Author: Heiner Marxen
Date: 16:38:49 06/28/99
Go up one level in this thread
On June 28, 1999 at 17:44:25, Dann Corbit wrote: >I have a lot of chess data, with a ludicrous amount coming soon. Here is my >problem. I want to read the data into a tree structure (don't worry about >space, I have big mainframes here) and then refine it. But the problem is that >the data is not all of the same quality and some of the data will be missing. >For instance, we have about 600K EPD rows at 12 minutes of PII 300 MHz >equivalent time. We also have about 10K at 8 hours of time. But we have tens >of millions at 2 seconds. So, I don't think that a simple mini-max is >appropriate because the data is of vastly different quality. > >Imagin, I am at some node. Now it has been analyzed to 16 plies and the >children (of which there are 40 possible) have only 14 entries. Of these two >are at 15 plies and 5 at 12 plies and 7 at 9 plies. How can I update the >parent from this? IMO, not at all. The 16-ply result is already build from 15-ply sub-results. Even those cannot improve the 16-ply result. You can look at this in a similar way as when you find a hash hit: smaller depth than wanted is just not usable. (Exception: mate scores) > It seems to me that only the 15 ply points can be used as a >reference because the root has already seen deeper than all the others. > >Does anyone have algorithm suggestions? Is there a modification to minimax >that will render it useful? You could operate with value pairs (depth, score), where the depth is the minimum of the subdepth plus one, and scores without enough depth are just not counted in. Mate scores should always be counted in. This is a somewhat pessimistic way to do it, but I fail to see, how this can be substantially improved on, without degrading the value of its result. Of course, you could have some optimistic result, also, but I'm not sure how to interpret such a result: what does it say? Another trick may be to argue with bounds instead of exact values, as is done with TTs, also. Now you could e.g. have a node with score 1.00 for depth=16, and a bound >= 0.89 for depth=17, and probably more bounds info.
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.