Section: New Results

Mixing myopic fast algorithms and asymptotically optimal algorithms

We made several works based on combining in sequential decision making:

• a fast algorithm providing quickly good heuristic results;

• an asymptotically optimal, too slow for real size problems.

Results are published in [31] and [38] , outperforming the state of the art for MineSweeper in reasonable time; an application to energy has been done, and a new one is under work (see Section 4.1 ). We believe that this diea of combining fast approximate solutions and slow asymptotically optimal algorithms is a key for improving the state of the art in high dimensional combinatorial planning and that our results on MineSweeper and moderate size energy problem are a solid first step in this direction.