Covariance Matrix Adaptation Evolution Strategy without a matrix Conference: the Genetic and Evolutionary Computation Conference; This is the authors’ version of the paper accepted at GECCO 2025.;

Jarosław Arabas, Adam Stelmaszczyk, Eryk Warchulski, Dariusz Jagodziński, Rafał Biedrzycki · 2026

Read Covariance Matrix Adaptation Evolution Strategy without a matrix Conference: the Genetic and Evolutionary Computation Conference; This is the authors’ version of the paper accepted at GECCO 2025.; by Jarosław Arabas, Adam Stelmaszczyk, Eryk Warchulski, Dariusz Jagodziński, Rafał Biedrzycki (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2601-00102.txt.

Description

Jarosław Arabas, Adam Stelmaszczyk, Eryk Warchulski, Dariusz Jagodziński, and Rafał Biedrzycki rework a well-known optimization method that adapts a covariance matrix. The method, called CMA-ES, searches by sampling from a bell-shaped distribution shaped by that matrix. Splitting the matrix into parts costs time that grows fast with dimension, which limits use on large problems. Their paper removes the matrix entirely.

An archive stores the gaps between past candidates and their midpoint, scaled by step size. New candidates are built as weighted mixes of those stored gaps. The authors prove the resulting distribution matches the standard method exactly. In tests, a small archive of recent rounds suffices, and the matrix-free version often converges faster.

Catalog record

CREATORJarosław Arabas, Adam Stelmaszczyk, Eryk Warchulski, Dariusz Jagodziński, Rafał Biedrzycki
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE39 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2601-00102

Browse the full library · About the Tower of Babel