← All papers
First page of Unbounded Holevo additivity gaps in finite dimensions

Unbounded Holevo additivity gaps in finite dimensions

Jinzhao Wang

quant-ph Sep 16, 2026 · v2 cs.IT math-ph
The paper's results and proofs were autoformalized in Lean by an AI system and checked with the Comparator tool, and are released as a Lean certificate.
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.

Known Holevo additivity violations for quantum channels stayed bounded as channel dimensions grew. The open question was whether the two-use Holevo gap can be unbounded in finite dimensions, and how it scales with input and output qubits.

Channels are built from structured tensor products of the complementary mixed-unitary channels in Collins's free-probabilistic proof. Minimum-output-entropy gaps come from combining the Collins–Youn product-group Haagerup inequality with the Bordenave–Collins quantitative strong-convergence estimates. These gaps are then converted into Holevo gaps. The proofs were autoformalized in Lean and checked against Comparator.

For large fixed K and large n, the channels T_n have output dimension K^n and input dimension exp(Θ_K(n^2)). Their two-use Holevo gap grows linearly in n. There are also channels whose single-use Holevo quantity tends to zero while their classical capacity diverges.