Full Enumeration

The idea of fully enumerating the solution space. In the case of a real-values solution space, full enumeration must be approximated by dividing each dimension into parcels. The size of such divisions must be chosen by the user.

We can generally think of “full enumeration” as being the worst case for performance. However there are a few exceptions. If the solution space is very small, it might make sense to simply fully enumerate the solution space.

A full enumeration search may be trivially performed in a “steady state” fashion, as the results from the search do not affect the search behaviour at all. Full enumeration may outperform some algorithms (such as Random Search) where the algorithm does not benefit from any form of search memory.