An optical capacity theorem needs the energy constraint in its headline

An October 15 preprint claims the classical capacity of every single-mode bosonic Gaussian channel at every input energy, with independent Gaussian signal ensembles attaining the optimum. The authors also distinguish strict photon-number cutoffs from mean-energy constraints when analyzing communication at nonzero error. The proof has not been independently audited here. This is a theoretical capacity claim, not a measured link rate, a quantum-capacity formula or a secure-key guarantee. The exact resource constraint matters because different restrictions on rare high-energy signals can produce different conclusions about communication above capacity.^[1]

Classical communication through a quantum channel sends ordinary messages using physical quantum states. The receiver tries to identify which message was encoded. The channel can include attenuation, amplification or Gaussian noise, while the sender is limited by an energy budget. Capacity asks for the best asymptotic rate under the specified error criterion, allowing the encodings and measurements admitted by the theory. It does not necessarily describe the rate of a particular transmitter and receiver available today, which may use a much narrower set of operations.

Single mode describes the action of one channel use; it does not automatically reduce the optimization to independent uses. A sender could, in principle, correlate or entangle inputs across many uses. Determining whether that helps is an additivity question. If independent inputs attain the unrestricted optimum for a channel family, the result removes a major layer of optimization for that family. It does not establish that entanglement never helps communication in general, because other channels and other communication tasks can have different mathematical structure.

Energy can be spent shaping a signal's fluctuations or separating the states used to encode messages. Squeezing changes the noise distribution between complementary quadratures, while displacement shifts a state's mean field. Under a limited budget, spending more on one leaves less for the other. An optimum therefore reflects both the channel's noise structure and the available energy. A state that minimizes output noise in isolation need not maximize transmitted information once the cost of preparing that state is included in the same resource account.

A simple original example explains why an average cap differs from a strict one. Suppose a scheme is allowed an average energy of ten units per transmitted block. A distribution that sends a fifty-unit block in one fifth of the attempts and zero in the rest meets that average. It violates a rule forbidding any block from exceeding ten units. This invented classical energy allocation is not a capacity-achieving code. It illustrates how rare high-energy events can be allowed by one constraint and excluded by another, changing the feasible set before optimization begins.

Error criteria add another boundary. A vanishing-error capacity asks for increasingly reliable decoding as block length grows. A statement about a fixed allowed average error permits a different trade-off. A strong converse says that exceeding a specified rate drives success toward failure in a particular asymptotic sense. Whether that conclusion holds can depend on the energy restriction and on how errors are averaged. These definitions are not fine print: they determine what the theorem actually rules out and which unusual coding strategies remain admissible.

Capacity also differs from security and from the ability to preserve quantum states. A receiver can decode classical messages reliably while an eavesdropper learns them too. Quantum capacity asks whether unknown quantum information can be transmitted, which is a different objective from discriminating an encoded classical alphabet. A secret-key rate needs an adversary model and security analysis. An optical channel can therefore have several distinct operational rates, and a result for one should not be relabeled as a complete solution to every communication problem on that channel.

The proposed theorem is consequential because it addresses a broad optimization question under a clear channel family. The next step for readers is independent scrutiny of the proof and a careful translation to constrained transmitter and receiver designs. Such translation should preserve the energy budget, error criterion and allowed measurements. It can then reveal how far a practical link lies from the relevant theoretical limit. Without those definitions, a single capacity number loses the conditions that make it meaningful and can be mistaken for a performance promise the theorem never made.