Author: KarinsDad
Date: 17:09:46 05/18/99
Go up one level in this thread
On May 18, 1999 at 18:15:23, Dann Corbit wrote: >On May 18, 1999 at 00:13:18, KarinsDad wrote: >[snip] >>I did not know why you posted your original message on this, so I did not know >>why you wanted to use it. >I am going to make a public interface to C.A.P. data. A sort of "permanent >hash" table. That way, you can provide a position and I will return an eval and >preferred move. The data, code, and specification will be publicly available. > >I wanted some discussion of the principles to ensure that nothing was overlooked >and that the functions would be useful to as large an audience as possible. Well, you could have people enter EPD positions and if you wanted to save them internally, you could use the new format that I specified. Unless you were willing to create and give out source for the format I specified elsewhere in this thread, it just wouldn't be worth it (it's too complex unless one person writes and debugs the code and then hands it out). However, my structure is 20 bytes for an extremely high percentage of positions and whenever someone sent you one that fails in my structure, you could store it in 40 bytes (although they would have to be doing this on purpose). And 20 bytes would save you 16% over storing each position internally in the EPD 24 bytes. Not a great savings, but some. But, all in all, I do not see a great need for my new format for what you are trying to do since there are probably better ways for you to store it internally. And externally, the 24 byte EPD format is probably sufficient (and it can handle all cases). KarinsDad :) PS. BTW, I am going to post a normal game (Shirov vs. Kasparov from today) later along with the number of bits required per move with my format. I haven't checked it all, but by move 13, the most bits that it required was about 131 (and dropping fast). It will be extremely difficult to find a position from a real game that uses over 155 bits and 161 bits seems virtually impossible in normal chess with this format. IMHO.
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.