2014/03/07 by Saikat Sarkar, Sarkar, Saikat, Debasish Roy +3
Computer Science · Physics and Astronomy · #Chaos control and synchronization #FOS: Computer and information sciences #Methodology (stat.ME) #Neural Networks and Applications #Statistical Mechanics and Entropy
paper · pdf · doi:10.48550/arxiv.1403.1680
openalex publication_date 2014/03/07 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28
A new global stochastic search, guided mainly through derivative-free\ndirectional information computable from the sample statistical moments of the\ndesign variables within a Monte Carlo setup, is proposed. The search is aided\nby imparting to a directional update term, which parallels the conventional\nGateaux derivative used in a local search for the extrema of smooth cost\nfunctionals, additional layers of random perturbations referred to as\n'coalescence' and 'scrambling'. A selection scheme, constituting yet another\navenue for random perturbation, completes the global search. The\ndirection-driven nature of the search is manifest in the local extremization\nand coalescence components, which are posed as martingale problems that yield\ngain-like update terms upon discretization. As anticipated and numerically\ndemonstrated, to a limited extent, against the problem of parameter recovery\ngiven the chaotic response histories of a couple of nonlinear oscillators, the\nproposed method apparently provides for a more rational, more accurate and\nfaster alternative to most available evolutionary schemes, prominently the\nparticle swarm optimization.\n