3 papers
math.OC2026
Inexact PDHG for Large-Scale Semidefinite Programming
Karthik Prakhya, Zimeng Wang
First-order methods, such as the primal-dual hybrid gradient (PDHG) method, have recently been explored for large-scale semidefinite programming (SDP). However, exact projections c…
math.OC2026
Universal Adaptive Proximal Gradient Methods via Gradient Mapping Accumulation
Zimeng Wang, Alp Yurtsever
We propose an adaptive proximal gradient method for minimizing the sum of two functions, where one is a simple convex function, and the other belongs to one of the three classes: n…
math.OC2026
Generalized Stochastic Gradient Descent with Momentum Methods for Smooth Optimization
Zimeng Wang, Alp Yurtsever
Stochastic gradient descent with momentum (SGDM) methods have become fundamental optimization tools in machine learning, combining the computational efficiency of stochastic gradie…