Legality Check on TT move

[Rein Halbersma]

Someone has an idea of how often happen the hash collisions? I did some tries and I'm not able to catch an illegal move coming from hash. Anyway I keep checking them for legality.

Here's some relevant math behind the question: http://en.wikipedia.org/wiki/Birthday_attack

This always comes up, but it is the wrong approach. The hash table does not remember all previous birthdays, only those that are still present.

"The wrong approach" seems a bit harsh. I like the sound of "conservative estimate" better.

I don't like the sound of "wrong" either, but I've seen the birthday paradox come up so often in this context...

I may have once come up with it myself as well in the days of rgcc. At least I remember working it out on a whiteboard.

Here you can find some good analysis. "Memory versus Search in Games": https://project.dke.maastrichtuniversit ... thesis.pdf

As usual, it's the wrong approach!!

It appears that Zobrist himself introduced this "wrong approach" http://research.cs.wisc.edu/techreports/1970/TR88.pdf

[syzygy]

It appears that Zobrist himself introduced this "wrong approach" http://research.cs.wisc.edu/techreports/1970/TR88.pdf

I have no idea what he is doing there. It is not the birthday paradox approach, but it is certainly wrong!! Still a very nice paper