Title :
Almost sure convergence of the BMP scheme with resetting
Author :
Gerencser, Laszlo ; Matyas, Zalan
Author_Institution :
MTA SZTAKI, Comput. & Autom. Inst., Budapest, Hungary
Abstract :
We consider stochastic approximation algorithms with Markovian dynamics as introduced in Benveniste, Métivier and Priouret [2]. Using a resetting mechanism with a fairly arbitrary truncation domain, the algorithm is shown to converge to the unique stationary point of the associated ODE with probability one. A self-contained outline to the basic technical aspects of the BMP theory will be also given.
Keywords :
Markov processes; approximation theory; convergence; BMP scheme; Markovian dynamics; ODE; almost sure convergence; arbitrary truncation domain; probability; resetting mechanism; stochastic approximation algorithms; Approximation methods; Convergence; Heuristic algorithms; Kernel; Markov processes; Poisson equations; Vectors; Markovian dynamics; almost sure convergence; recursive estimation; resetting; stochastic systems;
Conference_Titel :
Control Conference (ECC), 2007 European
Conference_Location :
Kos
Print_ISBN :
978-3-9524173-8-6
