{"ID":2893173,"CreatedAt":"2026-06-01T04:54:23.091178241Z","UpdatedAt":"2026-06-01T04:54:23.091178241Z","DeletedAt":null,"paper_url":"https://arxiv.org/abs/2507.14051","arxiv_id":"2507.14051","title":"cuPDLPx: A Further Enhanced GPU-Based First-Order Solver for Linear Programming","abstract":"We introduce cuPDLPx, a further enhanced GPU-based first-order solver for linear programming. Building on the recently developed restarted Halpern PDHG for LP, cuPDLPx incorporates a number of new techniques, including a new restart criterion and a PID-controlled primal weight update. These improvements are carefully tailored for GPU architectures and deliver substantial computational gains. Across benchmark datasets, cuPDLPx achieves 2.5x-5x speedups on MIPLIB LP relaxations and 3x-6.8x on Mittelmann's benchmark set, with particularly strong improvements in high-accuracy and presolve-enabled settings. The solver is publicly available at https://github.com/MIT-Lu-Lab/cuPDLPx.","short_abstract":"We introduce cuPDLPx, a further enhanced GPU-based first-order solver for linear programming. Building on the recently developed restarted Halpern PDHG for LP, cuPDLPx incorporates a number of new techniques, including a new restart criterion and a PID-controlled primal weight update. These improvements are carefully t...","url_abs":"https://arxiv.org/abs/2507.14051","url_pdf":"https://arxiv.org/pdf/2507.14051v4","authors":"[\"Haihao Lu\",\"Zedong Peng\",\"Jinwen Yang\"]","published":"2025-07-18T16:17:20Z","proceeding":"math.OC","tasks":"[\"math.OC\"]","methods":"[]","has_code":false,"code_links":[{"ID":612037,"CreatedAt":"2026-06-01T04:54:23.091178241Z","UpdatedAt":"2026-06-01T04:54:23.091178241Z","DeletedAt":null,"paper_id":2893173,"paper_url":"https://arxiv.org/abs/2507.14051","paper_title":"cuPDLPx: A Further Enhanced GPU-Based First-Order Solver for Linear Programming","repo_url":"https://github.com/MIT-Lu-Lab/cuPDLPx","is_official":false,"mentioned_in_paper":false,"mentioned_in_github":true,"github_stars":0}]}