一句话:研究图类上有界树深度、FO 有效广义量词消去与参数化模型检测复杂度之间的等价关系。
论文研究有限图类 K 上三个性质之间的关系:K 的树深度是否有界;一阶逻辑 FO 在 K 上是否具有有效的广义量词消去;以及 FO 在 K 上的参数化模型检测是否属于 para-AC0。作者证明由有界树深度可推出有效广义量词消去,而有效广义量词消去与参数化模型检测属于 para-AC0 等价。进一步地,若 K 对取子图封闭,则三者完全等价;但一般情况下,树深度有界并不被后两者蕴含,因此三者不等价。该结果用树深度与量词消去刻画 FO 模型检测的参数化复杂度上界。
For a class K of finite graphs we consider the following three statements. (i) K has bounded tree-depth. (ii) First-order logic FO has an effective generalized quantifier elimination on K. (iii) The parameterized model checking for FO on K is in para-AC0. We prove that (i) ⟹ (ii) and (ii) ⟺ (iii). All three statements are equivalent if K is closed under taking subgraphs, but not in general.
DOI 原文 ·
@inproceedings{paperbot4392,
title = {Tree-depth, quantifier elimination, and quantifier rank},
author = {Yijia Chen and Jörg Flum},
booktitle = {LICS},
year = {2018},
doi = {10.1145/3209108.3209160}
}