Finding all Solutions to a CNF. The DualSat SAT Solver E-Book

Elmar Stellnberger

• Herausgeber: GRIN Verlag
• Sprache: Englisch

Veröffentlichungsjahr: 2021

Beschreibung

Master's Thesis from the year 2020 in the subject Computer Science - Applied, grade: Sehr gut, Klagenfurt University (Institut für Angewandte Informatik), language: English, abstract: DualSAT is a performant new SAT-solver that is optimized to find and output all solutions to an input CNF usable by further processing steps that can follow. Solutions are stored as a number of disjoint solution classes that happen to be in Decomposable Negation Normal Form (DNNF). The solver implements new features like unentangled literal detection and backtracking with stack redo. Unentangled literal detection detects the remaining CNF to be in DNNF while a stack redo saves valuable work on conflict directed backjumps because much of the solver's stack content remains preserved. A combination of conflict dependent and independent heuristics ensures good results for known as well as novel and random CNFs. The dual data structures of DualSAT may also prove beneficial for new techniques with enhanced reasoning.