Question to Ed Schroeder

[Cardoso]

Hi,
about your technique on History Reductions:
during the search do use an aging mechanism for the entries?
If so how do you age them?
Also if after the reduced search a fail high occurs do you research again with "normal depth"?

thanks in advance,
Alvaro Cardoso

[ed]

Hi,
about your technique on History Reductions:
during the search do use an aging mechanism for the entries?
If so how do you age them?
Also if after the reduced search a fail high occurs do you research again with "normal depth"?

thanks in advance,
Alvaro Cardoso

Hi Alvaro,

The aging mechanism I use is that after a reduction the move value is reset to the threshold value, the effect is that the move won't be reduced so easily for a second/third time.

I have an option that researches a reduced move that fails high, it doesn't gave me an improvement just because of the above mentioned aging setting.

Good luck,

Ed

[Tony]

Hi Ed,

another question.

I was wondering, in your evaluation, do you use piece square tables ?

Or are you completely relying on the mobility calculation and using psq tables only for move ordering ?

Tony

[ed]

Hi Ed,

another question.

I was wondering, in your evaluation, do you use piece square tables ?

Or are you completely relying on the mobility calculation and using psq tables only for move ordering ?

Tony

Hi Tony,

I use both techniques, PS tables for evaluation (for me) is handy to solve small missing chess knowledge issues or conflicting chess knowledge without having to write the code for it in eval. An example of the latter, when there (for instance) is a bishop exchange on G3 my program tends to capture on G3 with the F2-pawn (instead of the H2-pawn) because of the opened F-file bonus for the rook on F1. A small PS-table fix solved the problem.

I have 2 sets of PS-tables, one for eval, one for move-gen. In move-gen you can do other tricks, such as pushing passed pawns or update the PS-table via the history counters.

Ed