Winboard Forum

Hi all,

I use an extension [ETD] which resembles the BM ext., in that it uses the null-move refutation .

It also resembles mate threat too, maybe more so.

But 1. I do not use the BM ext. because I still dont quite get it, and 2. I am not even sure what it is supposed to 'see' exactly

So, I am simply going to explain my extension, in hopes that someone will be able to point out the differences, or just help me understand it:

Basically, if not-moving loses a hung piece or is lost to a smaller valued piece, extend.

I limit this to PV nodes, and once used in a line, it is discontinued so that only one can be triggered along any variation.

Basically, it serves the same purpose as the recapture extension: to detect when something may be 'hidden' by eating up 2 ply.

IE, I see position has a problem, so I threaten your knight with my pawn, you move knight, and then the problem gets pushed over the horizon!

I figure this happens at least as often as the recapture trick does, if not more!

What does the BM ext see? What's the point of it? Detecting 'persistent' threats?

Is this right? :

If null-move refutation is the same now as it was at ply-2, extend [whole node? ie, ++depth]

Also, the mate threat extension:

Do most people trigger this upon *any* -mate score or just mate-in-one ?


-Thanks

-David

Hi David,

Botvinnik-Markoff Extensions are triggered, if a threat at ply N already occured two plies before, at ply N-2. Eg. two plies before - after null-move - there was a capture knight takes queen. In the meantime in the path to this current node, the queen moved to a safe square, but opponent threatens the queen again with the knight. After null the "same" threat nxq becomes possible, which triggers the extension. You are free to define what the "same" threat is, and there is much to try...

With fractional extensions, i also add some "centy-plies" for single threats or some more for mate threats and queen attacks.

Gerd

Thanks Gerd.