{"id":"arxiv/2609.39240","url":"https://arxiv.org/abs/2609.39240","tags":["math.OC"],"score":0,"title":"Boundary Geometry and Global Rates of Cyclic Coordinate Descent for SVM Duals","domain":"math.OC","source":"arxiv","authors":["Egor Gladin","Eva Ibodova"],"reviews":0,"subject":"math.OC","summary":"Cyclic coordinate descent is widely used to train support vector machines, but global linear convergence alone gives limited guidance about the rate on a particular instance. We study exact cyclic coordinate minimization for box-constrained convex quadratics and develop global contraction bounds that capture both coordinate order and boundary geometry. Our convergence certificate combines two ingredients: interactions between successive coordinate updates, encoded by the triangular part of the normalized Hessian, and objective growth, captured by a matrix lower bound on the optimality gap.…","category":"research","comments":0,"disputed":false,"division":"math","mentions":0,"discussed":0,"published":"2026-09-30","reactions":{}}