low-dimensional collective chaos in strongly- and globally-coupled noisy maps hugues chaté service...
Post on 18-Dec-2015
214 views
TRANSCRIPT
LOW-DIMENSIONAL COLLECTIVE CHAOS IN
STRONGLY- AND GLOBALLY-COUPLED NOISY MAPS
Hugues ChatéService de Physique de L'Etat Condensé, CEA-Saclay, France
Silvia De Monte Ecologie Théorique, Ecole Normale Supérieure, Paris, France
Francesco d'OvidioMétéorologie Dynamique, Ecole Normale Supérieure, Paris, France
Erik MosekildeDepartment of Physics, The Technical University of Denmark
OUTLINE
Phenomenology: identical and noisy globally-coupled maps
single-element versus mean-field
Motivation: a system globally-coupled biological oscillators
role of microscopic disorder? extract microscopic info from measurements?
Order parameter expansion: effective macroscopic dynamics
hierarchical structure and dimensionality of the macroscopic attractor
*
*
*
METABOLIC OSCILLATIONS IN YEAST CELLS
S. Dano, P. G. Soerensen, F. Hynne, Nature 402 (1999)
●macroscopic measurements only (so far)
●role of microscopic disorder (internal/external)
●use noise to extract microscopic information from measurements?
POPULATIONS
MICROSCOPIC FEATURES
SINGLE-ELEMENT DYNAMICSDISORDER DISTRIBUTION
MACROSCOPIC OBSERVABLES
AVERAGES OVER THE POPULATIONORDER PARAMETERS
Additive noise: stochastic process from a given distribution
MODEL: GLOBALLY-COUPLED NOISY MAPS
Single-element dynamics(chaotic, excitable)
Coupling strength
Global coupling
Large body of work on noiseless case, mostly in weak-coupling limit
(clustering, existence of collective dynamics, dimensionality…)
Here consider noisy chaotic maps in strong-coupling regime and
work around the fully-synchronized deterministic limit
This regime first considered by Teramae and Kuramoto
« anomalous scaling »
SPIRIT OF OUR APPROACH
Beyond finite-size effects: bifurcation diagram of collective observable vs noise strength
N=50
N=100
Beyond finite-size effects: noise-induced bifurcations
N=1000
N=10000
N=1000000
NOISE-INDUCED MACROSCOPIC REGIMES
Macroscopic bifurcation diagram
N 106
K 0.4
Snapshot pdfs
Macroscopic bifurcation diagram suggests theexistence of an effective, low-dimensional dynamical system acting on macroscopic observables such as X
This observation is at the heart of our approach,building hierarchically such an effective description
ORDER PARAMETER EXPANSION: dynamics of mean-field
ORDER PARAMETERS
Noise term
Change of variables Series expansion
ORDER PARAMETER EXPANSION
Infinite population size
n-th order REDUCED SYSTEM: truncation to
n-dimensional map
slaved variables
ZEROTH-ORDER REDUCED SYSTEMtrivial result for K=1
Reduced system 2
Population of logistic maps
● Interaction between nonlinearities of the single element and noise features● Classification of noise distributions according to their macroscopic effect
ZEROTH-ORDER APPROXIMATIONquartic maps with different noise distributions
Gaussian noise
Uniform noise
X 1 a 2 bm4 a 6b 2 X2 bX4
x 1 ax2 bx4Single-element dynamics:
Reduced system
HIGHER-ORDER APPROXIMATIONSmacroscopic bifurcation diagram of logistic maps
Period 2
Period 4Chaos
second order
fourth order
x 1 ax2Single-element dynamics:
Reduced system to the second order
X 1 aX2 a 2
22 1 K
2a2 m4 3 4 4X2
2 3 22
HIGHER-ORDER APPROXIMATIONSfine structure of the macroscopic attractor
Folding of the first return map The order parameter expansion captures the hierarchical structure of the macroscopic attractor
Zeroth-order
Second order
Fourth order
Population
HIGHER-ORDER APPROXIMATIONSmacroscopic Lyapunov exponents
calculate Lyapunov exponents/vectors directly from evolution of distribution p(x)
1st vector
HIGHER-ORDER APPROXIMATIONSmacroscopic Lyapunov exponents
compare with exponents of reduced system(s)
Largest Lyapunov exponent
HIGHER-ORDER APPROXIMATIONSmacroscopic Lyapunov exponents
ANOMALOUS SCALING...
K=0.3
K=0.315
K=0.32
Rather well captured by expansion for K not too small...
Over finite range of K, or only at loss of synchronisation?If only at Kc, continuous or subcritical transition?
= normal scaling=
ANOMALOUS SCALING AND LYAPUNOV ANALYSIS
K=0.315
K=0.4
One positive exponent and infinitely-many ~1/σ in anomalous scaling regionNormal scaling: all negative exponents finite as σ=0
Preliminary results from direct pdf simulations...
CONCLUSIONS AND PERSPECTIVES
Microscopic disorder 'unfolds' the synchronous dynamics of globally and strongly coupled maps.
Expansion provides a quantitatively-accurate hierarchical description of the collective dynamics in terms of macroscopic degrees of freedom and parameters.
Anomalous scaling:● existence over finite range of parameters?● only at breakdown of synchronisation?● universality of transition?
Applications: use effect of noise to learn about microscopicsfrom global measurements
GLOBALLY-COUPLED MAPS WITH PARAMETER MISMATCH
COHERENT REGIMES
Different effects of noise and parameter mismatch are captured by the order parameter expansion
Period 2Chaos
● Dependence on the system size● Convergence for maximal coupling
EFFECT OF NOISE ON INDIVIDUAL AND MACROSCOPIC TRAJECTORIES
Weak noise Strong noise
Mean field pdfsingle-element pdf
Time series
Onset of macroscopic oscillations
*
Population of excitable maps:
ZEROTH-ORDER APPROXIMATIONnon-polynomial maps
NOISE-INDUCED COHERENCE/COHERENCE RESONANCE
Breakdown of approximation at large
Reduced system for small and large K.