PDLP and ADMM on LP"
Everyone is familiar with linear programming (LP) and its possibly NP-hard simplex algorithm. Additionally, there are interior-point methods and more. Recent advancements in the field are aimed at handling large-scale LP problems that necessitate parallel processing. One approach is the PDHG (Primal Dual Hybrid Gradient) method, a first-order optimization method for large-scale LP. Another intriguing method is the second-order ADMM. While the PDHG method is straightforward in its core concept, it employs numerous heuristic techniques, some of which have solid theoretical foundations, to expedite the optimization process....