Plenarvorträge - DPG-Tagungen

Dynamik und Statistische Physik Dienstag
the true data generating distribution (the disorder) by an average over
an ensemble of surrogate data sets which are generated by sampling from
a set of available data. This offers the advantage that the data generating
process is known and can be controlled by the experimenter. Using tools
from the physics of disordered materials, we develop a general framework
to calculate approximate analytical Bootstrap averages. We apply
our method to the Bootstrap of Gaussian process models which can be
understood as (Bayesian) feed-forward neural networks with infinitely
many neurons in the hidden layer. Our method for the analytical calculation
of Bootstrap averages works on real data and yields quantitative
results which are reliable and faster to compute than Monte-Carlo averages.
The results can be used to evaluate and optimize the learning
performance.
DY 21.5 Di 11:15 H3
Strukturen polymer Netzwerke im Vergleich — •Michael Lang,
Dietmar Göritz und Stefan Kreitmeier — Fakultät für Physik,
Universität Regensburg, 93040 Regensburg
DY 22 Fractals and Nonlinearity II
Makroskopische Eigenschaften eines polymeren Netzwerkes werden auf
mikroskopischer Skala durch dessen chemische und topologische Struktur
definiert. Diese Struktur kann insbesondere mit statistischen Methoden
analysiert und beschrieben werden. Dieser Beitrag fokussiert dabei auf die
Unterschiede, die die Verwendung verschiedener Reaktionsmechanismen
bezüglich der entstehenden Struktur zeigen, und die Auswirkungen die
sich auf das Verhalten des Gesamtsystems abschätzen lassen. Die theoretischen
Überlegungen werden dabei durch zusätzliche Simulationen der
Netzwerkbildung ergänzt und gestützt.
Im Einzelnen wird eine allgemeine Theorie zur Beschreibung der Kettenlängen
zwischen den Netzstellen abgeleitet, die Berechnung des Zyklenranges
eines Netzwerkes verallgemeinert, und anschließend die gewonnenen
Ergebnisse für besondere Netzwerktypen im Vergleich vorgestellt.
Unter anderem kann dabei gezeigt werden, daß nahezu ideale endvernetzte
Syteme kaum realisiert werden können, und statistische Vernetzungsreaktionen
deutlich weniger anfällig gegenüber Störungen der
Reaktion sind.
Zeit: Dienstag 11:30–12:30 Raum: H2
DY 22.1 Di 11:30 H2
Noise Induced Transition from Translational to Rotational
Motion of Swarms — •Udo Erdmann 1 , Werner Ebeling 1 ,
and Alexander S. Mikhailov 2 — 1 Institut für Physik, Humboldt-
Universität zu Berlin, Newtonstraße 15, 12489 Berlin — 2 Department of
Physical Chemistry, Fritz-Haber-Institut der Max-Planck-Gesellschaft,
Faradayweg 4-6, 14195 Berlin
We report a new type of noise-induced transition represented by the
exchange of stability between the translational and rotational modes of
a system of Active Brownian particles. Active Brownian particles are a
promising model to resemble the basic features of biological active motion
of swarms. First we discuss the basic model of interacting Active
Brownian particles in two dimensions and introduce the Rayleigh model
of active friction. New solutions for the translational modes are presented
in the deterministic and in the stochastic descriptions. The border of stability
of the translational mode is investigated. A noise induced transition
to the rotational mode is observed and investigated. Finally the theory
is compared with simulations.
DY 22.2 Di 11:45 H2
Neural Cryptography by Synchronisation of Chaotic Maps —
•Wolfgang Kinzel 1 and Ido Kanter 2 — 1 Theoretische Physik, Universität
Würzburg — 2 Theoretical Physics, Bar Ilan Universty, Israel
Synchronisation of neural networks by mutual learning is combined
with synchronisation of chaotic maps by an external signal. Analytical
and numerical calculations show that the security of neural cryptography
is improved by this new mechanism.
Ref.: R. Mislovaty, E. Klein, I. Kanter and W. Kinzel, Phys. Rev. Lett.
91, 118701 (2003)
DY 23 Statistical Physics of RNA
DY 22.3 Di 12:00 H2
Detection of mutual phase synchronization by space-time clustering
— •Axel Hutt — Weierstrass-Institute fuer Angewandte Analysis
und Stochastik, Mohrenstr.39, 10117 Berlin
Recent findings in biomedical signal analysis revealed the importance
of phase synchronization for the understanding of the dynamics of biological
systems. Especially mutual phase synchronization is assumed
to reflect self-organized microscopic behaviour. The talk introduces the
major idea of a novel segmentation algorithm for mutual phase synchronization
in multivariate time series. The major feature is an extention of
the k-means algorithm in the topological plane to the k-means algorithm
on the n-dimensional torus. Applications to coupled chaotic systems and
empirical brain signals demonstrate properties of the method.
DY 22.4 Di 12:15 H2
Propagation von Reaktions-Diffusions-Wellen in einem
Belousov-Zhabotinsky-System mit anomaler Dispersion —
•Niklas Manz und Oliver Steinbock — Department of Chemistry
and Biochemistry, Florida State University, Tallahassee, FL 32306-4390,
USA
Wir präsentieren neuartige Wellendynamiken in einem homogen katalysierten
Reaktions-Diffusions-System mit anomaler Dispersion. In der
verwendeten Belousov-Zhabotinsky-Reaktion wird das klassische organische
Substrat Malonsäure durch 1,4-Cyclohexandion ersetzt. Die nichtmonotone
Dispersionsrelation führt zu Wechselwirkungen zwischen einer
führenden und einer nachfolgenden Welle, die in Systemen mit monoton
steigender Dispersionsrelation nicht zu finden sind. Dieses Verhalten
wird als ”stacking”, ”merging” und ”tracking” bezeichnet. Befindet sich
das System im ”tracking”-Regime ergibt sich ausserdem die Möglichkeit
einer bisher nicht beobachteten Spiraldrift.
Zeit: Dienstag 11:45–13:00 Raum: H3
DY 23.1 Di 11:45 H3
Dependence of the RNA secondary structure from the energy
model — •Bernd Burghardt and Alexander K. Hartmann —
Institut für Theoretische Physik, Universität Göttingen
In the literature mainly two different types of energy models are considered
for RNA secondary structures: On the one hand side every paired
base pair is assigned an energy, on the other hand only so called stacked
base pairs, e.g. two or more consecutive base pairs, are assigned an energy.
We examined a model that uses both types of energy contributions,
and therefore we were able to study the transition from one type of the
above mentioned energy models to the other one. We have done numerical
studies on statistical quantities, where we used algorithms that are able
to calculate the partition function and the ground-state energy exactly
in polynomial time.
DY 23.2 Di 12:00 H3
Bundle Formation of DNA like polyelectrolytes — •Christian
Holm, Hans Jörg Limbach, and Mehmet Sayar — Max-Planck-
Institut für Polymerforschung, Ackermannweg 10, 55128 Mainz
The physics of attractive interactions among like-charged polyelectrolytes
such as biopolymers like DNA and F-actin, as well as synthetic
polymers has been an active area of research in recent years. Such attractive
forces contradict the expectations of mean field theories. These forces
could lead to formation of well-defined nanoscale organizations, and
therefore a through understanding of this mechanism is of fundamental
importance for design and control of biological and synthetic nanoscale
systems. Here, bundle formation in semi-flexible polyelectrolytes with
short ranged hydrophobic interactions is studied via molecular dynamics
simulations. Bundles with finite radial and axial aggregation number
have been observed in experiments conducted on such polyelectrolytes.