A Unique Inverse Decomposition of Positive Definite Matrices under Linear Constraints

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Dolinsky, Yan, Zuk, Or
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866911399538589696
author Dolinsky, Yan
Zuk, Or
author_facet Dolinsky, Yan
Zuk, Or
contents We study a nonlinear decomposition of a positive definite matrix into two components: the inverse of another positive definite matrix and a symmetric matrix constrained to lie in a prescribed linear subspace. Equivalently, the inverse component is required to belong to the orthogonal complement of that subspace with respect to the trace inner product. Under a sharp nondegeneracy condition on the subspace, we show that every positive definite matrix admits a \emph{unique} decomposition of this form. This decomposition admits a variational characterization as the unique minimizer of a strictly convex log-determinant optimization problem, which in turn yields a natural dual formulation that can be efficiently exploited computationally. We derive several properties, including the stability of the decomposition. We further develop feasibility-preserving Newton-type algorithms with provable convergence guarantees and analyze their per-iteration complexity in terms of algebraic properties of the decomposed matrix and the underlying subspace. Finally, we show that the proposed decomposition arises naturally in exponential utility maximization, a central problem in mathematical finance.
format Preprint
id arxiv_https___arxiv_org_abs_2601_18662
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle A Unique Inverse Decomposition of Positive Definite Matrices under Linear Constraints
Dolinsky, Yan
Zuk, Or
Optimization and Control
Numerical Analysis
15A39, 15B48, 90C25 (primary), 91G10 (secondary)
We study a nonlinear decomposition of a positive definite matrix into two components: the inverse of another positive definite matrix and a symmetric matrix constrained to lie in a prescribed linear subspace. Equivalently, the inverse component is required to belong to the orthogonal complement of that subspace with respect to the trace inner product. Under a sharp nondegeneracy condition on the subspace, we show that every positive definite matrix admits a \emph{unique} decomposition of this form. This decomposition admits a variational characterization as the unique minimizer of a strictly convex log-determinant optimization problem, which in turn yields a natural dual formulation that can be efficiently exploited computationally. We derive several properties, including the stability of the decomposition. We further develop feasibility-preserving Newton-type algorithms with provable convergence guarantees and analyze their per-iteration complexity in terms of algebraic properties of the decomposed matrix and the underlying subspace. Finally, we show that the proposed decomposition arises naturally in exponential utility maximization, a central problem in mathematical finance.
title A Unique Inverse Decomposition of Positive Definite Matrices under Linear Constraints
topic Optimization and Control
Numerical Analysis
15A39, 15B48, 90C25 (primary), 91G10 (secondary)
url https://arxiv.org/abs/2601.18662