Author: Dann Corbit
Date: 16:19:19 12/17/04
Go up one level in this thread
On December 17, 2004 at 13:02:01, Mathieu Pagé wrote: >On December 16, 2004 at 21:30:41, Les Fernandez wrote: > >>On December 16, 2004 at 17:05:48, Nino wrote: >> >>>On December 16, 2004 at 16:38:07, F. Huber wrote: >>> >>>>On December 16, 2004 at 16:22:11, Nino wrote: >>>> >>>>>I came across your post not to long ago about some kind of special utility for >>>>>mates. HAve not seen anything about it nor did you say much about it. Can you >>>>>please tell me more about it and why was the results of those test positions >>>>>never reported. I would be interested in knowing more about this utility and >>>>>the results of the positions. Does anyone else know about this? >>>>> >>>>>Kindly, >>>>> >>>>>DM >>>> >>>>Have a look here: >>>> >>>>http://www.talkchess.com/forums/1/message.html?400706 >>> >>>I saw the post but was hoping to have Dann or anyone else tell me a little more >>>about it. Does anyone know anymore about it? Perhaps a buit more detail about >>>what it is and what it can do. >>> >>>Thanks >>>> >>>>Regards, >>>>Franz. >> >>Hello Nino, >> >>Well like Franz and Dann mentioned before I am just cleaning up some of the code >>and will make it available soon. I dont spend much time on it since I am busy >>during the day but do work on it at night. Basically what this utility does is >>find mate positions much faster (I believe) then what is done normally. It >>takes advantage of a special binary format which can store an entire EPD string >>along with side to move, all the pieces (b/w), their locations on the board, >>castling rights, enpassant, ce and pv. Whats nice about this special binary >>format may be explained better with an example: Lets say you have the following >>epd string (not a real position, just for example): >> >>3rn3/2p4k/8/8/3Q4/R7/Nn3P2/7K w - - pv Qd1; ce 32740; >> >>As an average I can store this position using about 9 bits. This is >>accomplished by taking advantage of the special binary format I have adopted. >>Whats nice about this as Dann has mentioned is that this utility can compose new >>mates also. Another interesting thing to note is that however many bits it >>takes me to store the above ficticious position in my special binary format I >>can store up to an additional 108 positions using not even 1 extra bit. So for >>example if it takes me 182 bits to store the above "1" position I can add an >>additional 108 positions to it and it would still only require 182 bits. FYI the >>least amount of bits I need for K K is 110 bits and the worse case scenarion >>(all pieces on board) would need 380 bits. These numbers may seem high but I >>assure you that an approximation as to how many bits are needed using my binary >>format would be closer to for K K ~ 5 bits and full board ~ 19 bits. As >>mentioned above it not only stores the pieces but everything else in the >>standard EPD string. >> >>HTH, >> >>Les > > >Hi Les, > >Did you mean bytes when you say bits ? It's hard to believe for me that you >encoded a position in about 19 bits (or 2.375 bytes). This encoding will only work for checkmates. He means bits. He gets an economy because similar positions can be used to compress it.
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.