Combinatorial Optimization Design Strategies

General strategies for designing algorithms that choose the best solution from a very large set of discrete candidates. Branch and bound enumerates candidate solutions systematically as a search tree and discards whole branches whose bounds show that they cannot contain the optimal solution, while a greedy algorithm builds a solution by making the locally best choice at each step and never reconsidering it. An algorithm belongs here when its cited article presents a design paradigm or strategy for combinatorial or discrete optimization rather than a method for one named problem. It excludes metaheuristics that borrow ideas from nature, and approximation algorithms with proven quality guarantees, which have their own groups.