Experimental results on subgoal reordering

The effect on the performance of a goal-oriented theorem power is studied on subgoal re-ordering using some simple synthetic heuristics. It is shown that subgoal reordering using these simple heuristics has a considerable impact on the performance of the prover on a large set of test problems. Some heuristics even provide equally good, and often better, performance as to the hand ordering of the input clauses. The merit of the approach seems to be that the syntactic aspect of theorem proving is considered. This approach is simple in form and cheap in its evaluation, and often provides good heuristics.<
>