Choosing restart strategy at partial knowledge of process statistics

TL;DR

This paper develops practical criteria for deciding when to restart a stochastic process with incomplete knowledge of its statistics, aiming to improve performance in mean completion time and success probability.

Contribution

It introduces new criteria for effective restart strategies that rely on partial statistical information, bridging a gap in existing theoretical understanding.

Findings

Proposed criteria for restart effectiveness based on easily estimated statistics

Criteria applicable to non-instantaneous restart protocols

Enhances decision-making with partial process knowledge

Abstract

Optimization of a random processes by restart is a subject of active theoretical research in statistical physics and has long found practical application in computer science. Meanwhile, one of the key issues remains largely unsolved: when should we restart a process whose detailed statistics are unknown to ensure that our intervention will improve performance? Addressing this query here we propose several constructive criteria for the effectiveness of various protocols of non-instantaneous restart in the mean completion time problem and in the success probability problem. Being expressed in terms of a small number of easily estimated statistical characteristics of the original process, these criteria allow informed restart decision based on partial information.