Order of search techniques

Discussion of chess software programming and technical issues.

Moderator: Ras

Sven
Posts: 4052
Joined: Thu May 15, 2008 9:57 pm
Location: Berlin, Germany
Full name: Sven Schüle

Re: Order of search techniques

Post by Sven »

jwes wrote:
Sven Schüle wrote:
AlvaroBegue wrote:
Sven Schüle wrote:
theturk1234 wrote:Then PVS if we're in a PV node.
Regular AB call otherwise.
I hope you do PVS in the way it is intended:
- full window for subtree of first move,
- zero window for all other subtrees, researching with full window in case of fail-high
If we are using PVS, can we define a PV node as a node where beta>alpha+1?

I realize this question is enough of a tangent that perhaps it should be in its own thread, but I expect a single answer will do, and perhaps it will be informative to other people reading this thread.
Yes, we can. But I would write "beta != alpha+1", even though it is practically the same.
If beta < alpha + 1, wouldn't you want to throw an exception or something?
Perhaps, but this should be irrelevant for the "PV node" condition. I simply don't want to think more than necessary. "beta == alpha+1" is the zero-window condition, and "not zero-window" means "PV node", and that's all I need. In practice also "beta >= alpha+1" is true (and can be asserted - at least at the beginning of a search function!) but why should we be too specific about it, where "!=" already implies "PV node"?