Performance of simulated annealing in p-spin glasses

TL;DR

This paper investigates the effectiveness of simulated annealing in dense 3-body spin glass models through numerical simulations, comparing results with various theoretical predictions to identify which best matches observed data.

Contribution

The study provides the first detailed numerical comparison showing that isocomplexity and following state theories align well with simulation data, while threshold states do not.

Findings

Isocomplexity and following state theories agree with numerical results

Threshold states can be excluded based on simulation data

Simulated annealing performance aligns with specific theoretical models

Abstract

We perform careful numerical simulations of slow Monte-Carlo annealings in the dense 3-body spin glass model and compare with the predictions from different theories: thresholds states, isocomplexity, following state. We conclude that while isocomplexity and following state both provide excellent agreement the numerical data, the influence of threshold states -- that is still the most commonly considered theory -- can be excluded from our data.