Hopping between distant basins

TL;DR

The paper introduces BH-S, an enhanced stochastic optimization algorithm that improves basin hopping by incorporating a skipping proposal, leading to better exploration of distant basins and improved performance on benchmark problems.

Contribution

It presents the BH-S algorithm, which replaces the perturbation step in basin hopping with a skipping proposal from rare-event sampling, enabling more effective non-local exploration.

Findings

BH-S outperforms traditional basin hopping on benchmark surfaces.

The skipping proposal facilitates hopping between distant basins.

Empirical results show improved optimization performance.

Abstract

We present the Basin Hopping with Skipping (BH-S) algorithm for stochastic optimisation, which replaces the perturbation step of basin hopping (BH) with a so-called skipping proposal from the rare-event sampling literature. Empirical results on benchmark optimisation surfaces demonstrate that BH-S can improve performance relative to BH by encouraging non-local exploration, that is, by hopping between distant basins.