Author: Heiner Marxen
Date: 15:28:25 12/31/00
Go up one level in this thread
On December 31, 2000 at 17:51:03, Steve Maughan wrote: >Heiner, > >>>As an example, suppose that initially Alpha = 1.00, Beta = 5.00 . From the >>>hash we find a deeper search with a lower bound of 3.00. It is possible that >>>had we not changed alpha and after the node has been searched the score for >>>this node would be +2.00 and it could have formed part of the PV. > >The key here is that the hash info pertains to a _deeper_ search. > >>I don't follow here: you just stated that the TT says "this node's value >>(for this depth) > >No not for this depth but for a deeper search. > >>is 3.00 or even higher". And we are trusting the statements >>from the TT, right? > >At a depth greater than the current search depth the value is indeed +3.00 or >higher but at the current depth it could be below this value (e.g. +2.00) and >this could still justify being part of the PV. > >>So, we expect a search here to not return anything below 3.00, especially not >>2.00 as value. > >At a 'shallow' depth it could be +2.00 > >Regards, > >Steve Ok, you are right, and I understand your point. OTOH, a general assumption underlying all this searching business is that deeper searches generally produce better (more accurate) results. I.e. if we alrady have an even more correct result than the one the normal search would yield, we tend to to happily accept that, and we do expect to improve by this. Right? Of course you could restrict TT hits to only occur if the draft exactly matches the depth. I suspect that would reduce the benefit from the TT in most cases, and quite drastically so in some cases (like the famous Fine 70). Do you really exclude TT entries with a larger draft? Heiner
This page took 0 seconds to execute
Last modified: Thu, 15 Apr 21 08:11:13 -0700
Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.