Modular Mix-and-Match Complementation of B”{u}chi Automata (Technical Report). (arXiv:2301.01890v1 [cs.FL])
![Modular Mix-and-Match Complementation of B”{u}chi Automata (Technical Report). (arXiv:2301.01890v1 [cs.FL])](https://likflix.com/wp-content/uploads/2023/01/12046-modular-mix-and-match-complementation-of-buchi-automata-technical-report-arxiv230101890v1-cs-fl.png?v=1672983927)
Play And Watch The Video Below
[Submitted on 5 Jan 2023]
Download PDF
Abstract: Complementation of nondeterministic Büchi automata (BAs) is an critical
dilemma in automata concept with many applications in formal verification,
such as termination assessment of applications, model checking, or in final decision
treatments of some logics. We make on thoughts from a the latest function on BA
determinization by Li et al. and propose a new modular algorithm for BA
complementation. Our algorithm will allow to mix a number of BA complementation
treatments jointly, with 1 process for a subset of the BA’s strongly
connected components (SCCs). In this way, 1 can exploit the construction of
certain SCCs (such as when they are inherently weak or deterministic) and
use much more efficient specialised algorithms, regardless of the construction of the
entire BA. We give a typical framework into which partial complementation
techniques can be plugged in, and its instantiation with various algorithms.
The framework can, in normal, generate a enhance with an Emerson-Lei
acceptance issue, which can typically be a lot more compact. Utilizing the algorithm, we
ended up able to establish an exponentially better new higher certain of $O(4n)$ for
complementation of the not long ago released class of elevator automata. We
implemented the algorithm in a prototype and executed a complete established of
experiments on a significant established of benchmarks, exhibiting that our framework
complements properly the point out of the artwork and that it can provide as a basis for
long term productive BA complementation and inclusion examining algorithms.
Submission historical past
From: Ondřej Lengál [view email]
[v1]
Thu, five Jan 2023 03:35:39 UTC (2,556 KB)