OKlibrary  0.2.1.6
AlgorithmSelection.hpp File Reference

Tools for choosing good local search algorithms from a collection More...

Go to the source code of this file.


Detailed Description

Tools for choosing good local search algorithms from a collection

Especially we consider running Ubcsat, while tools are often written in R.

Todo:
Connections
Todo:
Selecting the best algorithm
  • According to my experience (OK), for every (narrow) example class (like Ramsey formulas) there is one best algorithm in the Ubcsat selection, which is also relatively easy recognisable.
  • First we need to develop a method, and then we should try to automatise it.
  • The starting point is to find out on a single instance which algorithm performs best.
  • Then we consider three single instances, an "easy" one, a "middle easy" and a "harder" one, and see whether the predictions scale, where at least for the evaluation of our method we consider all algorithms.

Definition in file AlgorithmSelection.hpp.