Optimal Combinatorial Neural Codes With Matched Metric δ<sub>r</sub>: Characterization and Constructions

Aixian Zhang,Xiaoyan Jing,Keqin Feng
DOI: https://doi.org/10.1109/tit.2023.3266010
IF: 2.5
2023-01-01
IEEE Transactions on Information Theory
Abstract:Based on theoretical neuroscience, G. Cotardo and A. Ravagnani (2022) introduced a class of asymmetric binary codes called combinatorial neural codes (CN codes for short), with a “matched metric” <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$\delta _{r}$ </tex-math></inline-formula> called asymmetric discrepancy, instead of the Hamming distance <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$d_{H}$ </tex-math></inline-formula> for usual error-correcting codes. They also presented the Hamming, Singleton and Plotkin bounds for CN codes with respect to <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$\delta _{r}$ </tex-math></inline-formula> and asked how to construct CN codes <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">${\mathcal C}$ </tex-math></inline-formula> with large size <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$| {\mathcal C}|$ </tex-math></inline-formula> and minimum <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$\delta _{r}({\mathcal C})$ </tex-math></inline-formula> . In this paper, we first show that a binary code <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">${\mathcal C}$ </tex-math></inline-formula> reaches one of the above bounds for <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$\delta _{r}({\mathcal C})$ </tex-math></inline-formula> if and only if <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">${\mathcal C}$ </tex-math></inline-formula> reaches the corresponding bounds for <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$d_{H}$ </tex-math></inline-formula> and <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$r$ </tex-math></inline-formula> is sufficiently close to 1. This means that all optimal CN codes come from the usual optimal codes. Then, we present several constructions of CN codes with good and flexible parameters <inline-formula xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink"> <tex-math notation="LaTeX">$(n,K, \delta _{r}({\mathcal C}))$ </tex-math></inline-formula> by using bent functions.
What problem does this paper attempt to address?