A Riemannian quasi-Newton algorithm for optimization with Euclidean bounds

Mateusz BaranAGH University of Krakow30 Mickiewicz Ave., 30-059Kraków, Poland [email protected], Ronny BergmannDepartment of Mathematical Sciences,Norwegian University of Science and Technology,NO-7041 Trondheim, NorwayAGH University of Krakow30 Mickiewicz Ave., 30-059Kraków, Poland [email protected] Patryk PrzybyszAGH University of Krakow30 Mickiewicz Ave., 30-059Kraków, Poland · 2026

Read A Riemannian quasi-Newton algorithm for optimization with Euclidean bounds by Mateusz BaranAGH University of Krakow30 Mickiewicz Ave., 30-059Kraków, Poland [email protected], Ronny BergmannDepartment of Mathematical Sciences,Norwegian University of Science and Technology,NO-7041 Trondheim, NorwayAGH University of Krakow30 Mickiewicz Ave., 30-059Kraków, Poland [email protected] Patryk PrzybyszAGH University of Krakow30 Mickiewicz Ave., 30-059Kraków, Poland (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-2605-10573.txt.

Description

Mateusz Baran and two colleagues propose a Riemannian BFGS method for tuning with Euclidean bounds. The method pairs a limited-memory quasi-Newton update in the tangent space with a curved-space version of the Cauchy point plan from L-BFGS-B. That handles Euclidean bounds while using the domain's shape. Their 2026 paper gives a framework and code in the Manopt.jl library.

The setting matters for covariance matrix work with bounded variance, brain scans, and EEG signal sorting. It also covers tasks coupling manifold variables with constrained Euclidean values. Tests on problems show only minor speed loss on Euclidean problems versus L-BFGS-B, while beating older methods. On two mixed problems, the method beats methods by several orders of magnitude. The problems are capped-amplitude blind source separation and bounded-variance maximum likelihood common principal components analysis.

Catalog record

CREATORMateusz BaranAGH University of Krakow30 Mickiewicz Ave., 30-059Kraków, Poland [email protected], Ronny BergmannDepartment of Mathematical Sciences,Norwegian University of Science and Technology,NO-7041 Trondheim, NorwayAGH University of Krakow30 Mickiewicz Ave., 30-059Kraków, Poland [email protected] Patryk PrzybyszAGH University of Krakow30 Mickiewicz Ave., 30-059Kraków, Poland
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE45 KB
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2605.10573

Browse the full library · About the Tower of Babel