Almost sure convergence of randomly truncated stochastic algorithms under verifiable conditions

TL;DR

This paper establishes a new almost sure convergence theorem for randomly truncated stochastic algorithms, relaxing classical noise conditions and providing verifiable criteria for convergence.

Contribution

It introduces a novel convergence theorem that extends existing results by removing traditional noise assumptions in stochastic algorithms.

Findings

Proves almost sure convergence under relaxed conditions

Extends classical results on stochastic algorithm convergence

Provides easily verifiable criteria for convergence

Abstract

We study the almost sure convergence of randomly truncated stochastic algorithms. We present a new convergence theorem which extends the already known results by making vanish the classical condition on the noise terms. The aim of this work is to prove an almost sure convergence result of randomly truncated stochastic algorithms under easily verifiable conditions