|
Granichin O.N. (SPbSU) What is the actual structure of
complex information-control systems? 3 Kuchumov R. I. (PetrSU) Implementation and analysis of
work-stealing task scheduler20
Senin I. (SPbSU) A randomized algorithm for
processing ultrasound diagnostics data 87 Khlebnikov M. V. (ICS RAS) Extertnal estimation of
reachability sets for linear dynamical systems 112 Bykov A. V., Scherbakov P. S.
(ICS RAS) A nonconvex matrix sparsity detector with applications to optimal
control of linear systems 123 Abstracts . . . . . .
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142 |
ABSTRACTS
What is the Actual Structure of Complex
Information-Control Systems?
O. N. Granichin
Saint Petersburg State University
o.granichin@spbu.ru
Key words: clustering, self-organizing,
adaptive systems, multi-agent systems,
time-varying state space structure, airplane with “feathers”.
The paper describes the main underlying
features of the author’s new RNF project which is aimed at the development of
model predictive adaptive control with time-varying state space structure.
In the recent years, control of multi-agent
robotic systems has received a great deal of attention in science and
technology. The miniaturization and increased computing power of sensors and
actuators can further extend the range of applicability of the results of
modern control, identification, and estimation theories. In particular,
theoretical issues of adaptive control in changeable environment and under
time-varying structure of the state space have been only marginally considered
so far due to the limited ability of practical realization, and require more
attention.
Use of traditional mathematical models of
motion of systems with a large number of transducers / sensors and actuators
often leads to very complex problems involving extremely high-dimensional state
spaces. The multi-agent technology can effectively solve many of the problems arising in
this context by replacing the general model of interaction with a complex
system having multiple local models and their aggregation (clustering). In this
paper, for the case of time-varying structure of the wind disturbances, we
extend our previous results on synchronization and consensus in the network
control theory toward the flight control when a vast array of sensors and
actuators (“feathers”) is distributed over the surface of an airplane. We study
the possibilities of using Local Voting Protocol for the adaptations of
airplane’s feathers in a turbulence flow.
Bibliogr.: 14 refs.
Implementation
and Analysis of Work-stealing Task Scheduler
R. I. Kuchumov
Petrozavodsk State University
kuchumovri@gmail.com
Key words:
Task schedulers, work-stealing, deques, concurrent computing, multithreading.
Concurrent work-stealing task schedulers
yield near-optimal task distribution and also have low execution times, memory,
and synchronization overhead. The essence of this strategy is that when a
processor runs out
of tasks to execute, it starts to steal tasks from other processors. One
of the drawbacks of this strategy is a large number of steals when dealing with
relatively small tasks. This paper presents an implementation of a
work-stealing task scheduler that allows to take more than one task per single
steal operation. With this modification, the total number of steals can be
significantly reduced.
Bibliogr.: 14 refs.
Stochastic
Network Flow Processes
N. V. Malkovsky
Saint Petersburg State University malkovskynv@gmail.com
Key words:
network flow, stochastic systems, optimal control, capacity constraints.
Network flow processes are being intensively analyzed for many decades.
Such processes emerge naturally when the dynamics of a system is subject to
so-called capacity constrained. These type of constrains may represent a variety
of situations from the width of the road to the speed of communication channels
in information networks. In this paper, one class of such flow problems is
discussed. A general description of the problem is as follows: How can we
transit the system from one state to another in a minimum possible time by
applying a flow process? Methods for both deterministic and stochastic cases
are presented and analysed in detail.
Bibliogr.: 51 refs.
A Randomized Algorithm for Processing Ultrasound Diagnostics
Data
I. Senin
Saint Petersburg State University i.senin@2012.spbu.ru
Key words: compressive sensing, ultrasound tomography,
travel time tomography, randomized
measurements.
Ultrasound tomography has been successfully applied in medical diagnostics,
in particular, in breast cancer diagnostics. The progress in technologies
allows for a significant increase in the amount of sensors used within imaging
devices; this leads to a rapid growth of data to be processed. On the other
hand, tomography images are known to have has sparse representation. According
to the recently developed theory of Compressive Sensing (CS), it is possible to
recover such kind of images from undersampled data. In this paper, an improved
reconstruction algorithm for travel time tomography is proposed, which utilizes
image sparsity for CS. Numerous experiments show that the developed algorithm
exposes the same performance (quality of images) while using considerably less
amount of data.
Bibliogr.: 31 refs.
Extertnal Estimation of Reachability Sets for Linear
Dynamical Systems
M. V. Khlebnikov
Institute of Control Sciences RAS mkhlebnikov2008@yandex.ru
Key words: Linear dynamical system,
reachability set, invariant ellispoid, linear
matrix inequalities.
Simple extertnal estimates of reachability sets
for linear dynamical systems are proposed. They are represented as the
intersection of inva-riant ellispoids, whose number is less then the dimension
of the system. The explicit relations for the determination of such ellipoids
is given. A numerical exapmle is analyzed.
Bibliogr.: 12 refs.
A Nonconvex Matrix Sparsity Detector with Applications
to Optimal Control of Linear Systems
A. V. Bykov
P. S. Shcherbakov
Institute of Control
Science
alexey.bykov.mipt@gmail.com
cavour118@mail.ru
Key words:
linear systems, sparse feedback, H∞-optimization, LMIs, DC-functions, CCCP.
In [8], an
approach to sparse feedback design in linear control systems was proposed,
leading to row/column sparse controller gain matrices, i.e., those containing
zero rows/columns. To solve this problem, a special matrix norm was considered
such that it was used as a convex surrogate for the inherently nonconvex
original problem. The contribution of this current paper is twofold. First, we
propose a different
surrogate which shows itself more efficient, though also being nonconvex. Second, we
extend the original approach toward the classical H∞-optimization
problem. A heuristic in support of the efficiency of the new surrogate is given and
numerical examples are presented that testify to a better performance of the
new method.
Bibliogr.: 19 refs.