Private communication via zero-private-capacity quantum channels

Technical Summary: Private Communication via Zero-Private-Capacity Quantum Channels

Problem Statement
The paper addresses a longstanding open problem in quantum information theory: whether two quantum channels, each possessing zero private capacity individually, can jointly enable private communication when used together. Private communication requires reliable transmission to a receiver while ensuring secrecy from the environment (eavesdropper). While "superactivation" of quantum capacity (where two zero-capacity channels yield positive capacity) is known, and non-additivity of private capacity has been demonstrated, the specific case of superactivating private capacity using two channels that individually have strictly zero private capacity remained unresolved. Classical memoryless wiretap channels cannot exhibit this phenomenon, as their capacity is additive.

Methodology
The authors construct a specific counterexample involving two independent, finite-dimensional, memoryless quantum channels:

1. Channel $\mathcal{N}$ (The "Four-Level" Channel):

   • Structure: A channel acting on a 4-dimensional Hilbert space ($C^4$). Its noise model consists of a diagonal operator and three transition operators (and their reverses) mixed with an identity branch.
   • Zero Capacity Mechanism: The authors prove $\mathcal{N}$ has zero private capacity by demonstrating "transpose-antidegradability." They construct a completely positive trace-preserving (CPTP) map (a simulator) that allows the environment to reproduce the global transpose of the receiver's output. Since the Holevo information is invariant under transposition and the environment can simulate the receiver's statistics via a CPTP map, the environment's information is always at least as great as the receiver's, yielding zero private capacity for any blocklength.

2. Channel $\mathcal{E}_{2,p}$ (The "Helper" Channel):

   • Structure: A qubit erasure channel with erasure probability $p$ (where $1/2 \le p < 1$).
   • Zero Capacity Mechanism: For $p \ge 1/2$, the environment can simulate the receiver by retaining the qubit with probability $(1-p)/p$ and erasing otherwise. Thus, it also has zero private capacity.

3. Joint Encoding and Decoding Scheme:

   • Encoding: Alice encodes information into a superposition of a "background" state and a "signal" state. The background is a mixture of two orthogonal vectors, while the signal is a third orthogonal vector. Crucially, the encoding is designed such that all seven non-identity noise operators of $\mathcal{N}$ map the encoding space into its orthogonal complement.
   • Detection: The receiver performs a fixed binary measurement (a projective measurement onto the signal vector) on the joint output of the two channels.
   • Information Dynamics:
     • Receiver Gain: If the signal is present and the erasure channel does not erase the qubit, the receiver detects the signal with certainty. The mutual information gain scales linearly with the signal strength parameter $t$.
     • Environmental Leakage: The environment's Holevo information scales quadratically with $t$. This is proven by bounding the environmental states using a "classical coin" lemma, showing that the signal state is dominated by the background state in a specific operator sense ($\sigma_1 \preceq c \sigma_0$).

Key Results

• Superactivation Theorem: The paper proves that while $P(\mathcal{N}) = 0$ and $P(\mathcal{E}_{2,p}) = 0$ for $1/2 \le p \le 1$, the joint private capacity is strictly positive:
  $$P(\mathcal{E}_{2,p} \otimes \mathcal{N}) \ge \frac{6 \ln 2}{49(27 + 169p)(1 - p)^2} > 0$$
• Achievable Rate: At an erasure probability of $p = 1/2$, the achievable rate exceeds $1.903 \times 10^{-4}$ private bits per product use.
• Operational Constraints: The positive rate is achieved using a fixed joint measurement on each product output pair and classical wiretap coding across multiple uses. The authors prove that decoders restricted to Positive Partial Transpose (PPT) effects (including separable measurements and local operations with classical communication) cannot achieve a positive rate, highlighting the necessity of entangled measurements across the channel factors.
• Formal Verification: The core inequalities and the existence of the transpose simulator have been formally verified using the Lean 4 proof assistant and the Mathlib library, ensuring the mathematical rigor of the capacity bounds and the simulator construction.

Significance and Claims
The paper claims to resolve the private-capacity superactivation problem by providing the first explicit example where two channels with zero private capacity jointly yield a positive private capacity. This demonstrates that a channel's private capacity alone is insufficient to determine its utility for secure communication; the interaction between channels can unlock security properties that are absent individually.

The work distinguishes itself from previous non-additivity results by showing that the phenomenon is not merely a case of "small capacity + small capacity = large capacity," but rather "zero capacity + zero capacity > 0." The authors note that this result relies on the specific structural asymmetry where the receiver can detect a signal event that is absent in the background noise, while the environment's leakage remains suppressed.

The paper concludes that this result establishes a separation between the regularized "less-noisy" order (where the environment dominates the receiver at every blocklength) and the ability to activate private capacity, suggesting that information domination at the blocklength level does not guarantee stability under tensor products. The authors explicitly state that the initial activating example was identified through interactions with large language models, and the proof strategies were explored via these interactions, though the authors retain full responsibility for the correctness and formalization of the final results.