2016/05/31 by Jaehak Lee, Se-Wan Ji, Jiyong Park +1
Computer Science · Mathematics · Physics and Astronomy · #Algorithm #Artificial intelligence #Benchmark (surveying) #Computer science #Decoding methods #Encoding (memory) #Gaussian #Gaussian noise #Mathematics #Physics #Quantum Information and Cryptography #Quantum Mechanics and Applications #Quantum mechanics #Quantum optics and atomic interactions #Statistical physics #Theoretical computer science #quant-ph
paper · pdf · doi:10.1103/physreva.93.050302
published as Phys. Rev. A 93, 050302(R) (2016) · 9 pages, 6 figures, with supplemental material
openalex publication_date 2016/05/31 · arxiv created 2016/06/03 · arxiv updated 2016/06/06 · openalex created_date 2025/10/10 · openalex updated_date 2026/08/05
We establish the fundamental limit of communication capacity within Gaussian schemes under phase-insensitive Gaussian channels, which employ multimode Gaussian states for encoding and collective Gaussian operations and measurements for decoding. We prove that this Gaussian capacity is additive, i.e., its upper bound occurs with separable encoding and separable receivers so that a single-mode communication suffices to achieve the largest capacity under Gaussian schemes. This rigorously characterizes the gap between the ultimate Holevo capacity and the capacity within Gaussian communication, showing that Gaussian regime is not sufficient to achieve the Holevo bound particularly in the low-photon regime. Furthermore, the Gaussian benchmark established here can be used to critically assess the performance of non-Gaussian protocols for optical communication. We move on to identify non-Gaussian schemes to beat the Gaussian capacity and show that a non-Gaussian receiver recently implemented by Becerra et al. [F. E. Becerra et al., Nat. Photon. 7, 147 (2013)] can achieve this aim with an appropriately chosen encoding strategy.