Statistical behavior of adaptive multilevel splitting algorithms in simple models

Abstract

Adaptive multilevel splitting algorithms have been introduced rather recently for estimating tail distributions in a fast and efficient way. In particular, they can be used for computing the so-called reactive trajectories corresponding to direct transitions from one metastable state to another. The algorithm is based on successive selection-mutation steps performed on the system in a controlled way. It has two intrinsic parameters, the number of particles/trajectories and the reaction coordinate used for discriminating good or bad trajectories. We investigate first the convergence in law of the algorithm as a function of the timestep for several simple stochastic models. Second, we consider the average duration of reactive trajectories for which no theoretical predictions exist. The most important aspect of this work concerns some systems with two degrees of freedom. They are studied in details as a function of the reaction coordinate in the asymptotic regime where the number of trajectories goes to infinity. We show that during phase transitions, the statistics of the algorithm deviate significatively from known theoretical results when using non-optimal reaction coordinates. In this case, the variance of the algorithm is peaking at the transition and the convergence of the algorithm can be much slower than the usual expected central limit behavior. The duration of trajectories is affected as well. Moreover, reactive trajectories do not correspond to the most probable ones. Such behavior disappears when using the optimal reaction coordinate called committor as predicted by the theory. We finally investigate a three-state Markov chain which reproduces this phenomenon and show logarithmic convergence of the trajectory durations.

23 pages, 11 figures, accepted for publication in the Journal of Computational physics. A sign typo has been corrected in formula 11, at the bottom of page 6

References in corpus (4)

• The large deviation approach to statistical mechanics
• Random changes of flow topology in two dimensional and geophysical turbulence
• Computing transition rates for the 1-D stochastic Ginzburg--Landau--Allen--Cahn equation for finite-amplitude noise with a rare event algorithm
• On the length of one-dimensional reactive paths

Cited by in corpus (14)

• The instanton method and its numerical implementation in fluid mechanics
• Computing transition rates for the 1-D stochastic Ginzburg--Landau--Allen--Cahn equation for finite-amplitude noise with a rare event algorithm
• Computing return times or return periods with rare event algorithms
• Rare event computation in deterministic chaotic systems using genealogical particle analysis
• Applications of large deviation theory in geophysical fluid dynamics and climate science
• Extreme events in transitional turbulence
• Multistability and rare spontaneous transitions in barotropic $\beta$-plane turbulence
• Application of Adaptive Multilevel Splitting to High-Dimensional Dynamical Systems
• Coupling rare event algorithms with data-based learned committor functions using the analogue Markov chain
• Numerical study of extreme mechanical force exerted by a turbulent flow on a bluff body by direct and rare-event sampling techniques
• Collapse of transitional wall turbulence captured using a rare events algorithm
• On a new class of score functions to estimate tail probabilities of some stochastic processes with Adaptive Multilevel Splitting
• Computing non-equilibrium trajectories by a deep learning approach
• Does rare, noise-induced, bypass transition in plane Couette flow bypass instantons ?