|
|
HyperLTL Model Checking
Alexaj, Ondrej ; Strejček, Jan (oponent) ; Lengál, Ondřej (vedoucí práce)
HyperLTL model checking is an approach to verifying a system against a given hyperproperty, which is able to relate multiple executions of a system. The algorithmic approach based on automata which relies on standard -automata operations is well established. The aim of this work is to outperform the complete state-of-the-art HyperLTL model checker AutoHyper by employing more efficient partial automata operations, in particular complementation and inclusion. The implementation of HyperLTL model checking in a novel modular-based complementation tool Kofola resulted in a significant enhancement in performance compared to the reference tool. Finally, our approach to language inclusion checking shows a notable improvement in terms of the generated state space. As a commonly used automata operation, it could potentially contribute to the advancement of other areas of verification.
|
|
|
Abstraction in Automata Algorithms
Kocourek, Tomáš ; Lengál, Ondřej (oponent) ; Holík, Lukáš (vedoucí práce)
The goal of this thesis is to implement and experimentally compare antichain-based algorithms with and without abstraction, which decide the emptiness of alternating finite automata. The author also proposes his own algorithms using abstraction and comes up with a few optimizations of existing abstract algorithms. The thesis introduces the theoretical background of studied algorithms and describes efficient ways to implement data structures which are used by these algorithms. The experimental evaluation over random automata shows that the algorithms without abstraction give us better results in general because they do not perform costly evaluation of closed set intersection and complementation. However, in case of automata with high transition density, the algorithms without abstraction tend to decelerate, while the abstract ones accelerate.
|
|
|
Abstraction in Automata Algorithms
Kocourek, Tomáš ; Lengál, Ondřej (oponent) ; Holík, Lukáš (vedoucí práce)
The goal of this thesis is to implement and experimentally compare antichain-based algorithms with and without abstraction, which decide the emptiness of alternating finite automata. The author also proposes his own algorithms using abstraction and comes up with a few optimizations of existing abstract algorithms. The thesis introduces the theoretical background of studied algorithms and describes efficient ways to implement data structures which are used by these algorithms. The experimental evaluation over random automata shows that the algorithms without abstraction give us better results in general because they do not perform costly evaluation of closed set intersection and complementation. However, in case of automata with high transition density, the algorithms without abstraction tend to decelerate, while the abstract ones accelerate.
|
host ::
RSS.