Author: Paul Jacobean Sacral
Date: 14:38:25 12/02/05
Go up one level in this thread
On December 02, 2005 at 17:13:39, chandler yergin wrote:
>On December 02, 2005 at 15:39:26, Paul Jacobean Sacral wrote:
>
>>The Nalimov tablebases do not require an active "CDrom Fritz" engine. The code
>>to access the Nalimovs is included in the engines themselves which are capable
>>to access it during the search, each, and in the Fritz GUI which can use them
>>for tbs. positions. What's required is that the tbs. related options of the GUI
>>and in the engine options are set, and the tbs. path(s) where the Nalimov tbs.
>>are stored on the hd. See F3 and in tools/options/tablebases.
>>
>>Do you understand that Thompson and Nalimov tables are different formats which
>>cannot be used in the same way? It really seems as if you are confusing them.
>>The Thompsons have indeed been distributed on 3 and later on 4 CD ROMs. I know
>>that they require specific engines to use them in CB. programs.
>
>No they don't. Fritz 5.32 accesses them all the time.
>You just show the correct Path.
Ok here's my LAST reply for you.
First, you were talking of active-cdrom-engines-blah-blah. Now, after I was
writing that the Thompson tables indeed required special access engines in CB.
progs, you suddenly reply "no they don't."
You don't understand what is meant by access by the engines. Engines with
Nalimov access use the tbs. in positions where the number of pieces is for
example 8 or 10 (or even more) already, because they reach tbs. positions in the
search tree due to captures which occur in search, reducing the number of
pieces.
This is not possible with the Thompson tables. Of course tbs. are always
accessed when you have a tbs. positions on the board! But this is not what's
meant by tbs. access by engines, in general.
Example (just a randomly chosen endgame position), see the tb= numbers:
102023: Botvinnik,M - Szabo,L, EU-chT Hamburg 1965
[D]8/1R6/p2p4/k3b3/1r6/3K4/4N3/8 w - - 0 1
Analysis by Hiarcs 9:
61.Ra7
-+ (-3.05) Depth: 3/10 00:00:00
61.Rc7 Kb6 62.Rd7 a5 63.Nc3
-+ (-2.90) Depth: 3/10 00:00:00
-+ (-2.91) Depth: 5/14 00:00:00 6kN
61.Rd7 Ka4 62.Nc3+ Kb3 63.Ne4 Rd4+ 64.Ke3 a5
-+ (-2.90) Depth: 5/14 00:00:00 8kN
-+ (-2.99) Depth: 6/16 00:00:00 14kN
61.Rc7 Ka4 62.Kc2 Rb2+ 63.Kd3 Rb3+ 64.Kc2 Rh3 65.Nf4 d5 66.Nxh3 Bxc7
-+ (-2.98) Depth: 6/16 00:00:00 19kN
-+ (-3.00) Depth: 9/21 00:00:01 342kN, tb=55
61.Ra7 Kb6 62.Ra8 Kb7 63.Rd8 a5
-+ (-2.99) Depth: 9/21 00:00:01 429kN, tb=88
-+ (-2.99) Depth: 9/21 00:00:01 438kN, tb=88
61.Rf7 Kb5 62.Nc3+ Kc6 63.Rc7+ Kxc7 64.Nd5+
-+ (-2.98) Depth: 9/21 00:00:01 574kN, tb=128
-+ (-3.01) Depth: 10/23 00:00:03 1092kN, tb=178
61.Rc7 Kb6 62.Rc8 a5 63.Rb8+ Kc5 64.Rc8+ Kd5 65.Nc3+ Ke6 66.Ne2 Rb3+ 67.Kc2 a4
68.Ra8 Ra3 69.Nc1 Kd5 70.Nd3
-+ (-3.00) Depth: 10/23 00:00:03 1213kN, tb=228
-+ (-3.17) Depth: 15/33 00:01:57 42639kN, tb=21440
--------
Count the pieces in the position...
> But they cannot
>>be accessed by engines during the search!
>Of course they can I use them all the time.
Then try the position above with a tbs.-using engine, post the results, and we
will see if it contains any tbs. access numbers.
> Btw. some of the tables which were
>>distributed, had errors.
>
>No!
>Ask Thompson
Yes, some had errors. Guess who my info source was, for that? :-))
Yours truly Paul J. Sacral
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.