[Haskell-cafe] Memoized game-tree evaluation

[Felipe Lessa]

On Sat, Jul 25, 2009 at 10:08:20AM -0400, Matthew wrote:
> I'm working on defining a data type for representing game-trees
> for chess or other two-person board games. My plan is to
> memoize the evaluations of a game-state at each depth so that
> values don't need to be recomputed for future moves.

My suggestion is that it probably isn't worth it.  You'll consume
so much memory that your performance will actually get a lot
worse than if you recomputed everything everytime.  Well, at
least for chess this is true.

--
Felipe.