Primal Heuristics for Branch and Price: The Assets of Diving Methods
研究了在分支定价框架下构建原始解的通用启发式方法，重点探讨潜水方法及其与多样化强化策略的组合，在广义分配、下料和顶点着色问题上取得了优于专用启发式的结果。
Primal heuristics have become essential components in mixed integer programming (MIP) solvers. Extending MIP-based heuristics, our study outlines generic procedures to build primal solutions in the context of a branch-and-price approach and reports on their performance. Our heuristic decisions carry on variables of the Dantzig–Wolfe reformulation, the motivation being to take advantage of a tighter linear programming relaxation than that of the original compact formulation and to benefit from the combinatorial structure embedded in these variables. We focus on the so-called diving methods that use reoptimization after each linear programming rounding. We explore combinations with diversification-intensification paradigms such as limited discrepancy search, sub-MIP, local branching, and strong branching. The dynamic generation of variables inherent to a column generation approach requires specific adaptation of heuristic paradigms. We manage to use simple strategies to get around these technical issues. Our numerical results on generalized assignment, cutting stock, and vertex-coloring problems set new benchmarks, highlighting the performance of diving heuristics as generic procedures in a column generation context and producing better solutions than state-of-the-art specialized heuristics in some cases.