arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2302.05813cs.SC
收藏

Lazard 风格 CAD 与等式约束

Lazard-style CAD and Equational Constraints

James H. Davenport, Akshar S. Nair, Gregory K. Sankaran, Ali K. Uncu

AI总结:

本文研究如何将等式约束优化迁移至 Lazard 风格 CAD,分析由此重新出现的零化问题,并基于 Brown–McCallum 改进给出单约束与多约束情形的解决方案。

摘要:

McCallum 风格柱面代数分解(CAD)是对 Collins 原始版本的重大改进,并且已有许多后续进展,尤其是针对全部或部分等式约束的工作。但它存在零化(nullification)问题。最近得到论证的 Lazard 风格 CAD 没有这一问题。然而,将等式约束方面的工作迁移到 Lazard 风格时,确实会重新引入零化问题。本文基于第二作者的博士论文以及 Brown–McCallum 对 Lazard 的改进,解释了该问题及其解决方案。对于单个等式约束,我们可以在 Lazard 风格中获得与 McCallum 风格 CAD 相同的改进。此外,在 McCallum 方法会因零化而失败的情形下,我们的方法不会失败。不出所料,它所实现的改进程度不及非零化情形。我们还考虑了多个等式约束的情形。

英文摘要:

McCallum-style Cylindrical Algebra Decomposition (CAD) is a major improvement on the original Collins version, and has had many subsequent advances, notably for total or partial equational constraints. But it suffers from a problem with nullification. The recently-justified Lazard-style CAD does not have this problem. However, transporting the equational constraints work to Lazard-style does reintroduce nullification issues. This paper explains the problem, and the solutions to it, based on the second author's Ph.D. thesis and the Brown--McCallum improvement to Lazard. With a single equational constraint, we can gain the same improvements in Lazard-style as in McCallum-style CAD . Moreover, our approach does not fail where McCallum would due to nullification. Unsurprisingly, it does not achieve the same level of improvement as it does in the non-nullified cases. We also consider the case of multiple equational constraints.

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。

↑