Noomen.ctg: UPDATE

Discussion of anything and everything relating to chess playing software and machines.

Moderators: hgm, Harvey Williamson, bob

Forum rules
This textbox is used to restore diagrams posted with the [d] tag before the upgrade.
Jeroen
Posts: 501
Joined: Wed Mar 08, 2006 8:49 pm

Re: Noomen.ctg: UPDATE

Post by Jeroen » Sun Aug 21, 2016 1:43 pm

Here is the link from Ed's website:

http://rebel13.nl/download/books.html

Jeroen
Posts: 501
Joined: Wed Mar 08, 2006 8:49 pm

Re: Noomen.ctg: UPDATE

Post by Jeroen » Sun Aug 21, 2016 1:46 pm

Thanks. I am running a test against Cerebellum as well, same result: 3-2 to Cerebellum, with many draws.

In any case, this looks like a very good test candidate. So I am going to forward it to my testers and take a good look at it. I like new challenges :-).

Jeroen
Posts: 501
Joined: Wed Mar 08, 2006 8:49 pm

Re: Noomen.ctg: UPDATE

Post by Jeroen » Sun Aug 21, 2016 1:51 pm

Hi Harvey,

It is funny that people like Shrapnel evaluate a book after just two games :-). In any case, I am happy he showed me a good idea! IMO it would be a good test candidate for your book as well.

Wishing you well,

Jeroen

peter
Posts: 1775
Joined: Sat Feb 16, 2008 6:38 am
Full name: Peter Martan

Re: Noomen.ctg: UPDATE

Post by peter » Sun Aug 21, 2016 3:20 pm

Jeroen wrote: In any case, this looks like a very good test candidate. So I am going to forward it to my testers and take a good look at it. I like new challenges :-).
Sounds great, thanks again :!:
Peter.

Hurnavich
Posts: 718
Joined: Wed Jun 19, 2013 6:14 pm

Re: Noomen.ctg: UPDATE

Post by Hurnavich » Sun Aug 21, 2016 6:27 pm

Hi,

Many thanks for this
"May your next game be your best"

User avatar
Harvey Williamson
Posts: 1765
Joined: Sun May 25, 2008 9:12 pm
Location: Media City, UK
Contact:

Re: Noomen.ctg: UPDATE

Post by Harvey Williamson » Sun Aug 21, 2016 9:45 pm

Jeroen wrote:Hi Harvey,

It is funny that people like Shrapnel evaluate a book after just two games :-). In any case, I am happy he showed me a good idea! IMO it would be a good test candidate for your book as well.

Wishing you well,

Jeroen
Sounds like a good 3-way book test if I get the time to run it.

Gusev
Posts: 1475
Joined: Mon Jan 28, 2013 1:51 pm

Re: Noomen.ctg: UPDATE

Post by Gusev » Mon Aug 22, 2016 4:52 pm

Try the newest AsmFish with Noomen against BrainFish with Cerebellum.
Jeroen wrote:Thanks. I am running a test against Cerebellum as well, same result: 3-2 to Cerebellum, with many draws.

In any case, this looks like a very good test candidate. So I am going to forward it to my testers and take a good look at it. I like new challenges :-).

Ferdy
Posts: 4067
Joined: Sun Aug 10, 2008 1:15 pm
Location: Philippines

Re: Noomen.ctg: UPDATE

Post by Ferdy » Mon Aug 22, 2016 5:50 pm

I continue to run some tests, in the match-1, it was 50 games, but later matches I use just 20 games.

In Match-1, noomen is using up to move 60, but from match-2 to 4, increase it to 100.

Before running the match-2, I let noomen book to learn the wins and loses from match-1 games.

Before running the match-3, I let noomen book to learn the wins and loses from match-2 games.

Before running the match-4, I let noomen book to learn the wins and loses from match-3 games.

Cerebellum release 12 is used for match-1 to 3, and cerebellum release 13 is used in match-4.

The learn from games features in Fritz has an effect in Match-3, when noomen won by one game.
In match-4 the new cerebellum (release 13) regains the advantage again.

Before running match-5, I will first let noomen learn the wins and loses from match-4 games.

Here are the results of matches.

Code: Select all

Blitz 3m+2s, Match-1                                      
1   Brainfish 190816 64 POPCNT    +7  +3/=45/-2 51.00%   25.5/50
2   Stockfish 190816 64 POPCNT    -7  +2/=45/-3 49.00%   24.5/50

Blitz 3m+2s, Match-2                                      
1   Brainfish 190816 64 POPCNT   +53  +4/=15/-1 57.50%   11.5/20
2   Stockfish 190816 64 POPCNT   -53  +1/=15/-4 42.50%    8.5/20

Blitz 3m+2s, Match-3                                      
1   Stockfish 190816 64 POPCNT   +17  +1/=19/-0 52.50%   10.5/20
2   Brainfish 190816 64 POPCNT   -17  +0/=19/-1 47.50%    9.5/20

Blitz 3m+2s, Match-4                                      
1   Brainfish 190816 64 POPCNT   +35  +2/=18/-0 55.00%   11.0/20
2   Stockfish 190816 64 POPCNT   -35  +0/=18/-2 45.00%    9.0/20

Ferdy
Posts: 4067
Joined: Sun Aug 10, 2008 1:15 pm
Location: Philippines

Re: Noomen.ctg: UPDATE

Post by Ferdy » Wed Aug 24, 2016 8:19 am

Match continues, Match-5.

Sf book: noomen with learn from win and loss games from match-4.
Bf book: cerebellum release 13.

Match-5 start thinking data comparison, only wins are shown.

Code: Select all

rd 6, black Stockfish, thinkStartsFmvn 11, lastFmvn 72, eco B18, res 1-0
rd 6, white Brainfish, thinkStartsFmvn 14, lastFmvn 72, eco B18, res 1-0
rd 12, white Brainfish, thinkStartsFmvn 13, lastFmvn 56, eco B80, res 1-0
rd 12, black Stockfish, thinkStartsFmvn 15, lastFmvn 56, eco B80, res 1-0
rd 18, black Stockfish, thinkStartsFmvn 9, lastFmvn 56, eco C54, res 1-0
rd 18, white Brainfish, thinkStartsFmvn 13, lastFmvn 56, eco C54, res 1-0
rd 20, black Stockfish, thinkStartsFmvn 11, lastFmvn 53, eco C54, res 1-0
rd 20, white Brainfish, thinkStartsFmvn 22, lastFmvn 53, eco C54, res 1-0
Cerebellum dominates here with advantage to book exit especially on game 20.

Head to head:

Code: Select all

Blitz 3m+2s, Match-5                  
1   Brainfish 190816 64 POPCNT   +70  +4/=16/-0 60.00%   12.0/20
2   Stockfish 190816 64 POPCNT   -70  +0/=16/-4 40.00%    8.0/20
Download games for m2 to m5.
http://www.mediafire.com/download/cmi46 ... 5-games.7z

fantasmadel50
Posts: 112
Joined: Thu Apr 30, 2015 5:36 pm

Re: Noomen.ctg: UPDATE

Post by fantasmadel50 » Mon Aug 29, 2016 9:23 pm

Buenas alguien podria aclarar por favor esta duda referente a la optimización del libro, aquí en el post el señor Jeroen Noomen , pone optimización de 3 a 100 movidas. Y en la pagina REBEL 13, se muestra en la gráfica otra cosa(3 a 60), CUAL ES LA CORRECTA. Muchas gracias. Good someone could please clarify this doubt concerning the optimization of the book, here in the post Mr. Jeroen Noomen, puts optimization 3-100 moved. And in the REBEL 13 page, it is shown in the graph otherwise (3-60), which one is correct. Thank you very much.

Post Reply