Finding Minima in Complex Landscapes: Annealed, Greedy and Reluctant Algorithms

TL;DR

This paper introduces a new class of dynamical algorithms combining annealing with a balanced greedy-reluctant strategy to effectively find the deepest minima in complex, multivalleyed landscapes, demonstrated on spin-glass models.

Contribution

It presents a novel algorithmic approach that integrates annealing with a balanced greedy-reluctant strategy for optimization in complex systems.

Findings

Successfully finds deep minima in spin-glass models

Outperforms traditional methods in complex landscape optimization

Demonstrates effectiveness on the Sherrington-Kirkpatrick model

Abstract

We consider optimization problems for complex systems in which the cost function has a multivalleyed landscape. We introduce a new class of dynamical algorithms which, using a suitable annealing procedure coupled with a balanced greedy-reluctant strategy drive the systems towards the deepest minimum of the cost function. Results are presented for the Sherrington-Kirkpatrick model of spin-glasses.