NU 图的 Brightwell–Winkler 型刻画
A Brightwell-Winkler type characterisation of NU graphs
AI总结:
本文证明 H-扩展重构对所有实例均平凡当且仅当 H 为 NU 图,给出 NU 图的新刻画,并得到解空间直径界、参数化多项式最短路径算法及对最短路径重构的推广。
摘要:
2000 年,Brightwell 和 Winkler 将可拆除图刻画为这样的图 H:对于所有图 G,定义在从 G 到 H 的同态集合上的 Hom-图 Hom(G,H) 都是连通的。这表明,H-着色问题的重构版本 Recon_Hom(H)——即对于给定的 G,判断 Hom(G,H) 是否连通——是平凡的,当且仅当 H 是可拆除图。我们针对 H-扩展问题的重构版本证明了一个类似的出发点。其中 Hom(G,H;p) 是 Hom-图 Hom(G,H) 的子图,由扩展 G 的 H-预着色 p 的那些 H-着色诱导;H-扩展问题的重构版本 Recon_Ext(H) 对于给定的图 G 的 H-预着色 p,询问 Hom(G,H;p) 是否连通。我们证明,对每一种 (G,p) 的选择 Hom(G,H;p) 都连通的图 H,恰好就是 NU 图。这给出了 NU 图的一种新刻画;NU 图是一类良好的图,在 CSP 二分性的代数方法中十分重要。我们进一步给出了当 H 为 NU 图时 Hom(G,H;p) 直径的界,并证明 Hom(G,H;p) 中两个顶点之间的最短路径可在参数化多项式时间内找到。我们将结果应用于最短路径重构问题,显著推广了近期的结果。
英文摘要:
In 2000, Brightwell and Winkler characterised dismantlable graphs as the graphs $H$ for which the Hom-graph ${\rm Hom}(G,H)$, defined on the set of homomorphisms from $G$ to $H$, is connected for all graphs $G$. This shows that the reconfiguration version ${\rm Recon_{Hom}}(H)$ of the $H$-colouring problem, in which one must decide for a given $G$ whether ${\rm Hom}(G,H)$ is connected, is trivial if and only if $H$ is dismantlable. We prove a similar starting point for the reconfiguration version of the $H$-extension problem. Where ${\rm Hom}(G,H;p)$ is the subgraph of the Hom-graph ${\rm Hom}(G,H)$ induced by the $H$-colourings extending the $H$-precolouring $p$ of $G$, the reconfiguration version ${\rm Recon_{Ext}(H)}$ of the $H$-extension problem asks, for a given $H$-precolouring $p$ of a graph $G$, if ${\rm Hom}(G,H;p)$ is connected. We show that the graphs $H$ for which ${\rm Hom}(G,H;p)$ is connected for every choice of $(G,p)$ are exactly the ${\rm NU}$ graphs. This gives a new characterisation of ${\rm NU}$ graphs, a nice class of graphs that is important in the algebraic approach to the ${\rm CSP}$-dichotomy. We further give bounds on the diameter of ${\rm Hom}(G,H;p)$ for ${\rm NU}$ graphs $H$, and show that shortest path between two vertices of ${\rm Hom}(G,H;p)$ can be found in parameterised polynomial time. We apply our results to the problem of shortest path reconfiguration, significantly extending recent results.
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。