Skip to main content
QUICK REVIEW

[Paper Review] Algorithms for Computing the Petz-Augustin Capacity

Chun-Neng Chu, Wei-Fu Tseng|arXiv (Cornell University)|Jan 10, 2026
Wireless Communication Security TechniquesEngineering0 citations
TL;DR

The authors propose the first non-asymptotic algorithms with convergence guarantees for computing the Petz-Augustin capacity of classical-quantum (CQ) channels, via two complementary approaches for Petz-Rényi and Petz-Augustin information. They provide Hölder-smooth optimization and Blahut-Arimoto-type methods with proofs of convergence rates and a fixed-point inner solver.

ABSTRACT

We propose the first algorithms with non-asymptotic convergence guarantees for computing the Petz-Augustin capacity, which generalizes the channel capacity and characterizes the optimal error exponent in classical-quantum channel coding. This capacity can be equivalently expressed as the maximization of two generalizations of mutual information: the Petz-Rényi information and the Petz-Augustin information. To maximize the Petz-Rényi information, we show that it corresponds to a convex Hölder-smooth optimization problem, and hence the universal fast gradient method of Nesterov (2015), along with its convergence guarantees, readily applies. Regarding the maximization of the Petz-Augustin information, we adopt a two-layered approach: we show that the objective function is smooth relative to the negative Shannon entropy and can be efficiently optimized by entropic mirror descent; each iteration of entropic mirror descent requires computing the Petz-Augustin information, for which we propose a novel fixed-point algorithm and establish its contractivity with respect to the Thompson metric. Notably, this two-layered approach can be viewed as a generalization of the mirror-descent interpretation of the Blahut-Arimoto algorithm due to He et al. (2024).

Motivation & Objective

  • Generalize computation of channel capacities to the Petz-Augustin framework for CQ channels.
  • Develop non-asymptotic convergence guarantees for both Petz-Rényi and Petz-Augustin information maximizations.
  • Provide practical algorithms that extend mirror-descent interpretations of Blahut-Arimoto in the quantum setting.
  • Analyze algorithmic convergence rates and establish contractivity for the inner fixed-point solver.

Proposed method

  • Maximize Petz-Rényi information by exponentiating the objective to obtain a convex Hölder-smooth problem and solve with universal fast gradient methods.
  • Show the objective is ((1-α)/α, 1/α)-Hölder smooth on the simplex for α in [1/2,1), enabling Hölder-smooth optimization with a convergence rate of O(T^{1-1.5/α}).
  • Develop a Blahut-Arimoto-type algorithm for the Petz-Augustin capacity by nesting an inner fixed-point algorithm to compute I_α^A(p,W) and an outer entropic mirror-descent step to optimize over input distributions.
  • Prove the fixed-point inner loop is contractive in the Thompson metric, yielding linear convergence for the inner computation.
  • Establish the outer loop convergence of entropic mirror descent under relative smoothness of the negative Petz-Augustin information, achieving a O(1/T) rate.
  • Discuss implications for constrained input problems and relate the method to PDHG approaches.

Experimental results

Research questions

  • RQ1How can the Petz-Augustin capacity for CQ channels be computed with non-asymptotic convergence guarantees?
  • RQ2Can the Petz-Rényi and Petz-Augustin information maximize efficiently through distinct optimization frameworks (Hölder-smooth vs. mirror-descent)?
  • RQ3What are the convergence properties and rates of the proposed Hölder-smooth and Blahut-Arimoto-type methods in the CQ setting?
  • RQ4How can inner fixed-point computations for I_α^A(p,W) be made contractive and efficiently differentiable for outer optimization?
  • RQ5How do these algorithms extend to constrained input problems and relate to existing Arimoto-Blahut generalizations?

Key findings

  • The Hölder-smooth approach yields a convergence rate of O(T^{1-1.5/α}) for α in [1/2,1).
  • The inner fixed-point solver for I_α^A(p,W) converges linearly in the Thompson metric, enabling ε-approximate gradients and values in O(log(1/ε)) iterations.
  • The outer entropic mirror-descent method achieves a convergence rate of O(1/T) for the Petz-Augustin information optimization.
  • The Blahut-Arimoto-type algorithm and the Hölder-smooth approach provide complementary rates, with empirical results illustrating α-dependent performance trade-offs.
  • The methods extend the mirror-descent interpretation of Blahut-Arimoto to the Petz-Augustin capacity and can handle certain input-constraint scenarios.

Read this paper

Open the full text in the Nubint reader and ask about any passage as you read.

Read this paper on Nubint

This review was generated by AI from the original paper. Please check the paper before citing it.