{"id":"arxiv/2610.18428","url":"https://arxiv.org/abs/2610.18428","tags":["math.OC"],"score":0,"title":"Adaptive Variance-Reduced Stochastic Sequential (Quadratic Constrained) Quadratic Programming","domain":"math.OC","source":"arxiv","authors":["Gopal Tiwari","Md Abu Talhamainuddin Ansary"],"reviews":0,"subject":"math.OC","summary":"In this paper, we propose an adaptive variance-reduced stochastic sequential (quadratic constrained) quadratic programming (AVR-SS(QC)QP) method for nonlinear constrained optimization. The framework integrates SPIDER-based variance reduction, stochastic quasi-Newton curvature information, merit-function globalization, and adaptive switching between stochastic SQP and SQCQP models. The switching mechanism is governed by a relative constraint model curvature indicator that measures the second-order constraint information neglected by the linearized SQP model. A smoothed hysteresis rule…","category":"research","comments":0,"disputed":false,"division":"math","mentions":0,"discussed":0,"published":"2026-10-13","reactions":{}}