Singular extensions
Komputer Korner
How did Deeper Blue make them worthwhile?
In the latest ICCA journal, it is stated that Anantharaman in 1990
stated that they broke even because the extra search needed to detect
whether a move is singular cancelled out the gains. Obviously there
was an improvement between then and the Kasparov- Deeper Blue wars .
How did the Deeper Blue team improve the singular extension idea? It
is interesting that Fritz, Hiarcs, Shredder and Genius seem to use
singular extensions as well. Are there others ? Do singular
extensions work well only with full width programs?
--
--
Komputer Korner
The inkompetent komputer
To send email take the 1 out of my address. My email address is
kor...@netcom.ca but take the 1 out before sending the email.
Robert Hyatt
Komputer Korner <kor...@netcom.ca> wrote:
: How did Deeper Blue make them worthwhile?
: In the latest ICCA journal, it is stated that Anantharaman in 1990
: stated that they broke even because the extra search needed to detect
: whether a move is singular cancelled out the gains. Obviously there
: was an improvement between then and the Kasparov- Deeper Blue wars .
: How did the Deeper Blue team improve the singular extension idea? It
: is interesting that Fritz, Hiarcs, Shredder and Genius seem to use
: singular extensions as well. Are there others ? Do singular
: extensions work well only with full width programs?
Here's the issue: When Thomas said "broke even" he was looking carefully
at results on *tactical* suites only, and comparing DT with, to DT without,
and concluded that they didn't make a lot of difference. And in that context,
I believe he was correct.
*but*, in OTB games, I have seen them, game after game, simply blow the
opponent out tactically. I posted the game vs Cray Blitz here last year
where they saw winning a piece 10 full moves (not plies) before Cray Blitz
saw it. I watched them play *Socrates in either Indianapolis or Cape May,
and while *socrates was actually out-searching DT by a ply or so, if you
look at "iteration depth", DT entered into this wonderfully long combination
that they saw again about 10 moves or so before *socrates. The discussion
was around "bugs" for a while, until Mike Valvo finally said (about 6 moves
into the combination as the game actually moved forward) "wait, I see the
point of this finally... when we get *here* white has material hanging,
*and* is threatened with mate, and white has to respond to the threat of
mate and is going to lose material." He was correct.
So it is not clear that it is a "break even" event. I know another program
that is currently using a form of singular extensions and doing quite well.
I'm playing with it again myself, although it is not in current versions since
I am working on the parallel search stuff.
It is clear that most programs are good at finding tactical stuff in positions
that are already tactical. DB seems quite adept at finding tactics in positions
where no tactics seem to exist.
But whether SE is good, bad, or "so-so" will take some time. I lean toward
their working well, now, even though I tried them a long time ago and didn't
like the result. Once I test them some more in a month or two, I'll report
exactly what I find here. That will be one more data point. But the two points
already known are positive, IMHO.
Don't forget that Genius uses some form of SE, according to direct comments from
Lang a few years ago. There's another +. Kittinger said he was using them
over a year ago. That's another one...
b
: --
: --
: Komputer Korner
: The inkompetent komputer
: To send email take the 1 out of my address. My email address is
: kor...@netcom.ca but take the 1 out before sending the email.
--
Robert Hyatt Computer and Information Sciences
hy...@cis.uab.edu University of Alabama at Birmingham
(205) 934-2213 115A Campbell Hall, UAB Station
(205) 934-5473 FAX Birmingham, AL 35294-1170
bruce moreland
On Sun, 12 Apr 1998 02:23:50 -0400, "Komputer Korner"
<kor...@netcom.ca> wrote:
>How did Deeper Blue make them worthwhile?
>In the latest ICCA journal, it is stated that Anantharaman in 1990
>stated that they broke even because the extra search needed to detect
>whether a move is singular cancelled out the gains. Obviously there
>was an improvement between then and the Kasparov- Deeper Blue wars .
>How did the Deeper Blue team improve the singular extension idea? It
>is interesting that Fritz, Hiarcs, Shredder and Genius seem to use
>singular extensions as well. Are there others ? Do singular
>extensions work well only with full width programs?
Upon what evidence are you assuming that any of these programs use
singular extension?
bruce
Komputer Korner
Simple really. They have 2 columns. The first says how many plies full
width and the second gives the singular extension. I was wrong about
Shredder. It doesn't do singular extension per se but you can set
the selectivity to maximum and in effect it becomes like singular
extension because the depth of the longest selective line goes to 30
plies.
--
--
Komputer Korner
The inkompetent komputer
To send email take the 1 out of my address. My email address is
kor...@netcom.ca but take the 1 out before sending the email.
bruce moreland wrote in message
<3533f2ad....@news.seanet.com>...
bruce moreland
>Simple really. They have 2 columns. The first says how many plies full
>width and the second gives the singular extension. I was wrong about
>Shredder. It doesn't do singular extension per se but you can set
>the selectivity to maximum and in effect it becomes like singular
>extension because the depth of the longest selective line goes to 30
>plies.
Singular extension is a search heuristic.
You can do extensions without doing singular extensions.
bruce
Amir Ban
Komputer Korner wrote:
>
> Simple really. They have 2 columns. The first says how many plies full
> width and the second gives the singular extension. I was wrong about
> Shredder. It doesn't do singular extension per se but you can set
> the selectivity to maximum and in effect it becomes like singular
> extension because the depth of the longest selective line goes to 30
> plies.
>
Aren't you confusing singular exptensions with extensions (any sort) ?
Everybody does extensions. The only programmer who said he uses simgular
extensions is Diep, but I think he gave it up since.
Amir
Robert Hyatt
Amir Ban <ami...@m-sys.com> wrote:
: Amir
Actually, two others (besides the DB guys) that I know of: Richard Lang
said that he uses (used?) them. He said this was the main difference between
genius 2 and genius 3. This heard at a chess tournament somewhere, perhaps
at a WCCC event. Dave Kittinger is (was?) using them last year in wchessx
on ICC. I don't know if he continues to do so or not, but suspect he does
based on a few deep tactical moves I've seen him spot nearly instantly...
Chris Whittington
Amir Ban <ami...@m-sys.com> wrote in article <3534CD...@m-sys.com>...
> Komputer Korner wrote:
> >
> > Simple really. They have 2 columns. The first says how many plies full
> > width and the second gives the singular extension. I was wrong about
> > Shredder. It doesn't do singular extension per se but you can set
> > the selectivity to maximum and in effect it becomes like singular
> > extension because the depth of the longest selective line goes to 30
> > plies.
> >
>
> Aren't you confusing singular exptensions with extensions (any sort) ?
> Everybody does extensions. The only programmer who said he uses simgular
> extensions is Diep, but I think he gave it up since.
>
CSTal uses singular extensions in a roundabout sort of way.
Chris Whittington
> Amir
>
Tord Kallqvist Romstad
In article <6h2r22$4ot$2...@juniper.cis.uab.edu>, Robert Hyatt wrote:
>Actually, two others (besides the DB guys) that I know of: Richard Lang
>said that he uses (used?) them. He said this was the main difference between
>genius 2 and genius 3.
Are you sure? I have heard that Mephisto Lyon was the first Lang program
using singular extensions.
Tord
--
"The true Christian should be aware of mathematicians and those who make empty
prophecies. Chances are already there that mathematicians have made a covenant
with the Devil to darken the spirit and confine man in the bounds of Hell"
- St. Augustin
Robert Hyatt
Tord Kallqvist Romstad <rom...@pan.uio.no> wrote:
: In article <6h2r22$4ot$2...@juniper.cis.uab.edu>, Robert Hyatt wrote:
:>Actually, two others (besides the DB guys) that I know of: Richard Lang
:>said that he uses (used?) them. He said this was the main difference between
:>genius 2 and genius 3.
: Are you sure? I have heard that Mephisto Lyon was the first Lang program
: using singular extensions.
: Tord
All I can tell you is what I heard him say at a computer chess event
somewhere... And that was that the main difference between genius 2 and
3 was singular extensions.
I am certain he (nor anyone else) does singular extensions like DB because
it is quite expensive. But I have no idea whether he used them before genius
3 or not, and if he did, why not in genius 2. That's something Richard would
have to answer...
: --
: "The true Christian should be aware of mathematicians and those who make empty
: prophecies. Chances are already there that mathematicians have made a covenant
: with the Devil to darken the spirit and confine man in the bounds of Hell"
: - St. Augustin
--