A New Heuristic Algorithm

How to find a reasonable, universal heuristic algorithm with small bifurcation rate has always been a problem deeply explored by the artificial intelligence community. They use several common techniques:

The cost of solving some problems is usually assessable, and the cost of solving the whole problem is usually reasonable. For example, for a jigsaw puzzle of 10, the cost of solving the problem should be similar to the cost of moving the square of 1 to 5 back to the correct position. Usually the problem solver will first set up a pattern database to store the cost of some problems to evaluate the problems. Solving the problem that is easy to approximate can usually be used to evaluate the original problem reasonably. For example, the Manhattan distance is a simple version of the n-puzzle problem, because we assume that we can move a square to the position we want independently, and we don't consider moving to other squares for the time being. Give us a set of reasonable heuristic functions h 1(n), h2(n), ..., hi(n), and the function H (n) = Max {H 1 (n), H2 (n), ..., Hi (n)} can predict these functions. A program, prolim r, written by A.E. Prieditis in 1993, uses these technologies and can automatically generate a heuristic algorithm for the problem. The heuristic algorithm generated by proquir for 8-puzzle is better than any previous one! It also found the first useful heuristic program to solve the magic square.