Enhanced partial expansion A

摘要: When solving instances of problem domains that feature a large branching factor, A* may generate number nodes whose cost is greater than the optimal solution. We designate such as surplus. Generating surplus and adding them to OPEN list dominate both time memory search. A recently introduced variant called Partial Expansion (PEA*) deals with aspect this problem. expanding node n, PEA* generates all its children puts into only f = f(n). n reinserted in f-cost best discarded child. This guarantees are not inserted OPEN. In paper, we present novel Enhanced (EPEA*) advances idea address aspect. Given priori domain-and heuristic-specific knowledge, EPEA* Although always applicable or practical, study several variants EPEA*, which make it heuristics. In particular, ideas IDA* where pattern databases traditionally used. Experimental studies show significant improvements run-time performance for standard benchmark applications. provide theoretical facilitate an understanding new algorithm.