On Lagrange multipliers in convex entropy minimization

TL;DR

This paper demonstrates that solutions derived via Lagrange multipliers in convex entropy minimization problems are indeed optimal, based on a new characterization of optimality.

Contribution

It provides a theoretical validation that Lagrange multiplier solutions are optimal in convex entropy minimization, clarifying their correctness.

Findings

Lagrange multiplier solutions are optimal for convex entropy minimization.

A new characterization of optimality supports the validity of these solutions.

The approach confirms the theoretical soundness of using Lagrange multipliers.

Abstract

Based on a characterization of the optimality of a feasible solution of a convex entropy minimization problem, one shows that the feasible solutions obtained using formally the Lagrange multipliers method are optimal.