Talk:Optimization

From DDL Wiki

--J 11:13, 24 September 2008 (EST) I suggest that we include the brief table here for overview and then move the more detailed text about solvers in each of the problem classes to the appropriate page for each (e.g.: NLP, convex MINLP, etc). Each such page should include a heading for 1) formulation, 2) methods, and 3) software. Aida's text can fall under software. We should also add a column for heuristic/deterministic (is there a more precise way to say whether the algorithm can guarantee global solutions in the class), a column for black-box function compatibility, and maybe a column for "large scale" capabilities. One idea is, rather than X's to represent classes of problems the algorithm is designed for, we could use L, G, or S to designate that the algorithm finds local, global, or satisficing solutions. We may do something similar with constraint handling. We could also convert the environment category into columns like the problem class category.