Øystein Schønning-Johansen wrote:
I think the price indication of the IPR is a bit high. (I guess Xavier
reads this)
It might be high, but I don't think that's necessarily the right way to
calculate it. First of all, if Xavier doesn't find a buyer, I don't think
he loses too much. So he
On Sun, 11 Jun 2023, Lasse Hjorth Madsen wrote:
Thanks, Tim. I don't think the problem is simply that I fail to factor
gammons in, because my dead-cube take points agree with GNU, also for
gammonish positions. It is only when I try to *both* account for gammons
and a fully live cube, that I
Lasse Hjorth Madsen wrote:
Funny enough, when I try examples with no gammons, I can reproduce GNUs
fully live TPs exactly.
If anybody can explain what I’m doing wrong, I would be forever
grateful.
I'd also be interested in the answer to this question. Given your comment
about gammons, I
In chess, there have been some attempts to measure the "complexity" of a
position. Guid and Bratko have written a couple of a papers that discuss
the topic, and you can find a few other papers by using Google Scholar to
find papers that cite these papers.
Turker Eflanli wrote:
Does anyone know a way to convert a GNU analyzed sgf file to xg format? If
not, is there documentation that explains the sgf format?
Here's a link to some code that someone wrote to convert a different
format to sgf.
There is a subtle point about luck that is not well understood even by
some professional mathematicians.
In a series of games (or matches, but for simplicity let me focus on
games), one must distinguish between
1. counting the number of games in which I was luckier, and
2. determining who
Rich Heimlich wrote:
Well, let's also be clear about some harsh realities here. Backgammon is
always going to be seen in the same light as checkers. It just is. No
matter how big the following, it's never going to be chess or have that
sort of following. It's pretty much seen as a beginner's
Just to be clear, when I talked about regaining ground, I was thinking
about "market share" and not speed or playing strength. Market share, of
course, is controlled by many factors, but if one product has active
development and another product does not, then the product with active
On Wed, 2 Dec 2020, Ian Shaw wrote:
I think XG has gained in popularity over gnubg mainly because it is
faster, and only marginally because of the slight playing strength
advantage.
I think it's hard to be sure why XG has gained in popularity over GNUBG,
but I would agree with you that the
Joseph Heled wrote:
My recollection is that GNUBG was much stronger than Snowie even 15
years ago. And 2020 GNUBG is stronger than 2005 GNUBG.
Am I missing something?
According to Frank Berger, Torsten Schoop conducted a "Big Bot Shootout"
back in 2006.
pierre zakia wrote:
I am aware of what you wrote, but the position doesn't look so tricky to
get fooled at 4 ply (compare to roll home a massive backgame for
instance), hence my question.
If the decision looks obvious to you, then I suspect that it means that
you have a general tendency to
Pierre Zakia wrote:
In the position here below, I am curious to understand what in Gnubg
engine yields such discrepancy between 4 ply and roll out.
GNU Backgammon ID de position: G27HAAiY2+ABAw
ID de match : MAHyAAAE
Thanks for posting this position---it's very
Aaron Tikuisis wrote:
That is interesting, I did not realize that gnubg misplays race
positions much. What are some examples?
There was some discussion of this issue on BGOnline back in 2013. It was
in connection with eXtreme Gammon rather than GNU, but I wouldn't be
surprised if GNU
Øystein Schønning-Johansen wrote:
And again: You should not build such a one-sided database in the first
place. You will not get a better player if you use such a large bearoff
database.
I'm not saying that anyone should put in the time to fix this bug. But
the last time I tried to
I recently learned about makehyper. A friend of mine ran it on his
computer and it seemed to terminate successfully. I would like to be able
to examine the equities using my own C programs, rather than using them
inside GNUBG. But to do this, I need to understand the format of the
database
Lots of interesting comments by Sarah Payne and Chris Bray, most of which
make sense to me. But there's one part that I have trouble believing.
In my opinion, a phone version would broaden appeal / access but the
most critical issue is the neural nets. XG feels like a very different
animal
Philippe Michel wrote:
The engine doesn't "plan ahead", does it ? It approximates the
probabilities of the game outcomes from the current position (or we can
say its equity for simplification).
My understanding is that its potential accuracy depends on the neural
network (architecture +
On Sun, 8 Dec 2019, Joseph Heled wrote:
Agreed, but from a practical point of view, not caring about
non-reachable positions and positions with a very low probability is
good enough for a playing-bot.
We probably shouldn't clutter the mailing list further with this debate,
but *from a
On Sun, 8 Dec 2019, Joseph Heled wrote:
Of course you need to weight every position with the probability it
occurs in actual play.
You say "of course," but I don't agree. Weighting things in that way
amounts to demanding perfection only from the starting position. In my
book, perfection
On Sun, 8 Dec 2019, Joseph Heled wrote:
Yes. But there is the question of how easy it is to "navigate" to those
positions. can you reliably get to those positions against a bot and win
from their ignorance?I have my doubts.
I've done some experimentation of this sort, but rather than quote my
On Sat, 7 Dec 2019, Nikos Papachristou wrote:
The moral: If one needs to experience the full power of the bg bots one
needs to change the default settings which are configured for the
average user. Whatever errors bots occasionally make at their
evaluations, they make up by searching deeper.
On Thu, 5 Dec 2019, Nikos Papachristou wrote:
My personal view on improving GNUBG: Why not try to "upgrade" your
existing supervised learning approach? There have been lots of advances
in optimization/regularization algorithms for neural networks in the
past years and it might be less
On Thu, 5 Dec 2019, Joseph Heled wrote:
To do a "AlphaGammon" would require, at a minimum, one or more googlers
who have worked on AlphaGo or AlphaChess and are willing to contribute
time and resources.
The situation is rosier than that, in my opinion. In chess, for example,
Leela Chess
On Thu, 5 Dec 2019, Joseph Heled wrote:
I had the same idea the day I heard they cracked go, but just saying
something is a good idea is not helpful at all in my book.
Well, other people may have other books.
Also, it's my impression that many people *don't* think this is even a
worthwhile
On Wed, 4 Dec 2019 at 16:19, I wrote:
It's surely occurred to some people that it would be interesting to see
what would happen if one were to apply "AlphaZero methods" to
backgammon.
On Wed, 4 Dec 2019, Joseph Heled wrote:
Well, are you?
Not any time soon.
Tim
Just wanted to put this out there...
It's surely occurred to some people that it would be interesting to see
what would happen if one were to apply "AlphaZero methods" to backgammon.
I suspect that doing so would finally plug the one glaring hole in the
repertoire of the current leading bots
Ian,
The topic of "fish METs" is an interesting one. I think that they can be
useful for non-contact race cubes, where humans can make precise
calculations over the board using race formulas. Beyond that, I'm a
little skeptical about how practical they are.
One thing that I feel that
On Wed, 13 Nov 2019, Joseph Heled wrote:
I thought it was clear that what we want to establish a difference
between two (say players X and Y) by running games and testing that
gammon-rate(X) != gammon-rate(Y).
Thanks for the clarification.
If you're pitting X against Y and your null
On Wed, 13 Nov 2019, Joseph Heled wrote:
"but for most practical purposes this is an irrelevant technicality"
Are you saying that I can treat each of the 4 estimates independently?
That is, use sqrt(pq/N) as the std for each? seems problematic to me :)
No, I didn't say that. As I said, the
On Tue, 12 Nov 2019, Joseph Heled wrote:
Hi Timothy,
Here is a stats question I encounter from time to time.
Suppose I run N BG games and collect the average win rates and gammon
rates.
4 estimates which are dependent as they sum to 1. How do I determine
the confidence intervals for each?
Ian,
Thanks for putting all this effort into a new MET!
I don't know too much about the innards of GNU Backgammon, but I do know
something about math and statistics.
In terms of how many matches you would have to play between GNU-old-MET
and GNU-new-MET, that depends on how much stronger
It's been a long time since I've tried to download and install gnubg, but
I just got a new laptop, and I was surprised to find that the website
www.gnubg.org doesn't seem to work any more. Using a link in an old email
message of Michael Petch's---
I'm interested in learning a bit more about the details of how GNU's
neural net was trained, and in particular what was done differently
compared to other neural nets. I found Oystein Johansen's paper from
December 2007 on the GNUBG website. It credits Joseph Heled with the main
ideas, and
Lucas wrote:
Last year i tested using Fibs,( were i did in the past 8 bots), 2 bots
one set to play Worldclass and the other at grandmaster
so 2 ply against 3 ply
they played 3000 5 point matches
Worldclass the lesser setting had a winrate of 55 %
Ian Shaw wrote:
I'd be surprised if just 3000
Luck-adjusted results are computed using MWC while designations such as
Expert/Beginner/Awful! are computed using EMG. I suspect that the same is
true of Go to bed/Go to Las Vegas. My guess is that Richard made some
relatively small EMG errors at a late stage in the match where they made a
Michael Petch wrote:
On 2014-08-17 7:01 PM, jl...@juno.com wrote:
I just downloaded the update version of GNU Backgammon. I thought of a
feature that could be added. When there is no valid move (no matter
what is rolled), then skip the roll entirely. It does not make sense to
waste time
On Sun, 2 Mar 2014, boomslang wrote:
I cannot give you Snowie error rates for the several settings, but I can
give you an estimated FIBS rating for various noise settings.
[...]
Looking at the table, I would say that the noise settings for
intermediate, casual player and beginner are set too
In the GNU documentation it says the following about predefined settings:
Beginner: This setting uses no lookahead and adds up to 0.060 noise to the
evaluation. With this setting GNU Backgammon will evaluate like a
beginner.
Casual play: This setting uses no lookahead and adds up to 0.050
This may be just another manifestation of the same bug that I reported
recently, but just in case...
As before, start up the graphical version of GNU on a Windows machine.
Enter edit mode, and paste in the following ID:
XGID=--BB-BB---a-B--bdA:1:1:1:61:0:0:1:0:10
Click on the little
Michael Petch wrote:
In step 5, I assume you meant control instead of shift.
Whoops, yes...that's what I meant.
Glad you're able to confirm the bug. As I've mentioned before, this is
not a rare occurrence...it happens to me frequently, though I'm not sure
exactly what triggers it.
I
I'm not sure if this has been fixed in the newest release, but I'm going
to report this anyway. (I've experience this behavior frequently before,
and have even reported it to this list, but until now was having trouble
reproducing it.) I'm running 0.90-mingw 20111003 on a Windows XP box.
1.
On Saturday, 7 January, I wrote:
I wanted to study the following position from a match between Falafel and
Neil Kazaross.
XGID=-b--BBCA-Bb-bcae-C--AA:0:0:1:00:5:3:0:11:10
On a Windows 7 machine, I copied the XGID, then fired up GNU and clicked
on the Edit button. Then I pasted the XGID
Suppose I do everything that I reported in my previous email, except that
instead of clicking on the White checker to set White on roll, I exit edit
mode by clicking the Edit button. Then I try to double by pressing
Ctrl-D. The computer complains that I can't double because it's not my
turn
I wanted to study the following position from a match between Falafel and
Neil Kazaross.
XGID=-b--BBCA-Bb-bcae-C--AA:0:0:1:00:5:3:0:11:10
On a Windows 7 machine, I copied the XGID, then fired up GNU and clicked
on the Edit button. Then I pasted the XGID into the GNU window. The
desired
1. Start up GNU and click on the New button, and click the button to set
up a new position. You'll be placed in edit mode.
2. Click on the checker tray to clear the board, and then click on various
points to set up some checker position. As the final step, click on the
white checker icon
What I am about to describe is something that I have unfortunately been
unable to reproduce on demand. However, I'm reporting it anyway in case
someone else has had similar problems.
Sometimes, I will hold down the Ctrl key and drag a checker across the
board. In the process of doing
On the Windows version of gnubg, I recently started a long rollout of a
cube decision in order to View Statistics. At some point I decided to
click on Stop rather than wait for the rollout to complete. The window
said that it had completed 18408 trials for gnubg owns 2-cube and 18407
trials
Ian Shaw wrote:
Gnubg used to be able to break down the data from a rollout, showing the
cube-level at which games finished, games with hits, etc.
I can't find this option anymore. Is it still available? I don't use it
often, but there are times when I want to.
This is bad news if it's
I use gnubg on a Linux system, a snapshot from a few months ago.
Sometimes, when I am in edit mode, dragging checkers around the board by
holding down the ctrl key, some random checker (usually a blot, but not
always) will suddenly get transferred to the bar when I release the mouse
at the
On Sat, 31 Jul 2010, Christian Anthon wrote:
I set up a position, type hint to get a window with evaluations of the
moves, click on a move that I want a deeper-ply evaluation of, and click
(say) on the 4-ply button. GNUBG goes into a deep think, which I cannot
interrupt.
There is a
On Fri, 30 Jul 2010, Christian Anthon wrote:
Stopping seems to work for me. Can you describe the exact steps to
reproduce the problem.
I set up a position, type hint to get a window with evaluations of the
moves, click on a move that I want a deeper-ply evaluation of, and click
(say) on the
I just installed the latest version of gnubg. It's been about a year
since my previous update.
It used to be possible to stop a computation (like a 4-ply evaluation) if
I got tired of waiting for it to finish. I don't seem to be able to do
that any more. I see a stop sign next to the Gnubg
Currently I do the majority of my rollouts (on my Linux box) by preparing
a text file that looks something like this---
set gnubgid 0DNjwDDg6+ABDA:cAkT
hint
cmark move set rollout 1 2 3
analyse rollout move
save position mypos.sgf
---and then piping it to gnubg -t. What I want to do
I'd like to request that a feature be added to GNU Backgammon to allow the
user to save the statistics about number of games ending with the cube on
various values, and whether it was a gammon/backgammon or not.
I raised this issue on Stick's forum and got three responses. Joe Russell
was
On Sun, 10 Jan 2010, Michael Petch wrote:
Someone will have to correct me if I am wrong but I don't believe that
info is actually saved to the SGF file, so when it gets reread it
disappears.
That is unfortunate if true. I've been studying cube efficiency and at
the moment it seems to me
Suppose I have a .sgf file in which I've saved the results of rolling out
a cube decision or some checker plays. I load it into GNUBG and then I
want to View Statistics of the number of times the game ended at various
cube levels. Is this possible? If I click on Hint then a box pops up
but
I've installed a version of gnubg from early August 2009 on a
multi-processor machine running Fedora 10, and the way I do long rollouts
is to type something like
gnubg -t inputfile outputfile
This works fine, but I would like to take advantage of the multiple
processors to get the
I've done a rollout of the top 8 candidate moves in a particular position
and have the SGF file. Now I want to save the result to HTML format.
When I do so, however, only the top 5 moves are exported. How do I get
all 8 moves to be exported?
Tim
On Tue, 17 Nov 2009, Massimiliano Maini wrote:
2009/11/16 Timothy Y. Chow tc...@alum.mit.edu:
The multivariate tail probability, for
example, tells you only the probability that some strange event will occur
*under the assumption that the equities are equal to the estimated
equities
Hades wrote:
Hmm ok I see. Too bad, I thought there is an option to enable that
feature. Can someone tell me how often the developers update particular
things or if you can contact them?
The developers follow this email list closely so you're doing the right
thing already. They update
On Tue, 17 Nov 2009, Massimiliano Maini wrote:
If the above is true, on top of assuming a uniform prior, you assume
that the real pdf are the ones you have on your last step.
This is not far from assuming that the real equities are the estimated
ones. Duh, I must be missing something ...
Massimiliano Maini maxma...@gmail.com wrote:
Why don't we show the % instead of the JSD ? It's much more reasonable.
The trouble with this is that the percentages don't mean what you think
they mean.
In the bgonline thread, some people got the misimpression that the
points I was making were
Myshkin LeVine mysh...@verizon.net wrote:
The standard rules of backgammon do not allow for Beavers and
Raccoons in match play.
[...]
So yes it is your mistake. :-)
It's a bit harsh to call it a mistake, even in jest. I would say
instead: Using beavers and raccoons in match play is
On Sun, 4 Oct 2009, Philippe Michel wrote:
save match will save a money session.
There is a analyse session synonym to analyse match but no such
thing for save or load...
Thanks...that helped!
Now suppose I want to review the session. I tried
load match savedmoneysession.sgf
which
Occasionally I'm at a library or something where there's a public computer
from which I can ssh to my home machine to use gnubg in line-terminal
mode, but cannot bring up the GUI.
I want to be able to play a money-game session and save all the games.
Better still would be to analyze the session
Roy A. Crabtree roy.crabt...@gmail.com wrote:
What I am denoting is that the NNP learns the attributes of the
specific roll generator you use. And does it probably better than a
human can.
This is interesting. The Mersenne Twister is linear, so in principle it
is easy to learn how to
I just rolled out a cube decision and gnubg evaluated no double at
greater than +1.000 equity despite the fact that a gammon was no longer
possible. Below is the text output; I can email the .sgf file if anyone
wants it.
Tim
The score (after 0 games) is: gnubg 0, tchow 0
Match Information:
I thought that gnubg did cubeful rollouts using a live cube, rather than
doing cubeless rollouts and using some formula to convert them to cubeful
values. However, David Ullrich found some evidence to the contrary:
http://groups.google.com/group/rec.games.backgammon/msg/f639d6f0ad2497cd
Tim
On Tue, 11 Aug 2009, Christian Anthon wrote:
If you have a fairly recent version you can save your rollout settings
in individual files by clicking saveas in the rollout settings dialog
in the graphical user interface. You can then make customized files
that you can load like this
load
Michael Petch wrote:
Put this in a file rollout1
set player 0 name player1
set player 1 name player2
set gnubgid HN/ACAZiWzAMIw:sAFnAaAAEAAA
hint
cmark move set rollout 1 2
analyse rollout move
save position rollout1.sgf
And then did
nice -19 gnubg -t rollout1 rollout.log 21
This
On Wed, 17 Jun 2009, Michael Petch wrote:
To get that data, I do cube rollouts by clicking on the position then go
to the Analyze menu button, select Rollout/Cube. That analysis
window should have an ok button as you describe. Click on Ok. Now
the window is gone. Now make sure on the View
I'm having trouble with rollout results. Some years ago when I used
gnubg, after I did a rollout, I would click OK and another window would
pop up, giving the equities for double/take and no double. Now when I
click OK the window simply disappears and I can't even figure out how to
find that
Now that I have installed gnubg in my own directory on a shared Linux
machine, I am wondering if I can do rollouts as background jobs. That is,
I would like to be able to prepare an input file with a given position,
and then do something like
/bin/nice -19 gnubg -rolloutoption inputfile
On Fri, 12 Jun 2009, Christian Anthon wrote:
cvs -z3 -d:pserver:anonym...@cvs.savannah.gnu.org:/sources/gnubg co
This didn't work for me. Is the syntax correct? But it doesn't really
matter, because your other instructions did work. Thanks!
Let me mention that what I originally did was to
On Mon, 8 June 2009, I wrote:
On Sun, 7 Jun 2009, Michael Petch wrote:
This should help you:
http://www.rpm.org/max-rpm/s1-rpm-anywhere-different-build-area.html
You may get errors building with that rpm. I'd probably recommend
downloading a latest snapshot or retrieve the latest from CVS.
I have still been unsuccessful at trying to build gnubg using rpmbuild.
Everything seems to work more or less O.K. until I get this message:
checking for GtkGLExt - version = 1.0.0...
*** pkg-config cannot find gtkglext-1.0 = 1.0.0
*** Set the environment variable PKG_CONFIG_PATH to point to the
I am trying to install gnubg on a university computer where I have an
account but do not have admin privileges. I believe the machine is
running RedHat. I downloaded
gnubg-0.15-4.src.rpm
which I renamed gnubg.rpm for brevity, and tried to build it:
% rpmbuild --rebuild gnubg.rpm
77 matches
Mail list logo