Skip to main content
European Commission logo print header
Contenido archivado el 2022-12-23

Statistical physics of collective behaviour in disordinate systems and information processing

Objetivo



The Budapest team can contribute to the collaboration within the Network in three major areas:
(i) The techniques developed over the past years for long-ranged or high dimensional spin glasses Will be generalized to realistic, low dimensional systems and also to related problems like random manifolds. Through the calculation of finite size corrections a connection will be established between analytic theory and numerical simulations, while the study of the chaotic nature of the spin glass phase will try to make contact with ageing experiments and off-equilibrium investigations. (ii) In the field of neural networks we intend to work on quasiphysical modeling of Kohonen's algorithms, carrying out vector quantization in conjunction with some learnable tasks (classification or topological ordering). In addition we start research on the combination of genetic algorithms with neural networks.
(iii) In our studies of the dynamics of driven interfaces we shall concentrate on the critical behaviour of the surfaces in disordered systems at the depinning transition. We plan to investigate the above problem by examining Edwards-Wilkinson and Kardar-parisi-Zhang type stochastic differential equations using numerical methods as well as scaling arguments. Interface evolution equations where disorder is represented by quenched additive noise will be considered. It is expected that quenched noise leads to a number of novel results for the critical exponents and other parameters associated With the transition.

Convocatoria de propuestas

Data not available

Régimen de financiación

CSC - Cost-sharing contracts

Coordinador

Centre National de la Recherche Scientifique (CNRS)
Aportación de la UE
Sin datos
Dirección
24 rue Lhomond
75231 Paris
Francia

Ver en el mapa

Coste total
Sin datos

Participantes (1)