Options
Convergence with probability one of stochastic approximation algorithms whose average is cooperative
Auteur(s)
Date de parution
2000
In
Nonlinearity
Vol.
3
No
13
De la page
601
A la page
616
Mots-clés
Résumé
We consider a stochastic approximation process Xn+1 - x(n) = Yn+1 (F(x(n)) + Un+1) where F : R-m --> R-m is a C-2 irreducible cooperative dissipative vector field, {y(n)}(n greater than or equal to 0) is a sequence of positive numbers decreasing to 0 and {U-n}(n greater than or equal to 0) a sequence of uniformly bounded R-m martingale differences. We show that under certain conditions on {y(n)} and {U-n} the sequence {x(n)}(n greater than or equal to 0) converges with probability one toward the equilibria set of the vector field F.
Type de publication
Resource Types::text::journal::journal article
