Evolutionary Hessian Learning: Forced Optimal Covariance Adaptive Learning (FOCAL)

TL;DR

This paper introduces FOCAL, a novel method to accurately learn the Hessian matrix in high-dimensional optimization landscapes, demonstrated on model and quantum control systems, enhancing understanding of the search landscape.

Contribution

FOCAL is a new technique that explicitly retrieves the Hessian matrix at the global basin, improving covariance learning in high-dimensional, complex landscapes.

Findings

FOCAL accurately recovers the Hessian matrix in model landscapes.

FOCAL successfully applied to quantum control systems.

The method aligns with physical knowledge of the systems.

Abstract

The Covariance Matrix Adaptation Evolution Strategy (CMA-ES) has been the most successful Evolution Strategy at exploiting covariance information; it uses a form of Principle Component Analysis which, under certain conditions, is suggested to converge to the correct covariance matrix, formulated as the inverse of the mathematically well-defined Hessian matrix. However, in practice, there exist conditions where CMA-ES converges to the global optimum (accomplishing its primary goal) while it does not learn the true covariance matrix (missing an auxiliary objective), likely due to step-size deficiency. These circumstances can involve high-dimensional landscapes with large condition numbers. This paper introduces a novel technique entitled Forced Optimal Covariance Adaptive Learning (FOCAL), with the explicit goal of determining the Hessian at the global basin of attraction. It begins by introducing theoretical foundations to the inverse relationship between the learned covariance and the Hessian matrices. FOCAL is then introduced and demonstrated to retrieve the Hessian matrix with high fidelity on both model landscapes and experimental Quantum Control systems, which are observed to possess a non-separable, non-quadratic search landscape. The recovered Hessian forms are corroborated by physical knowledge of the systems. This study constitutes an example for Natural Computing successfully serving other branches of natural sciences, and introducing at the same time a powerful generic method for any high-dimensional continuous search seeking landscape information.