Hill search results start from the beginning with the initial guess of the solution, and then iteratively make local changes until a solution is found or the heuristic is stuck at a local maximum. There are many ways to avoid getting stuck at local maxima, such as performing many searches at the same time or probabilistic choice of successor state, etc. In many cases, mountain climbing algorithms will quickly converge on the correct answer. However, none of these approaches guarantees an optimal solution.
, , , . , . , .
, , . Branch-and-bound , .
, !