Described herein is a meta-algorithm adaptable to different types of optimization problems and different computing platforms. A problem space is (i) browsed using a heuristic that computes objectives locally and (ii) while constructing a solution, key decisions are performed globally. A simple data structure—a probabilistic cache—is used to efficiently store intermediate sub-solutions. As an example, the meta-algorithm is applied to find an algorithm for solving the graph coloring problem.
展开▼
