{"id":"arxiv/2603.12838","url":"https://arxiv.org/abs/2603.12838","tags":["math.OC","cs.DC","stat.ML"],"score":0,"title":"A New Kernel Regularity Condition for Distributed Mirror Descent: Broader Coverage and Simpler Analysis","domain":"math.OC","source":"arxiv","authors":["Junwen Qiu","Ziyang Zeng","Leilei Mei","Junyu Zhang"],"reviews":0,"subject":"math.OC","summary":"Existing convergence analyses of distributed optimization methods in non-Euclidean geometries typically rely on kernel assumptions: (i) global Lipschitz smoothness and (ii) bi-convexity of the associated Bregman divergence function. Unfortunately, these conditions are violated by nearly all kernels used in practice, leaving a huge theory-practice gap. This work closes this gap by developing a unified analytical tool that guarantees convergence under mild conditions. Specifically, we introduce Hessian relative uniform continuity (HRUC), a regularity satisfied by nearly all standard kernels.…","category":"research","comments":0,"disputed":false,"division":"math","mentions":0,"discussed":0,"published":"2026-03-13","reactions":{}}