{"id":"arxiv/2610.18785","url":"https://arxiv.org/abs/2610.18785","tags":["quant-ph","cs.IT"],"score":0,"title":"Relative-Entropy Geometry and Convergence in Nonconvex Wiretap Optimization","domain":"quant-ph","source":"arxiv","authors":["Farzin Salek","Masahito Hayashi"],"reviews":0,"subject":"quant-ph","summary":"We develop and analyze an entropic descent method for a nonconvex optimization problem on probability simplices arising in classical and quantum secure communication. The optimization variables describe a randomized encoder, allowing its input distribution and stochastic preprocessing to be optimized jointly. An exact relative-entropy identity yields a channel-independent majorization bound and an explicit mirror-descent iteration with a fixed unit step size. We establish quantitative descent, convergence of objective values, and stationarity of accumulation points on their support faces for…","category":"research","comments":0,"disputed":false,"division":"physics","mentions":0,"discussed":0,"published":"2026-10-13","reactions":{}}