at what point is material dangerously low for zugzwang?

[AndrewShort]

I have null move implemented, and I'm just amazed at the speed increase. In my code, if the side to move is "low on material", I don't try the null move, since the side to move might be in zugzwang. Presently, "low on material" means the side to move has 1500 or less in material (equivalent of 15 pawns or less).

In many games my engine plays, you can see very rapid responses, then, as soon as the tree or game reaches a player who is low on material, the engine slows down considerably, and I feel like I should set 1500 to a lower number, but I don't know enough about zugzwang to know what a safe number is.

What do you guys do? 1,000? 800 ? Are 99% of zugzwang positions with, say, less than 1,000 in material ?

[pedrox]

Most programs prove null move when the side on move has a piece (without count pawns).

In the endgame is true that you can verify whether zugzwang, I make that verification with less than 1300 (without count pawns). One thing is certain, with queens has zugzwang.

[michiguel]

Most programs prove null move when the side on move has a piece (without count pawns).

In the endgame is true that you can verify whether zugzwang, I make that verification with less than 1300 (without count pawns). One thing is certain, with queens has zugzwang.

In this case, I believe that a Knight should not count as a piece. In fact, endgames with knights and pawns are very prone to zugzwangs. I believe that the number of pawns should be irrelevant too. We should look at the presence of either bishops, rooks or queens. Those are the ones that avoid zugzwangs, because they can defend the same square moving back and forth. Knights and pawns can't do that.

Miguel

[Harald Johnsen]

VERIFIED NULL-MOVE PRUNING can help and is nearly free.

HJ.