The present invention provides improved algorithms existing satisfiability by changing the representation of the problem that exploits the structure of natural problems. It specifically considers the common algorithms for satisfiability, including WSAT and RELSAT. Each of these algorithms involves running, at least once, of a specific calculation, or 'sub-search', in which the search algorithm in the original theory, fundamental terms satisfactory to any of the numerical properties . In the case of realistic problems, the invention that the time spent in sub-search is negligible compared to the investment represented by the calculation of the algorithm. Indeed, the invention shows the sub-mark in terms of S (C, P, us), in which case the fundamental embodiments of literals u C have not recovered in P s for literal checked for the values assigned to P. this representation allows the execution of an intelligent search to resolve sub-search problems in terms of s (C, P, u, s). The intelligent sub-search proceeds by assignment of truth values, to atoms, so as to remove sets of links universally quantified variables within a quantified clausal constraint. These links will also eliminate themselves because they can not meet a specific proposal. Moreover, it is possible to make a recovery from poor choices on the occasion of a bond research leading to variables within quantified clauses. Generally, the intelligent subsystem search reduces the verification time O problems (D? U) O (DαU¿) for all α∫1.
展开▼
