Page 1 of 2

Paper "Singular Extensions" by Anantharaman, Campb

Posted: Mon Sep 15, 2014 8:22 pm
by RolandoFurioso
Hi,

I'm looking for the classical publication "Singular Extensions: Adding Selectivity to Brute-Force Searching" by Anantharaman, Campbell, and Hsu (1988).

Unfortunately, I don't have easy access to the Proceedings / Journal in which this seminal paper originally appeared, and I have not been able to locate it in the web either.

Is there anyone out there who could point me to an online resource where I could obtain a legal (prefereably free) copy of this paper?

Thanks a lot!
Roland

Re: Paper "Singular Extensions" by Anantharaman, C

Posted: Mon Sep 15, 2014 9:23 pm
by noctiferus
I believe I found it:

http://www.sciencedirect.com/science/ar ... 0290900739

not so cheap :( :(

Re: Paper "Singular Extensions" by Anantharaman, C

Posted: Tue Sep 16, 2014 1:09 pm
by Henk
noctiferus wrote:I believe I found it:

http://www.sciencedirect.com/science/ar ... 0290900739

not so cheap :( :(
Downloading source code of Stockfish is cheaper. It implements singular extensions, but perhaps no one understands why it works (in Stockfish) or it is kept a secret.

[Singular extensions in my chess program never worked or I did not detect any improvement]

Re: Paper "Singular Extensions" by Anantharaman, C

Posted: Tue Sep 16, 2014 1:41 pm
by Daniel Anulliero
Henk wrote:
Downloading source code of Stockfish is cheaper. It implements singular extensions, but perhaps no one understands why it works (in Stockfish) or it is kept a secret.

[Singular extensions in my chess program never worked or I did not detect any improvement]
Do you have any working heuristics in your search ? ;-)

Re: Paper "Singular Extensions" by Anantharaman, C

Posted: Tue Sep 16, 2014 1:46 pm
by Henk
Daniel Anulliero wrote:
Henk wrote:
Downloading source code of Stockfish is cheaper. It implements singular extensions, but perhaps no one understands why it works (in Stockfish) or it is kept a secret.

[Singular extensions in my chess program never worked or I did not detect any improvement]
Do you have any working heuristics in your search ? ;-)
No. But principal variation search does work for Skipper, but I'm not sure about that too. Heuristics and other statistical data (/lies) are not for me.

No one can say that they are well tested for long time controls if elo gain is small.

Re: Paper "Singular Extensions" by Anantharaman, C

Posted: Tue Sep 16, 2014 1:54 pm
by Daniel Anulliero
RolandoFurioso wrote:Hi,

I'm looking for the classical publication "Singular Extensions: Adding Selectivity to Brute-Force Searching" by Anantharaman, Campbell, and Hsu (1988).

Unfortunately, I don't have easy access to the Proceedings / Journal in which this seminal paper originally appeared, and I have not been able to locate it in the web either.

Is there anyone out there who could point me to an online resource where I could obtain a legal (prefereably free) copy of this paper?

Thanks a lot!
Roland
Hi rolando
It seems this article was pubished inthe ICGA journal volume 11 n°4
You may ask a scan to mr Hyatt , he said hd have all the volumes
Bests!

Re: Paper "Singular Extensions" by Anantharaman, C

Posted: Tue Sep 16, 2014 2:08 pm
by RolandoFurioso
Henk wrote:
noctiferus wrote:I believe I found it:

http://www.sciencedirect.com/science/ar ... 0290900739

not so cheap :( :(
Downloading source code of Stockfish is cheaper. It implements singular extensions, but perhaps no one understands why it works (in Stockfish) or it is kept a secret.

[Singular extensions in my chess program never worked or I did not detect any improvement]
Hi Henk,

afaik, Stockfish restricts Singular Extensions to TT candidate moves the value of which is marked as lower bound.

Starting, however, with the seminal publication on SEs seems to be worthwhile anyway. Unfortunately, while there exists a bunch of excellent books with reprints of historical papers on computer chess, no one covers the publication by Anantharaman, Campbell, and Hsu.

I'm fully aware that it will be probably quite a challenge to get an overall positive contribution out of a Singular Extension implementation; of course, there are interdependencies with the other extension and reduction strategies implemented by the particular engine, and there are quite a few parameters to experiment with.

The only way to find out whether it works in one's particular case / engine (in this case: Fischerle) is to experiment, experiment, experiment .... .

I have just started with my experiments, implementing the SE strategy as employed by Stockfish and playing with some parameters, and I'm very keen to see whether I'll be able to distill some SE settings which yield an overall positive contribution.

But I admit that I'm quite a bit sceptical though ... .

Roland

Re: Paper "Singular Extensions" by Anantharaman, C

Posted: Tue Sep 16, 2014 2:13 pm
by RolandoFurioso
Daniel Anulliero wrote:
RolandoFurioso wrote:Hi,

I'm looking for the classical publication "Singular Extensions: Adding Selectivity to Brute-Force Searching" by Anantharaman, Campbell, and Hsu (1988).

Unfortunately, I don't have easy access to the Proceedings / Journal in which this seminal paper originally appeared, and I have not been able to locate it in the web either.

Is there anyone out there who could point me to an online resource where I could obtain a legal (prefereably free) copy of this paper?

Thanks a lot!
Roland
Hi rolando
It seems this article was pubished inthe ICGA journal volume 11 n°4
You may ask a scan to mr Hyatt , he said hd have all the volumes
Bests!
Hi Daniel,
this might be an option as well - thanks!

I have just sent a respective request to Murray Campbell (one of the authors), whom I have been able to locate via ResearchGate.

Roland

Re: Paper "Singular Extensions" by Anantharaman, C

Posted: Tue Sep 16, 2014 2:30 pm
by Ferdy
As an option just search the SE topic, there are very good posts by Bob and others here.

All right here for example.
http://www.talkchess.com/forum/viewtopi ... +extension

Re: Paper "Singular Extensions" by Anantharaman, C

Posted: Tue Sep 16, 2014 2:42 pm
by RolandoFurioso
Ferdy wrote:As an option just search the SE topic, there are very good posts by Bob and others here.

All right here for example.
http://www.talkchess.com/forum/viewtopi ... +extension
I already had a look at several respective posts (thanks!), and as well at an excellent paper by Marsland and Björnsson ("VARIABLE DEPTH SEARCH") that provides pseudocode specifications of two SE algorithms, distinguishing between PV and Cut nodes. Taking a look at the original publication could be worthwhile anyway.