arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 0905.4405math.CO
收藏

拟阵多面体:算法、理论与应用

Matroid Polytopes: Algorithms, Theory, and Applications

David C. Haws

摘要:

本论文提出了三个不同的主题都与拟阵多面体的新结果。首先研究拟阵多面体、独立拟阵多面体和多拟阵的Ehrhart多项式的性质。我们证明了,对于固定的秩,他们的Ehrhart多项式在多项式时间内是可计算的。证明依赖于这些多面体的几何形状,以及一个新的精制分析托德多项式的评价。其次,我们讨论了关于拟阵多面体的代数组合学的理论结果。讨论了拟阵多面体的Ehrhart多项式的h^*-向量和系数的两种形式,并从理论和计算上证明了它们的正确性.我们还探讨了怀特猜想的一个变种,其中指出,每一个拟阵多面体有一个定期的幺模三角剖分。我们提供了广泛的计算证据支持这一新的猜想,并提出了一个组合条件的单形足够的单模性。最后,最近的非线性和多准则拟阵优化的算法理论的工作的动机,我们已经开发了算法和approxistics针对这些困难的问题的大型实例的实际解决方案。我们的方法主要是利用拟阵多面体中固有的局部邻接结构来旋转到可行解,这可能是也可能不是最优的。我们还提出了一种改进的广度优先搜索启发式算法,使用邻接枚举的一个子集的可行解。我们提出了其他的物理学,并提供支持这些新技术的计算证据。我们实现了我们所有的算法在软件包MOCHA(拟阵优化组合启发式和算法)。

英文摘要:

This dissertation presents new results on three different themes all related to matroid polytopes. First we investigate properties of Ehrhart polynomials of matroid polytopes, independence matroid polytopes, and polymatroids. We prove that for fixed rank their Ehrhart polynomials are computable in polynomial time. The proof relies on the geometry of these polytopes as well as a new refined analysis of the evaluation of Todd polynomials. Second, we discuss theoretical results regarding the algebraic combinatorics of matroid polytopes. We discuss two conjectures about the h^*-vector and coefficients of Ehrhart polynomials of matroid polytopes and provide theoretical and computational evidence for their validity. We also explore a variant of White's conjecture which states that every matroid polytope has a regular unimodular triangulation. We provide extensive computational evidence supporting this new conjecture and propose a combinatorial condition on simplices sufficient for unimodularity. Finally, motivated by recent work on algorithmic theory for non-linear and multicriteria matroid optimization, we have developed algorithms and heuristics aimed at practical solutions of large instances of these difficult problems. Our methods primarily use the local adjacency structure inherent in matroid polytopes to pivot to feasible solutions which may or may not be optimal. We also present a modified breadth-first-search heuristic that uses adjacency to enumerate a subset of feasible solutions. We present other heuristics, and provide computational evidence supporting these new techniques. We implemented all of our algorithms in the software package MOCHA (Matroids Optimization Combinatorial Heuristics and Algorithms).

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑