Yes, you are right, iterate.c contains aspiration window code for a long time already, sorry. I was reading the code too quickly.bob wrote:Iterative deepening is exactly what goes on in iterate.c, and yes, I have used aspiration windows in every version of Crafty that was distributed. As has PVS always been there. That is what I call "iterative deepening" which was as defined by Slate in chess 4.x...Sven Schüle wrote:He is not talking about "internal iterative deepening", and nor am I. What you call "traditional definition" (of iterative deepening) seems to be different from what other people think about it. Talking about Crafty, we are at "iterate.c" level (but with aspiration window - I did not look into recent Crafty code to check whether that is present there but older Crafty did not have it as far as I know).bob wrote:When someone says "iterative deepening" I use the traditional definition, which is what is done at the root. "internal iterative deepening" is a different topic and there you are correct. Only question is, which is he talking about???Sven Schüle wrote:[...]
I think the problem comes from misunderstanding the different roles of the ID loop and the full-width search processing the root node.
Still my previous comments are valid. The ID loop calls the main search function with "root_alpha, root_beta, depth" and not with "-root_beta, -root_alpha, depth - 1".