arXiv:2609.18222
Unbounded Holevo additivity gaps in finite dimensions
Jinzhao Wang
Abstract
We establish unbounded two-use Holevo additivity gaps in finite dimensions. For each sufficiently large fixed integer $K$ and all sufficiently large $n$, we construct channels $T_n$ with output dimension $K^n$, input dimension $\exp(Θ_K(n^2))$, and \[ χ(T_n^{\otimes2})-2χ(T_n) \ge n\left[\frac{\log_2K}{K}-2\log_2(1+9/K)\right]-O(1/n). \] The gap is linear in output qubits, with an explicit quadratic input-qubit cost and an explicit threshold on $n$. We also obtain channels whose single-use Holevo quantity tends to zero while their two-use Holevo information, and hence classical capacity, diverges. The channels arise from structured tensor products of the complementary mixed-unitary channels used in Collins's free-probabilistic proof. We establish minimum-output-entropy gaps by combining Collins--Youn's product-group Haagerup inequality with Bordenave--Collins's quantitative strong-convergence estimates. We also provide a Lean certificate for our proofs.
AI Summary
The paper asks whether the two-use Holevo additivity gap $\Delta_\chi(T)=\chi(T^{\otimes2})-2\chi(T)$ can grow without bound for finite-dimensional channels, rather than remain bounded as in previous von Neumann entropy constructions. For fixed sufficiently large $K$ and sufficiently large $n$, it constructs channels with output dimension $K^n$ and input dimension $2N^nK^{2n}$, where $N=\lceil e^{b_Kn}\rceil$ and $b_K=40(7\ln K+2)$. They satisfy $\Delta_\chi(T)\ge n\delta_K-O(1/n)$, with $\delta_K=(\log_2K)/K-2\log_2(1+9/K)$. The sufficient condition $\ln K>18$ makes this coefficient positive. Thus the gap is linear in output qubits, with quadratic input-qubit cost. A second regime gives $\chi(T)\to0$ while $\chi(T^{\otimes2})/2\to\infty$, hence diverging classical capacity. The construction uses tensor blocks of complementary mixed-unitary channels. A product-group Haagerup inequality bounds output purity uniformly over inputs entangled across each block; strong-convergence estimates transfer the bound to finite matrices. Bell inputs to conjugate blocks provide the joint-output entropy upper bound. An input switch and Weyl-covariant extension convert entropy deficits into Holevo gaps. The quantitative dimension bound combines an observable net, short-word embeddings into two generators, and polynomial linearization with Haar-matrix estimates. A Lean certificate is supplied. The stated limitations include extremely small guaranteed gap per output qubit, very large sufficient dimensions and thresholds, and a probabilistic construction without efficient methods for finding or verifying suitable unitaries. Linear input-qubit cost remains open; additivity of classical capacity between distinct channels is not addressed.
GPT-6 Astra (OpenAI) assisted with proof development and checking, literature searches and reference verification, and the drafting and revision of this manuscript.
Written by gpt-6.1-sol from the paper's full text. Automated: it can be wrong.
- Comments
- 22+8pp; v2:added a short discussion of concurrent works; corrected estimates in Appendix B; added a Lean certificate: https://github.com/JWang226/Holevo-Additivity-Gap