicine
(Regular Session)
Room 12
Chair: Andras Balogh
10.30 – 11.00
Bifurcation Control in an Infectious Disease Model
Andras Balogh, Roberto Castillo, Noel Cavazos Jr.
Abstract – In
mathematical model describing the spread of a class of
infectious diseases. A system of four integral equations
represents the SEIRS model, where individuals go through
stages of being susceptible (S), exposed (E), infective (I), and
recovered (R) for constant periods of time intervals.
Transcritical bifurcation of steady state solutions can be
observed in the system as the basic reproduction
number increases. The eigenvalue analysis of the linearized
equations provides local stability results. A stable numerical
algorithm is developed that demonstrate the theoretical
11.00 – 11.30
Identifiability Analysis of an Epidemiological PDE Model
Antoine Perasso, Béatrice Laroche, Suzanne Touzeau
Abstract – We investigate the parameter identifiability
problem for a SIR system of nonlinear integro-partial
differential equations of transport type, representing the
spread of a disease with a long infectious but undetectable
period in an animal population. After obtaining the expression
of the model inputoutput (IO) relationships, we give sufficient
conditions on the boundary conditions of the system that
guarantee the parameter identifiability on a finite time horizon.
We finally illustrate our findings with numerical simulations.
11.30 – 12.00
Unification of Accelerated and Proportional Hazard Rate
Models and Application for Data of the Hungarian National
Cancer Registry
Lídia Rejtı
Abstract – In the literature of multifactorial survival
analysis, individual survival curves are described eventually
with a single parameter deeming all survival curves to be
parallel ([1], [3], [6], [9]). Previously described theories are not
able to produce all form of survival curves we may meet;
in contrast, we propose a novel method which can handle
cases, for example, with constant hazard rate and rapidly
decreasing ones simultaneously. Using our methods we
estimated survival chances of 189,026 tumor cases, recognized
between 2001 and 2005 in Hungary and recorded in NCR.
List of Abstracts
Control for Markov and Nonlinear Markov
(Invited Session)
Room 13
Organizers: William M. McEneaney, Vassili N. Kolokoltsov
Chair: Vassili N. Kolokoltsov
10.30 – 11.00
Nonlinear Markov Games
Vassili N. Kolokoltsov
Abstract – A program of the analysis of a new class of
stochastic games is put forward, which I call nonlinear
Markov games, as they arise as a (competitive) controlled
version of nonlinear Markov processes (an emerging field of
intensive research, see e.g. [2], [5], [6]). This class of games can
model a variety of situation for economics and epidemics,
statistical physics, and pursuit - evasion processes. The
discussion below will be presented i more detail in the author’s
monograph [1].
11.00 – 11.30
Stochastic Adaptive Nash Certainty Equivalence Control:
Self-Identification Case
Arman C. Kizilkale, Peter E. Caines
Abstract – For
Certainty Equivalence (NCE), or Mean Field (MF)
developed in
decentralized strategies which asymptotically yield Nash
equilibria. The NCE (MF) control laws use only the local
information of each agent on its own state evolution and
knowledge of its own dynamical parameters, while the
behaviour of the mass is precomputable from knowledge of the
distribution of dynamical parameters throughout the mass
Relaxing the a priori information condition introduces
the methods of parameter estimation and stochastic adaptive
control (SAC) into MF control theory. In particular one
may consider incrementally the problems where the agents
must estimate: (i) its own dynamical parameters, (ii) the
distribution of the population’s dynamical parameters [1], and
(iii) the distribution of the population’s cost function
parameters [2]. In this paper we treat the first problem.
Each agent estimates its own dynamical parameters via
the recursive weighted least squares (RWLS) algorithm.
Under reasonable conditions on the population dynamical
parameter distribution, we establish: (i) the strong consistency
of the selfparameter estimates; and that (ii) all agent systems
are long run average L² stable; (iii) the set of controls yields a
(strong) -Nash equilibrium for all ; and (iv) in the population
limit the long run average cost obtained is equal to the nonadaptive long run average cost.
11.30 – 12.00
Stopping Problems of Markov Processes with Discontinuous
Lukasz Stettner
Abstract – The paper summarizes recent results on
optimal stopping of Feller Markov processes with time or space
discontinuous functionals. We characterize value functions and
their potential discontinuity points for various cost functionals:
finite time horizon, first exit from an open set horizon and
infinite horizon. Formulae for optimal or optimal stopping
times are also given.
12.00 – 12.30
Observation Process Control in Support of Stochastic
Tasking Operations
William M. McEneaney, Ali Oran
Abstract – We consider a problem of observation
control, specifically a problem where one chooses which aspects
of the state to observe at each time-step. The state takes
values in a finite set, and the conditional probability updates
by Bayes’ rule. The payoff for observation takes the form of
a finite maximum of linear functions of the final
observationconditioned probability distribution, and so is a
convex function of the distribution. However, the goal is
maximization, not minimization. Through use of the max-plus
distributive property, we are able to use a max-plus curse-ofdimensionalityfree computational method for solution of the
control problem. Complexity attenuation of the algorithm is
Algebraic Systems Theory, Behaviors, and
Codes: Design, Analysis, and Decoding of
Convolutional Codes
(Invited Session)
Room 14
Organizers: Heide Glüsing-Lürssen, Eva Zerz
Chair: Heide Glüsing-Lürssen
10.30 – 11.00
Decoding of a Class of Convolutional Codes
Heide Glüsing-Lürssen, Uwe Helmke, José Ignacio Iglesias
Abstract – A
convolutional codes will be exposed. Under certain conditions
this algorithm will allow to correct a high number of errors in
an interval of fixed length. We will show that a class of Cyclic
Convolutional Codes is particularly well suited for this
iterative algorithm in two aspects: first, it satisfies the
conditions so that the error correcting capacities of the
algorithm are optimized; secondly, the computations needed at
each iteration are feasible. Consequently, the application of the
algorithm to this class of codes results in an efficient decoding
11.00 – 11.30
On the Determination of an Input-State-Output Realization
of a Secure McEliece-Like Cryptosystem Based on
Convolutional Codes
Joan Josep Climent, Victoria Herranz, Carmen Perea,
Virtudes Tomás
Abstract – In this paper we present a public key
cryptosystem based on the McEliece scheme, but using a
convolutional code, instead of a block code. Firstly we present
some conditions about the convolutional code C to construct
the public key cryptosystem and then, starting with the parity
check matrix H of a good block code, we find an input-stateoutput representation of C such that the controllability matrix
of C is Ht. This cryptosystem is constructed so that any user
can encrypt a message by introducing the largest number of
possible errors.
11.30 – 12.00
Reverse-Maximum Distance Profile Convolutional Codes
over the Erasure Channel
Virtudes Tomás, Joachim Rosenthal, Roxana Smarandache
Abstract – The loss of transmitted packets over an
erasure channel, such as the Internet, can generate delay of the
received information due to retransmission, and this can have
Friday, 9 July
adverse effects in real-time applications. Error forward
correction is a technique used to avoid this delay. Until now
mainly block codes have been used for this purpose and
convolutional codes have been much less studied. In this paper
we study in detail the use of convolutional codes over this
channel and we show that the complexity of decoding is
polynomial. We see how maximum distance profile (MDP)
convolutional codes can deal with situations which are not
possible for a maximum distance separable (MDS) block code
and we introduce a new concept: reverse-MDP convolutional
codes. Reverse-MDP codes double the potential of MDP
convolutional codes since they behave as MDP codes in a
forward and a backward sense. Due to this fact, we propose
this new kind of codes as very good candidates to improve the
decoding process. In addition, we provide a particular
construction for reverse-MDP convolutional codes.
12.00 – 12.30
Column Distances for 2D-Convolutional Codes
Diego Napp Avelli, Carmen Perea, Raquel Pinto
Abstract – In this work we introduce the concept of
column distance for delay-free two dimensional (2D) finite
support convolutional codes. We present its principal
properties and an upper and lower bound for the column
called oblique manifold, intrinsic Newton’s method is
developed to optimize two popular cost functions for the
simultaneous diagonalization of symmetric matrices: the offnorm function and the log-likelihood function. Performance of
the proposed algorithms is investigated and compared by
several numerical experiments.
11.30 – 12.00
A Geometric Revisit to the Trace Quotient Problem
Hao Shen, Klaus Diepold, Knut Hüper
Abstract – This paper studies the problem of trace
quotient, or trace ratio maximization, which has enormous
applications in computer vision, pattern recognition and
machine learning. We provide a geometric revisit to the
problem in the framework of optimization on smooth
manifolds. The set of critical points of the trace quotient is
analyzed. Local quadratic convergence properties of the socalled Iterative Trace Ratio (ITR) scheme, which recently
became an attractive solver to the problem, is studied. Based
on this result, different from a popular realization of ITR,
which requires to solve a symmetric eigenvalue problem at
each iteration, we propose a simple, efficient algorithm, which
employs only one step of the parallel Rayleigh quotient
iteration at each iteration. An numerical experiment
demonstrates the local convergence properties of ITR.
12.00 – 12.30
Differential Geometric Methods for
Computational Engineering Applications – 1
(Invited Session)
Room 15
Organizers: Knut Hüper, Christian Lageman
Chair: Knut Hüper
10.30 – 11.00
Adaptive Filtering for Estimation of a Low-Rank Positive
Semidefinite Matrix
Silvère Bonnabel, Gilles Meyer, Rodolphe Sepulchre
Abstract – In this paper, we adopt a geometric viewpoint to
tackle the problem of estimating a linear model
whose parameter is a fixed-rank positive semidefinite matrix.
We consider two gradient descent flows associated to two
distinct Riemannian quotient geometries that underlie this set
of matrices. The resulting algorithms are non-linear and can be
viewed as a generalization of Least Mean Squares that
instrically constrain the parameter within the manifold search
space. Such algorithms designed for low-rank matrices find
applications in high-dimensional distance learning problems
for classification or clustering.
11.00 – 11.30
Intrinsic Newton’s Method on Oblique Manifolds for
Overdetermined Blind Source Separation
Martin Kleinsteuber, Hao Shen
Abstract – This
Overdetermined Blind Source Separation (OdBSS), a
challenging problem in signal processing. It aims to recover
desired sources from outnumbered observations without
knowing either the source distributions or the mixing process.
It is well-known that performance of standard BSS algorithms,
which usually utilize a whitening step as a pre-process to
reduce the dimensionality of observations, might be seriously
limited due to its blind trust on the data covariance matrix. In
this paper, we develop and compare two locally quadratic
OdBSS algorithms that forgo the dimensionality reduction
step. In particular, our algorithms solve a problem of
simultaneous diagonalization of a set of symmetric matrices. By
exploiting the appropriate underlying manifold, namely the so-
Local Minima of the Best Low Multilinear Rank
Approximation of Tensors
Mariya Ishteva, Pierre-Antoine Absil, Sabine van Huffel,
Lieven de Lathauwer
Abstract – Higher-order tensors are generalizations of
vectors and matrices to third- or even higher-order arrays
of numbers. We consider a generalization of column and
row rank of a matrix to tensors, called multilinear rank.
Given a higher-order tensor, we are looking for another tensor,
as close as possible to the original one and with
multilinear rank bounded by prespecified numbers. In this
paper, we give an overview of recent results pertaining the
associated cost function. It can have a number of local minima,
which need to be interpreted carefully. Convergence to the
global minimum cannot be guaranteed with the existing
algorithms. We discuss the conclusions that we have drawn
from extensive simulations and point out some hidden
problems that might occur in real applications.
Semiplenary Lecture
Room 1
Chair: Tyrone E. Duncan
14.00 – 15.00
Nonlinear Filtering and Systems Theory
Ramon van Handel
Abstract – The fundamental connection between the
stability of linear filtering and linear systems theory was
already remarked in Kalman’s seminal 1960 paper.
Unfortunately, the linear theory relies heavily on the
investigation of the explicit Kalman filtering equations, and
sheds little light on the behavior of nonlinear filters.
Nonetheless, it is possible to establish surprisingly general
connections between the stability of nonlinear filters and
nonlinear counterparts of basic concepts in linear systems
theory: stability, observability, detectability. The proofs of
these results are probabilistic in nature and provide significant
insight into the mechanisms that give rise to filter stability. The
aim of this paper is to review these recent results and to discuss
some of their applications.
List of Abstracts
Semiplenary Lecture
Room 14
Chair: Clyde F. Martin
Distributed Parameter Systems IV:
Computational Issues
(Invited Session)
Room 1
14.00 – 15.00
Organizers: Birgit Jacob, Michael A. Demetriou, Miroslav
Krstic, Kirsten Morris, Hans Zwart
Chair: Hans Zwart
Controllability of Networked Systems
Magnus Egerstedt
Abstract – In
controllability properties associated with networked control
systems whose information exchange takes place over a static
communication network. The control signal is assumed to be
injected into the network at a given input node and its
influence is propagated through the network through a
nearest-neighbor interaction rule employed to ensure network
cohesion. In particular, the problem of driving a collection of
mobile robots to a given target destination is studied, and
conditions are given for this to be possible, based on tools from
algebraic graph theory. The main result is a necessary and
sufficient condition for an interaction topology to be
controllable, given in terms of the network’s external, equitable
Semiplenary Lecture
Room 15
Chair: Doreen Thomas
14.00 – 15.00
Quantum Linear Systems Theory
Ian R. Petersen
Abstract – This paper surveys some recent results on
the theory of quantum linear systems and presents them
within a unified framework. Quantum linear systems are a
class of systems whose dynamics, which are described by the
laws of quantum mechanics, take the specific form of a set of
linear quantum stochastic differential equations (QSDEs).
Such systems commonly arise in the area of quantum optics
and related disciplines. Systems whose dynamics can be
described or approximated by linear QSDEs include
interconnections of optical cavities, beam-spitters, phaseshifters, optical parametric amplifiers, optical squeezers, and
cavity quantum electrodynamic systems. With advances in
quantum technology, the feedback control of such quantum
systems is generating new challenges in the field of control
theory. Potential applications of such quantum feedback
control systems include quantum computing, quantum error
correction, quantum communications, gravity wave detection,
metrology, atom lasers, and superconducting quantum circuits.
A recently emerging approach to the feedback control
of quantum linear systems involves the use of a controller
which itself is a quantum linear system. This approach to
quantum feedback control, referred to as coherent quantum
feedback control, has the advantage that it does not destroy
quantum information, is fast, and has the potential for efficient
implementation. This paper discusses recent results concerning
the synthesis of H-infinity optimal controllers for linear
quantum systems in the coherent control case. An important
issue which arises both in the modelling of linear quantum
systems and in the synthesis of linear coherent quantum
controllers is the issue of physical realizability. This issue
relates to the property of whether a given set of QSDEs
corresponds to a physical quantum system satisfying the laws
of quantum mechanics. The paper will cover recent results
relating the question of physical realizability to notions
occuring in linear systems theory such as lossless bounded real
systems and dual J-J unitary systems.
15.30 – 16.00
Numerical Approximation of Exact Controls for Vibrating
Nicolae Cîndea, Sorin Micu, Marius Tucsnak
Abstract – The numerical study of the exact controls of
infinite dimensional systems started in the 90's with a series of
papers of Glowinski and Lions (see [3, 4]) where algorithms to
determine the minimal L²-norm exact controls (sometimes
called HUM controls) are provided. Several abnormalities
presented in these works stand at the origin of a large number
of articles in which a great variety of numerical methods are
presented and analyzed (see, for instance, [5], [2] and the
references therein). However, except the recent work [1], where
the approximation of the HUM controls for the one
dimensional wave equation is considered, to our knowledge,
there are no results on the rate of convergence of
the approximative controls.
Our main theoretical result gives the rate of convergence of
our approximations to an exact control. Moreover, to illustrate
the efficiency of this approach, we apply it to several systems
governed by PDE's and we describe the associated numerical
Our methodology
“stabilizability implies controllability" principle with error
estimates for finite element type approximations of the
considered infinite dimensional systems.
16.00 – 16.30
Robust Output Controller Design Based on Adaptive Model
Reduction for Parabolic PDE Systems
Sivakumar Pitchaiah, Antonios Armaou
Abstract – The problem of designing robust feedback
controller for spatially distributed processes, described
by parabolic PDE systems, is addressed by designing
robust output feedback controllers using adaptive proper
orthogonal decomposition methodology (APOD). Initially, an
ensemble of eigenfunctions is constructed based on a relatively
small data ensemble which is then recursively updated as
additional process data becomes available periodically. These
eigenfunctions are then utilized in deriving a reduced order
model (ROM) of the PDE system by employing the Galerin’s
method. The obtained ROM is further utilized for the synthesis
of robust feedback controllers via geometric techniques. Under
the assumption that the number of measurements sensors is
equal to the number of modes of the ROM, we obtain the the
estimates for the states of the ROM using a static observer.
Utilizing these estimated states in the robust controllers leads
to robust output feedback controllers that guarantee
boundedness of the state along with uncertainty attenuation in
the infinite-dimensional system. As new data from the closedloop process becomes available we update the ROM (and hence
the robust controller) by employing APOD. The theoretic
results are successfully applied to a representative example of
dissipative PDEs with nonlinearities and uncertainty.
Friday, 9 July
16.30 – 17.00
16.30 – 17.00
Overview of Consensus Filters for Distributed Parameter
Systems Utilizing Sensor Networks
Michael A. Demetriou
Optimal Event-Triggered Control under Costly Observations
Adam Molin, Sandra Hirche
Abstract – This work summarizes two types of consensus
filters for a class of distributed parameter systems: consensus
and adaptive-consensus filters. Furthermore, it proposes a
metric for comparing the disagreement among the spatially
local filters. It is assumed that a sensor network consists of
groups of sensors, each of which provides a number of state
measurements from sensing devices that are not necessarily
identical to each other and which only transmit their
information to their own sensor group. A metric for examining
the disagreement of the local filters, as extended from the finite
dimensional case, essentially yields a deterministic analog of
the standard deviation of the spatially local filter errors. The
disagreement metric is examined for both consensus and
adaptive consensus filters. The measure of disagreement is
subsequently shown to be linked to the state estimation errors
thereby simplifying the performance analysis to simply that of
stability of the estimating scheme.
17.00 – 17.30
Sensor Network Design for Inverse Problems
Dariusz Uciński
Abstract – The aim of this work is to expose an optimal
node activation algorithm in sensor networks whose
measurements are supposed to be used to estimate unknown
parameters of the underlying process model in the form of a
partial differential equation. By partitioning the observation
horizon into a finite number of consecutive intervals, the
problem is set up to select nodes which will be active over each
interval while the others will remain dormant such that a
general convex design criterion defined on the Fisher
information matrix associated with the estimated parameters is
minimized. The search for the optimal solution is performed
using the branch-and-bound method in which an efficient
technique is employed to produce a lower bound to the
minimum of the objective function.
New Paradigms for Control
(Regular Session)
Room 2
Chair: Fritz Colonius
15.30 – 16.00
Minimal Data Rates and Invariance Entropy
Fritz Colonius
Abstract – For compact locally controlled invariant subsets
of the state space, minimal data rates for achieving invariance
are characterized by the invariance entropy. In particular, for
linear control systems with bounded control range, locally
invariant sets are constructed and the associated minimal data
rates are computed.
16.00 – 16.30
Fast Controls and Their Calculation
Alexander Daryin, Yulia Minaeva
Abstract – New technologies, such as control in
quantum systems, may require that the control would act on a
very small time horizon. Another requirement is that the
control should be designed in a closed-loop form. A possible
response to this demand is the use of fast controls [1]. They are
introduced as bounded approximations of generalized impulse
controls (belonging to the class of higher-order distributions).
Abstract – Digital control design is commonly constrained to
sampling intervals. The emergence of more and more complex
and distributed systems urges the development of advanced
triggering schemes
communication resources efficiently. This paper considers a
linear stochastic continuous-time setting, where the design
objective is to find an event-triggered controller that optimally
meets the trade-off between control performance and resource
utilization. This is reflected by imposing a cost penalty on
updating the controller by current observations that is added
to a quadratic control cost. It is shown that the underlying
optimization problem results in an event-triggered controller,
where the controller is updated, when the estimation error of
the controller exceeds an apriori determined threshold. The
controller design is related to linear quadratic Gaussian
regulation and to optimal stopping time problems. Contrary to
the initial problem, these can be solved by standard methods of
stochastic optimal control. Numerical examples underline the
effectiveness compared to optimal time-triggered controllers.
Electrical Circuits
(Regular Session)
Room 3
Chair: Luigi Fortuna
15.30 – 16.00
Chaos Control in Inductor-Based Chaotic Oscillators
Arturo Buscarino, Luigi Fortuna, Mattia Frasca, Gregorio
Abstract – In this work a new chaos control technique
for inductor-based chaotic oscillators is introduced. The
technique consists of coupling the original circuit with a
further passive control circuitry, made of a coupled inductor
and a variable resistor. The dynamics of the oscillator can be
controlled by varying a single external control parameter, i.e.
the resistor value. The technique has been applied to two case
studies, leading to circuits exhibiting a rich dynamics,
including stable limit cycles of different periods, bistability,
and new regions of chaos.
16.00 – 16.30
A Reduced Model of Reflectometry for Wired Electric
Mohamed Oumri, Qinghua Zhang, Michel Sorine
Abstract – Reflectometry is a technology frequently used for
the diagnosis of failures in wired electric networks. For the
purpose of developing advanced diagnosis methods, a reduced
mathematical model of reflectometry is proposed in this paper.
Based on the telegrapher’s equations and on the Kirchhoff’s
laws, this model leads to a simple algorithm for
the computation of frequency domain reflection coefficients
from the characteristic parameters of the transmission lines
and their connections in a star-shaped or a tree-shaped
network. This algorithm implemented in a digital computer
can easily simulate networks composed of different and
inhomogeneous transmission lines. Comparisons between
simulated reflection coefficients and real reflectometry
measurements confirm the validity of the proposed model.
List of Abstracts
16.00 – 16.30
(Regular Session)
Room 4
Overdetermined Systems on Lie Groups and Their Transfer
Eli Shamovich
Chair: Pierre-Antoine Absil
15.30 – 16.00
An Efficient BFGS Algorithm for Riemannian Optimization
Chunhong Qi, Kyle A. Gallivan, Pierre-Antoine Absil
Abstract – In this paper, we present a convergence result for
superlinear convergence. We also present a theory of building
vector transports on submanifolds of Rn and discuss its use to
assess convergence conditions and computational efficiency of
the resulting Riemannian optimization algorithms. We
illustrate performance and check predictions of our theory
using a version of a Riemannian BFGS algorithm we proposed
16.30 – 17.00
On the Class RSI of Rational Schur Functions Intertwining
Solutions of Linear Differential Equations
Daniel Alpay, Andrey Melnikov, Victor Vinnikov
16.00 – 16.30
Properties of a Parameterized Model Reduction Method
Aivar Sootla
Abstract – In this contribution a recently proposed
model reduction method for a class of linear time-invariant
(LTI) parameterized models is investigated. The method is
based on matching of the frequency response samples using the
semidefinite programming methods. The main focus of this
contribution is the properties of the obtained approximations.
Among those properties is stability of individual LTI systems,
continuity with respect to parameters, error bounds on
approximation quality.
16.30 – 17.00
A Notion of Approximation for Systems Over Finite
Danielle C. Tarraf
17.00 – 17.30
On Ranks of Noncommutative Polynomials
Victor Vinnikov
Abstract – We consider the problem of approximating
plants with discrete sensors and actuators (termed ‘systems
over finite alphabets’) by deterministic finite memory systems
for the purpose of certified-by-design controller synthesis. We
propose a new, control-oriented notion of input/output
approximation for these systems, that builds on ideas from
robust control theory and behavioral systems theory. We
conclude with a brief discussion of the key features of the
proposed notion of approximation relative to those of two
existing notions of finite state approximation and abstraction.
Shadows of Multidimensionality:
Multidimensional Systems with Applications to
1-D Systems – 2
(Invited Session)
Room 6
Organizers: Joseph A. Ball, Victor Vinnikov
Chair: Joseph A. Ball
Abstract – We consider the minimum of the rank of a
given noncommutative polynomial in d noncommuting
indeterminates when evaluated on d-tuples of nxn matrices. It
is conjectured that the minimum of the normalized rank, that
is of the rank divided by n, tends to 0 as n tends to infinity;
an even stronger conjecture is that the minimum of the
rank stabilizes for sufficiently large n. These conjectures
appeared in free noncommutative algebra (the study of two
sided principal ideals in the ring of noncommutative
polynomials). They are important for free noncommutative
algebraic geometry that has emerged recently as a powerful
tool when dealing with dimension independent problems in
systems and control. Of course rank minimization problems
are in general ubiquituous in control and optimization.
Consensus and Games
(Regular Session)
Room 7
15.30 – 16.00
Chair: Xiaoming Hu
Multievolution Scattering Systems and the Multivariable
Schur Class
Dmitry S. Kaliuzhnyi-Verbovetskyi
15.30 – 16.00
Abstract – We show using the multievolution scattering
systems formalism how to obtain the decompositions of
multivariable Schur-class functions which are analogous, to
a certain extent, to Agler’s decompositions of Schur–Aglerclass functions. In particular, this gives a new class of d-tuples
of commuting strict contractions on a Hilbert space which
satisfy the multivariable von Neumann inequality.
Optimal Output Consensus Control and Outlier Detection
Johan Thunberg, Xiaoming Hu
Abstract – In this paper we study the output
consensus problem for systems of agents with linear continuous
time invariant dynamics, and derive control laws that minimize
a conical combination of the energies of the agents control
signals, while only using local information. We show that the
optimal control requires the connectivity graph to be complete
Friday, 9 July
and in general requires measurements of the state errors. We
identify the cases where the optimal control is only based on
output errors, and show that in the infinite time horizon case,
the optimal control can always be expressed as a dynamic
control that is only based on the output errors. We also give a
Lemma for the position of the equilibrium point for a large
class of agent dynamics. As a second part of this paper we
consider the problem of outlier detection, in which an agent
wants to deduce if an other agent is using the consensus
controller, or if it is an outlier that uses a different controller.
We introduce the outlier detection equation.
16.00 – 16.30
Distributed Consensus Under Limited Information
Haopeng Zhang, Qing Hui
Abstract – Consensus for networked control systems has
a significant application in civil and military applications,
while most of the literature focus on the research of
consensus for the networked control system with ideal
measurements. However, in practice, those assumptions can
not be guaranteed properly. Due to the communication link
and information storage memory limitations, quantization
consensus is more reasonable for the networked control system
since the quantized values are less ideal than the perfectly
measured values and much more easier to access and transmit
in practice. In this paper, we present a novel quantized
consensus protocol for the networked control system and prove
that near-consensus is achieved under this protocol. To obtain
the exact-consensus for the quantized system, a distributed
consensus algorithm is further investigated. Finally, the Matlab
simulations are provided to verify our theoretical results.
16.30 – 17.00
Emergence of Lévy Flights in Distributed Consensus
Jing Wang, Nicola Elia
Abstract – In this paper, we consider a multi-agent
model which generates a collective super-diffusion behavior.
Although such complex behaviors are ubiquitous in many
natural and human-made systems, discovering mechanisms of
their emergence is mostly an open research area. Our model is
motivated to solve consensus problem under limitations on
information exchange including link failures and additive
noise. We use orthogonal decomposition approach to analyze
the system and establish several equivalent necessary and
sufficient conditions for Mean Square (MS) stability of part of
this system. We show that the emergence of the super-diffusion
behavior is introduced by the loss of MS stability and prove it
to be Lévy flights for a special system. This work is the first, to
the best of our knowledge, to establish the intimate relationship
between propagation of uncertainties in networks, the MS
stability robustness and the emergence of Lévy flights, which
may have far reaching consequence on the understanding and
engineering of complex systems.
17.00 – 17.30
Anti-Palindromic Pencil Formulation for Open-Loop
Stackelberg Strategy in Discrete-Time
Marc Jungers, Cristian Oară
Abstract – The
gametheoretic framework an adapted concept to obtain
equilibrium for hierarchical games. For linear-quadratic
games, Stackelberg strategy with open-loop information
structure leads to solve non-symmetric Riccati equations, by
assumming the invertibility of some weighting matrices. This
paper provides an antipalindromic pencil approach to
formulate it. Thus it allows to relaxe invertibility assumptions
and furthermore to take advantage of the recent literature of
numerical methods for palindromic pencils.
Robust Convex Control
(Regular Session)
Room 8
Chair: Debraj Chakraborty
15.30 – 16.00
Bang-Bang Solutions to the Open Loop Maximal Time
Debraj Chakraborty
Abstract – The problem of maximizing the duration of
open loop operation of a perturbed linear time invariant
system, while keeping performance errors within bound, is
considered. It was shown in an earlier article that the optimal
control for this problem is purely bang-bang if an associated
switching function is non-zero almost everywhere. Sufficient
conditions are derived in this article to guarantee this
16.00 – 16.30
Robust Stabilization with Real Parametric Uncertainty via
Linear Programming
Svetlana Iantchenko, Andrey Ghulchak
Abstract – A numerical method is proposed for
optimal robust control synthesis. A dual interpretation of the
problem is derived. In the special case, when uncertainty
parameter is real-valued, it is shown that the dual problem
becomes essentially finite dimensional in the space of variables
(semiinfinite convex programming). It makes possible to
efficiently obtain a numerical solution of the dual problem and
to construct the optimal robustly stabilizing controller via
the alignment principle. In order to illustrate the method,
several examples on the robust stabilization with real
uncertainties are solved (both analytically and numerically).
16.30 – 17.00
Treatment of Systems Nonlinearities by a Multiplier Method
Éva Gyurkovics, Tibor Takács
Abstract – This paper investigates the conditions under
which an abstract matrix multiplier method can be applied to
determine guaranteeing cost controls for systems containing
nonlinear/ uncertain elements via linear matrix inequalities
(LMIs). Quadratically
nonlinearities are considered which comprehend the cases of
norm-bounded, positive-real
uncertainties/nonlinearities. Both the discrete-time and the
continuous-time cases are discussed. Necessary and sufficient
conditions are formulated in case of unstructured uncertainty.
The conditions are sufficient in the structured case. The cost
guaranteeing controls can be determined by solving LMIs. The
proposed method provides a guideline to treat system
nonlinearities, if the system dynamics can be formulated as
considered in the paper by an approriate choice of system
17.00 – 17.30
Gain-Scheduled H2 Filter Synthesis via Polynomially
Parameter-Dependent Lyapunov Functions with Inexact
Scheduling Parameters
Masayuki Sato
Abstract – This paper addresses the design problem of GainScheduled (GS) H2 filters for Linear Parameter-Varying
(LPV) systems under the condition that only inexactly
measured scheduling parameters are available. The state-space
matrices of the LPV systems are supposed to be polynomially
parameter dependent and those of filters which are to be
designed are supposed to be rationally parameter dependent.
The uncertainties in the measured scheduling parameters are
List of Abstracts
supposed to lie in a priori defined convex set. Using structured
polynomially Parameter-Dependent
(PDLFs), we give a design method of GS H2 filters, which are
robust against the uncertainties in the measured scheduling
parameters, in terms of parametrically affine Linear Matrix
Inequalities (LMIs). Our proposed method includes robust
filter design as a special case. A numerical example
demonstrates the effectiveness of our method.
global optimization problem with that of the optima reached
under limited network information. In this way, we are able to
incorporate the effect of network control topology into the
computational aspects of the optimization problem. We also
characterize the topological properties of command and
control structures that result in global optima, without the
need to use the full information of the grid.
Robust Control
(Regular Session)
Networked Control – 2
(Regular Session)
Room 10
Room 9
Chair: Shivkumar V. Iyer
15.30 – 16.00
Application of Graph Theory in Stability Analysis of Meshed
Shivkumar V. Iyer, Madhu N. Belur, Mukul C. Chandorkar
Abstract – This paper studies microgrids where loads are
supplied by parallel connected inverters controlled by
decentralized active power/voltage frequency and reactive
magnitude droop
implementation of droop control laws for sharing of power
between inverters has been known to present stability
problems particularly for large values of active power/voltage
frequency droop control gains. Stability analysis of the
microgrid requires a mathematical formulation of
the interaction between the inverters due to their droop control
laws. However, a simple and elegant mathematical model
resulting in a conclusive proof of stability has been found to be
lacking in reported literature. In this paper, a state dynamical
model has been derived by combining active power flow
equations with the active power/voltage frequency droop
control laws. Using an analogy between the model matrix and
connected graphs, a proof of stability of the microgrid has been
stated as a theorem. The paper further examines the
limitations of the proof and the difference between the results
of the proof and reported practical results.
16.00 – 16.30
Power-Aware Sensor Coverage: an Optimal Control
Patrick Martin, Rosalba Galvan-Guerra, Magnus Egerstedt,
Vadim Azhmyakov
Abstract – Sensor
competing objectives: they must sense as much as possible, yet
last as long as possible when deployed. In this paper, we
approach this problem using optimal control. We describe a
model that relates each sensor’s “footprint” to their power
consumption and use this model to derive optimal control laws
that maintain the area coverage for a specified operational
lifetime. This optimal control approach is then deployed onto
different sensor networks and evaluated for its ability to
maintain coverage during their desired lifetime.
16.30 – 17.00
Distributed Electrical Power Distribution Using Evolutionary
Variational Inequalities
Abubakr Muhammad
Abstract – In this paper, we study a projected dynamical
system PDS suggested by Nagurney et al. for the solution of an
electrical power distribution problem. We study the PDS from
the point of view of the topology of the grid control network. In
particular, we first cast this PDS in a graph theoretic (or
topological) form by using the strategy outlined by the author
in a previous paper. Next, by studying the critical points of this
PDS for various topologies, we compare the solution of the
Chair: Didier Henrion
15.30 – 16.00
Convex Inner Approximations of Nonconvex Semialgebraic
Sets Applied to Fixed-Order Controller Design
Didier Henrion, Christophe Louembet
Abstract – We describe an elementary algorithm to
build convex inner approximations of nonconvex sets. Both
input and output sets are basic semialgebraic sets given as lists
of defining multivariate polynomials. Even though no
optimality guarantees can be given (e.g. in terms of volume
maximization for bounded sets), the algorithm is designed to
preserve convex boundaries as much as possible, while
removing regions with concave boundaries. In particular, the
algorithm leaves invariant a given convex set. The algorithm is
based on Gloptipoly 3, a public-domain Matlab package
solving nonconvex polynomial optimization problems with the
help of convex semidefinite programming (optimization over
linear matrix inequalities, or LMIs). We illustrate how the
algorithm can be used to design fixed-order controllers for
linear systems, following a polynomial approach.
16.00 – 16.30
Stability Analysis of Discrete-Time Systems with TimeVarying Delays via Integral Quadratic Constraints
Chung-Yao Kao
Abstract – This manuscript presents certain l2-gain
properties of and the integral quadratic constraint
characterizations derived from these properties for the
time-varying operator.
characterizations are crucial for the IQC analysis to be applied
to study robustness of discretetime systems in the presence of
time-varying delays. One new contribution of this manuscript
is to utilize the information of the variation of the delay
parameter to derive less conservative IQCs. The effectiveness
of the proposed IQC analysis is verified by numerical
experiments, the results of which are compared with those
recently published in the literature.
16.30 – 17.00
Robust Stability Analysis of Inverse LQ Regulator for Linear
Systems with Input Delay
Takao Fujii, Osamu Kaneko
Abstract – We analyze the robust stability of Inverse LQ
regulators for single-input linear systems with uncertain input
delay. Unlike the usual LQ regulator, the Inverse LQ regulator
has a gain tuning pa-rameter that can be chosen freely to some
extent without losing its LQ optimality. Utilizing this freedom
we seek the range of tuning parameter that ensures the robust
stability against the uncertain de-lay time, as well as the robust
stability condition of the Inverse LQ regulator for some gain
tuning parameter. The result is based on the quadratic
stabilization problem.
Friday, 9 July
17.00 – 17.30
16.30 – 17.00
Periodically Time-Varying Controller Synthesis for
Multiobjective H2/H∞ Control of Discrete-Time Systems and
Analysis of Achievable Performance
Yoshio Ebihara, Jun Yamaguchi, Tomomichi Hagiwara
Symmetry in Quantum System Theory of Multi-Qubit
Robert Zeier, Uwe Sander, Thomas Schulte-Herbrüggen
Abstract – In this paper, we propose a linear
periodically time-varying
approach for the multiobjective H2/H∞ control problem of
discrete-time linear time-invariant (LTI) systems. By
artificially regarding the LTI plant as N-periodic and applying
the discrete-time system lifting, we first derive an SDP for the
synthesis of suboptimal multiobjective LPTV controllers.
Furthermore, we show that we can reduce the conservatism
and improve the control performance gradually by simply
increasing the controller period. On the other hand, in the
latter part of the paper, we propose another SDP for the
computation of a lower bound of the control performance that
is achievable via LPTV controllers of any period and order.
Similarly to the LPTV controller synthesis, the SDP is derived
based on the liftingbased treatment of the LTI plant, and it is
shown that we can improve the lower bound gradually by
increasing the fictitious period N. We validate all of these
theoretical results through an illustrative example.
Quantum Systems and Control
(Regular Session)
Room 11
Chair: Francesco Ticozzi
15.30 – 16.00
Schrödinger Bridges for Discrete-Time, Classical and
Quantum Markovian Evolutions
Michele Pavon, Francesco Ticozzi
Abstract – The theory of Schrödinger bridges for
diffusion processes is extended to discrete-time Markov chains,
and to some problems for quantum discrete-time processes.
Taking into account the past-future lack of symmetry of the
discrete-time setting, results bear a striking resemblance to the
classical ones. In particular, the solution of the path space
maximum entropy problems is always obtained from the prior
model by means of a suitable multiplicative functional
16.00 – 16.30
A Unified Approach to Controllability of Closed and Open
Quantum Systems
Indra Kurniawan, Gunther Dirr, Uwe Helmke
17.00 – 17.30
Quantum Stochastic Stability and Weak-* Convergence of
System Observables
Ram Somaraju, Ian R. Petersen
Abstract – The evolution of open quantum systems can be
equations (QSDEs). The solution of QSDEs leads to a one
parameter semigroup of completely positive operators with
which one can associate a minimal quantum Markov dilation.
In this paper, we use a Lyapunov type theorem to prove
asymptotic stability in the weak-* operator norm for such
minimal Markov dilations provided some assumptions are
satisfied. This theorem uses the fact that the unit ball in the
space of bounded operators on a Banach space is weak-*
Systems Biology
(Regular Session)
Room 12
Chair: Krisztina Kiss
15.30 – 16.00
Prey and Polyphagous Predator Species with Diffusion
Zsuzsanna Barta, Krisztina Kiss
Abstract – This paper deals with a ratio–dependent
polyphagous predator–prey system taking into account the
spatial movement of the species. We will investigate
under what conditions Turing stability or instability occurs
in higher dimensions.
16.00 – 16.30
Robust Dynamical Network Reconstruction
Ye Yuan, Guy-Bart Stan, Sean Warnick, Jorge Gonçalves
Abstract–This paper addresses the problem of
robustly reconstructing network structure from input-output
data. Previous work identified necessary and sufficient
conditions for network reconstruction of LTI systems,
assuming perfect measurements (no noise) and perfect system
identification. This paper assumes that the previously
identified necessary and sufficient conditions for network
reconstruction are satisfied but here we additionally take into
unmodelled dynamics
nonlinearities). In order to identify the network structure that
generated the data, we compute the smallest distances between
the measured data and the data that would have been
generated by particular Boolean networks. By striking a
compromise between such distance and network complexity,
we provide methods for revealing the correct network
structure from data despite the presence of noise
and nonlinearities.
List of Abstracts
16.30 – 17.00
MAPK Module: Biological Basis, Structure, Mathematical
Model and Dynamical Analyse
Nataša A. Kablar
Abstract – In this paper we present mitogen-activated
protein kinase (MAPK) module: its biological definition,
structure, and model. In modelling stage, we build on result of
[9], and we include newly experimentally observed processes to
capture more on real dynamic of cell: cross-linking among the
different modules of MAPK and/or cross-linking with other
pathways; influence of Phosphatase’s, and influence of
phosphorylated kinase kinase (KKP) found to have profound
effect on module dynamics. For the chosen set of
experimentally verifiable parameters we perform dynamic
analyze. In investigation of bifurcation, we find Hopf
Bifurcation as the only type of bifurcation observed.
17.00 – 17.30
Estimation of Efficacy of HIV Nucleoside-Analogue Reverse
Transcriptase Inhibitor (AZT) via Stochastic Modeling
Samira Khalili, James M. Monaco, Antonios Armaou
Abstract – In this work, the mechanisms by which
nucleosideanalogue reverse transcriptase inhibitors (NRTIs),
the most common class of drugs used in the treatment of HIV1, exert their antiviral effects are analyzed and methods in
which those known mechanisms could be employed to generate
mathematical models for drug efficacy in terms of measurable
physical values are identified. Drug concentration is considered
as a time variant parameter which depends on the drug
administration time and dosage.
closedloop system. We obtain equations for the optimal
controller and outline a homotopy method which reduces the
solution of the problem to the numerical integration of a
differential equation initialized by the standard linear
quadratic Gaussian controller.
16.30 – 17.00
Anisotropy-Based Bounded Real Lemma
Alexander P. Kurdyukov, Eugene A. Maximov, Michael M.
Abstract – This paper extends the Bounded Real Lemma
of the H∞-control theory to stochastic systems under
random disturbances with imprecisely known probability
distributions. The statistical uncertainty is measured in
entropy theoretic terms using the mean anisotropy functional.
The disturbance attenuation capabilities of the system are
quantified by the anisotropic norm which is a stochastic
counterpart of the H∞- norm. We develop a state-space
criterion for the anisotropic norm of a linear discrete time
invariant system to be bounded by a given threshold value. The
resulting Anisotropy-based Bounded Real Lemma involves an
inequality on the determinant of a matrix associated with a
parameter-dependent algebraic Riccati equation.
Economics and Systems Theory
(Regular Session)
Room 14
Chair: Giacomo Como
15.30 – 16.00
Stochastic Control
(Regular Session)
Room 13
Chair: Alexander Yu. Mazurov
15.30 – 16.00
Risk-Sensitive Dissipativity and Relevant Control Problems
Alexander Yu. Mazurov, Pavel V. Pakshin
Abstract – The paper is focused on control-affine
stochastic Itô systems with control-quadratic storage functions.
The concept of dissipativity with risk-sensitive storage function
(RSSF) is proposed, with dissipativity criterion derived
involving generalized Hamilton-Jacobi-Bellman inequalities.
The proof utilizes a certain version of stochastic Artstein’s
inequality. Connections to risk-sensitive suboptimal control,
the theory of games, invariant probabilistic measure and
deterministic H∞-control are established. In linear-quadratic
case the results are expressed via linear matrix inequalities
(LMI). An example is provided.
16.00 – 16.30
Hardy-Schatten Norms of Systems, Output Energy
Cumulants and Linear Quadro-Quartic Gaussian Control
Igor G. Vladimirov, Ian R. Petersen
Abstract – This
stochastic control systems in state space. The integral of the
squared norm of the system output over a bounded time
interval is interpreted as energy. The cumulants of the output
energy in the infinite-horizon limit are related to Schatten
norms of the system in the Hardy space of transfer functions
and the risksensitive performance index. We employ a novel
performance criterion which seeks to minimize a combination
of the average value and the variance of the output energy of
the system per unit time. The resulting linear quadro-quartic
Gaussian control problem involves the H2 and H4-norms of the
On Robustness Analysis of Large-Scale Transportation
Giacomo Como, Ketan Savla, Daron Acemoglu, Munther A.
Dahleh, Emilio Frazzoli
Abstract–In this paper, we study robustness properties
of transportation networks with respect to its predisturbance equilibrium operating condition and the agents’
response to the disturbance. We perform the analysis within a
dynamical system framework over a directed acyclic graph
between a single origin-destination pair. The dynamical system
is composed of ordinary differential equations (ODEs), one for
every edge of the graph. Every ODE is a mass balance equation
for the corresponding edge, where the inflow term is a function
of the agents’ route choice behavior and the arrival rate at
the base node of that edge, and the outflow term is function of
the congestion properties of the edge.We consider disturbances
that reduce the maximum flow carrying capacity of the links
and define the margin of stability of the network as the
minimum capacity that needs to be removed from the network
so that the delay on all the edges remain bounded over time.
For a given equilibrium operating condition, we derive
upper bounds on the margin of stability under local
information constraint on the agents’ behavior, and
characterize the route choice functions that yield this bound.
We also setup a simple convex optimization problem to find the
most robust operating condition for the network and determine
edge-wise tolls that yield such an equilibrium operating
16.00 – 16.30
Combining the Frisch Scheme and Yule-Walker Equations
for Identifying Multivariable Errors-in-Variables Models
Roberto Diversi, Roberto Guidorzi
Abstract – Errors–in–Variables
models whose stochastic environment considers measurement
errors on both inputs and outputs are intrinsically more
realistic than representations assuming an exact knowledge of
Friday, 9 July
the input but are also more difficult to estimate. The difficulties
increase in a non trivial way passing from the SISO and MISO
cases to the MIMO one. This paper proposes a procedure
for EIV identification of MIMO processes based on the
Frisch scheme that assumes additional white noises on all
inputs and outputs and shows its effectiveness by means of
Monte Carlo simulations.
16.30 – 17.00
Recursive Estimation of GARCH Processes
László Gerencsér, Zsanett Orlovits, Balázs Torma
Abstract – ARCH processes and their extensions known
as GARCH processes are widely accepted for modelling
financial time series, in particular stochastic volatility
processes. The offline estimation of ARCH and GARCH
processes have been analyzed under a variety of conditions in
the literature. The main contribution of this paper is a rigorous
convergence analysis of a recursive estimation method for
GARCH processes with large stability margin under
reasonable technical conditions. The main tool in the
convergence analysis is an appropriate modification of the
theory developed by Benveniste, Métivier and Priouret.
17.00 – 17.30
Stochastic Calculus of Heston's Stochastic-Volatility Model
Floyd B. Hanson
Abstract – The Heston (1993) stochastic–volatility model is a
square–root diffusion model for the stochastic–variance.
It gives rise to a singular diffusion for the distribution
according to Feller (1951). Due to the singular nature, the timestep must be much smaller than the lower bound of the
variance. Several transformations are introduced that lead to
proper diffusions including a transformation to an additive
noise model with perfect-square solution, an exact, nonsingular
solution special case and an alternate model. Simulation
solution examples are also given.
Differential Geometric Methods for
Computational Engineering Applications – 2
(Invited Session)
Room 15
Organizers: Knut Hüper, Christian Lageman
Chair: Christian Lageman
15.30 – 16.00
A Filtering Technique on the Grassmann Manifold
Quentin Rentmeesters, Pierre-Antoine Absil, Paul van
Abstract–In this paper, a filtering technique that deals
with subspaces, i.e., points on the Grassmann manifold, is
proposed. This technique is based on an observer design where
the data points are seen as the outputs of a constant velocity
dynamical model. An explicit algorithm is given to efficiently
compute this observer on the Grassmann manifold. This
approach is compared to a particle filtering technique and
similar results are obtained for a lower computational cost.
Some extensions of the filter are also proposed.
16.00 – 16.30
Rotation Averaging and Weak Convexity
Richard Hartley, Jochen Trumpf, Yuchao Dai
Abstract–We generalize the concept of geodesic convexity
in the Special Orthogonal Group SO(3) and apply the
generalization to the discussion of rotation averaging. As a
result we are able to derive strong and new theorems about the
location of global minima of the rotation averaging cost
function. A brief discussion of the relationship of our results to
previous results from the literature will be provided, as well as
an application to camera rig calibration in computer vision.
16.30 – 17.00
On the Computation of Means on Grassmann Manifolds
Knut Hüper, Uwe Helmke, Sven Herzberg
Abstract–Given a set of data points on a
Grassmann manifold sufficiently close to each other, one way
to define their centroid or geometric mean is via the minimizer
of a certain cost function. If one chooses the cost as the sum of
squared geodesic distances between a given point and all the
data points we end up with the definition of the Karcher mean.
In this paper we analyze the critical points for this cost
17.00 – 17.30
Joint Subspace Intersections as a Fitting Problem
Christian Lageman, Knut Hüper
Abstract–In this paper we consider the task of estimating a
joint intersection of several subspaces from perturbed
measurements of the subspaces. We treat this problem as a
fitting problem on a Grassmann manifold. A potential
application in face recognition is discussed.
Was this manual useful for you? yes no
Thank you for your participation!

* Your assessment is very important for improving the work of artificial intelligence, which forms the content of this project

Related manuals

Download PDF