Abstract
We prove that Gaussian thermal input states minimize the output von Neumann entropy of the one-mode Gaussian quantum-limited attenuator for a fixed input entropy. This result addresses a long-standing conjecture in quantum information theory, establishing that the Gaussian thermal state—the quantum analogue of thermal noise—is the unique minimizer. The proof utilizes a majorization result to restrict the search to passive states and introduces a new isoperimetric inequality relating the derivative of output entropy to the entropy itself. Beyond quantum optics, these findings settle a parallel question in classical probability regarding the "thinning" operation and provide the necessary converse proof for the capacity region of the Gaussian degraded quantum broadcast channel.
Introduction
In classical information theory, the effect of signal attenuation on entropy is straightforward: scaling a real-valued signal by a factor results in a predictable change in Shannon entropy. However, in the quantum regime, where signals are composed of discrete photons, this simplicity vanishes. The one-mode Gaussian quantum-limited attenuator, which models photon loss in media such as optical fibers or satellite links, does not treat all input states equally. The output von Neumann entropy is not merely a function of the input entropy; it depends fundamentally on the specific "shape" of the quantum state.
For decades, researchers conjectured that Gaussian thermal states are the most efficient at preserving low entropy through such channels. While previous attempts using the quantum Entropy Power Inequality (qEPI) provided some bounds, they were not "tight"—meaning they failed to reach the specific values produced by Gaussian states. This paper provides the rigorous proof that Gaussian states are indeed the unique minimizers, filling a critical gap in our understanding of quantum communication capacities.
Methodology: Bridging Quantum and Classical Domains
The proof employs a multi-step analytical framework that reduces a complex quantum optimization problem into a manageable discrete mathematical proof.
1. Majorization and Passive States
The authors first leverage majorization theory to simplify the search space. They demonstrate that the entropy minimizer must be found among "passive states"—states that are diagonal in the Fock (photon number) basis with eigenvalues arranged in non-increasing order. This reduction is crucial as it transforms a problem over all possible quantum density operators into a problem over discrete probability distributions.
2. The Isoperimetric Inequality
The core innovation of the paper is the derivation of a new quantum isoperimetric inequality. This inequality relates the rate of change (the derivative) of the output entropy with respect to the attenuation parameter to the entropy itself. By proving this relationship at an infinitesimal level, the authors can "integrate" the result to cover any degree of attenuation (λ). This shows that if Gaussian states are optimal for a marginal amount of loss, they remain optimal for any level of loss.
3. Classical Thinning and KKT Conditions
The paper establishes a formal bridge between quantum mechanics and classical probability. The action of a quantum attenuator on Fock states is shown to be mathematically identical to "thinning"—a classical process where elements of a set are independently removed with a certain probability. The authors treat the minimization as a functional optimization under entropy constraints, applying Karush-Kuhn-Tucker (KKT) conditions to prove that the geometric distribution (the photon-number distribution of a Gaussian thermal state) is the unique solution.
Key Results and Implications
The primary result is formalized in Theorem 1, which establishes the inequality:
$S(\Phi_\lambda(\rho)) \ge g(\lambda g^{-1}(S(\rho)))$
This confirms that for any input state $\rho$ with a given entropy, the output entropy of the attenuator $\Phi_\lambda$ is bounded below by the output entropy of a Gaussian thermal state with the same initial entropy.
Impact on Communication Theory
This theorem has immediate consequences for the Gaussian degraded quantum broadcast channel. It provides the "converse" proof—the mathematical upper bound—for its capacity region. It confirms that Gaussian encodings are the optimal strategy for transmitting information to multiple receivers through lossy media.
Advancements in Probability
The work also settles a question in classical probability theory. It proves that geometric distributions minimize the output Shannon entropy of the thinning operation for a fixed input entropy. Notably, this proof does not require the "ultra log-concavity" (ULC) assumptions that previous researchers had to rely on, providing a more general and robust result.
Conclusion and Future Work
While this paper settles the one-mode case, it highlights the challenges remaining for multimode systems. The majorization techniques used here do not scale easily to multiple modes, suggesting that a different mathematical strategy may be required to characterize complex "triple trade-off" scenarios involving the simultaneous transmission of classical and quantum information. Nonetheless, this work provides the fundamental building block for the multimode generalization and the ultimate characterization of Gaussian quantum channels.
Summary updated