Creative Commons Attribution 3.0 Unported license
We consider nested fixed-point expressions like μ z. ν y. μ x. f(x,y,z) evaluated over a finite lattice, and ask how many queries to a function f are needed to find the value. The previous upper bounds for a monotone function f of arity d over the lattice {0,1}ⁿ were of the order n^{𝒪(d)}, whereas a lower bound of Ω(n²/(lg n)) is known in case when at least one alternation between the least (μ) and the greatest (ν) fixed point occurs in the expression. Following a recent development for parity games, we show here that a quasi-polynomial number of queries is sufficient, namely n^{lg(d/lg n)+𝒪(1)}. The algorithm is an abstract version of several algorithms proposed recently by a number of authors, which involve (implicitly or explicitly) the structure of a universal tree. We then show a quasi-polynomial lower bound for the number of queries used by the algorithms in consideration.
@InProceedings{arnold_et_al:LIPIcs.CSL.2021.9,
author = {Arnold, Andr\'{e} and Niwi\'{n}ski, Damian and Parys, Pawe{\l}},
title = {{A Quasi-Polynomial Black-Box Algorithm for Fixed Point Evaluation}},
booktitle = {29th EACSL Annual Conference on Computer Science Logic (CSL 2021)},
pages = {9:1--9:23},
series = {Leibniz International Proceedings in Informatics (LIPIcs)},
ISBN = {978-3-95977-175-7},
ISSN = {1868-8969},
year = {2021},
volume = {183},
editor = {Baier, Christel and Goubault-Larrecq, Jean},
publisher = {Schloss Dagstuhl -- Leibniz-Zentrum f{\"u}r Informatik},
address = {Dagstuhl, Germany},
URL = {https://drops.dagstuhl.de/entities/document/10.4230/LIPIcs.CSL.2021.9},
URN = {urn:nbn:de:0030-drops-134430},
doi = {10.4230/LIPIcs.CSL.2021.9},
annote = {Keywords: Mu-calculus, Parity games, Quasi-polynomial time, Black-box algorithm}
}