For development/debugging, it would be useful to compute some stats on local search operators. Typically how many moves of each type are tested and how many are applied (maybe with associated gain). I don't think we have a sound way to decide how many of the tried moves are actually valid because most of them are usually discarded without even checking validity, based on the potential gain (or a lower bound of the gain).
For development/debugging, it would be useful to compute some stats on local search operators. Typically how many moves of each type are tested and how many are applied (maybe with associated gain). I don't think we have a sound way to decide how many of the tried moves are actually valid because most of them are usually discarded without even checking validity, based on the potential gain (or a lower bound of the gain).