[Bug-gnubg] GnuBg settings to mimic XGR++

[M. J. Mannon]

Can anyone give me guidance on setting up a GnuBg rollout that mimics the XG 
Roller++ settings?
 
According to the XG help system, an XGR++ rollout uses these XG settings:
 
XGRoller++ Checker Play
-------------------------
360 games with VR
Truncated after 5 moves using 3-ply
First 2 decisions in 3-ply
Then 2-ply checker and 3-ply cube
 
XGRoller++ Cube Decision
-------------------------
360 games with VR
Truncated after 7 moves using 3-ply
Stop when reaching 0.010 Confidence (minimum 180 games)
First 2 decisions in 3-ply
Then 2-ply checker and 3-ply cube 
 
I am trying to create a rollout setting for GnuBg that mimics the truncated 
rollout used by XG Roller++.
 
I am confused by the settings in the GnuBg Rollout dialog. In order truncate a 
GnuBg rollout after 5 plies, should I enter the number 4 or the number 5 into 
the field “Truncate at ply?” 
 
Similarly, I am not sure what number to enter into the field “Change eval after 
ply.”
 
----------------------------------------------------------------------
The issue is not as simple as it may seem. See the messages in the following 
thread at BgOnline. 
 
Stick vs Merriman: XG++ Data Makes No Sense 
Posted By: Daniel Murphy <address@hidden>
Date: Monday, 30 July 2012, at 7:27 p.m.
http://www.bgonline.org/forums/webbbs_config.pl?noframes;read=127133
 
----------------------------------------------------------------------
The following text explains my difficulties best. It is a part of the 
aforementioned thread.
 
GnuBg settings to mimic XGR++
Posted By: Taper_Mike <address@hidden>
Date: Tuesday, 31 July 2012, at 6:28 a.m.
http://www.bgonline.org/forums/webbbs_config.pl?noframes;read=127164
 
As a computer programmer myself, I must confess that sometimes we can be an odd 
bunch. For instance, if you ask a computer programmer to count to ten, he (or 
she) will often start at zero. Then, as if to compensate, the programmer stops 
at nine! 

GnuBg seems to do the same thing. It uses the name “0-ply” for the same 
analysis other bots describe as “1-ply.” If you set GnuBg to truncate a rollout 
after 7 plies, it performs evals at plies 0-6. 
 
In its dialog box for Rollout Settings, GnuBg has a section called 
“Truncation.” Within it is a field called “Truncate at ply.” I am a bit 
confused by this language. Should my response here use the lower ply numbers 
GnuBg prefers to describe its rollouts? If I want to simulate the 7 plies of an 
XGR++ eval, does that mean I should enter a 6 in this field? 
 
The way I see it, if you truncate a rollout “at ply 6,” that means that you 
have performed evals at plies 0-6. Because you begin with ply 0, stopping at 
ply 6 means you made evals at 7 different plies. 
 
The answer to my question, however, may be no. In the rollout posted by Daniel, 
he selected 7. In the output for the rollout, this generates the message, 
“Truncated cubeful rollout (depth 7) with var. redn.” Depth 7 sounds like what 
we want. “Truncate at ply 7” means “depth 7.” 
 
Even if this is the case, the language is awkward. “Truncate after 7 plies 
(plies 0-6)” would be better. Given GnuBg’s numbering scheme, “Truncate at ply 
7” ought to mean depth 8. 
 
Similarly, the section entitled “Evaluation for later plies” has a field called 
“Change eval after ply.” In order to mimic XGR++, we want all decisions after 
the first two to be made using lower settings. Well, the first decision (in the 
parlance of GnuBg) happens at ply 0, and the second, at ply 1. The third 
decision occurs at ply 2. When selecting a value for the field “Change eval 
after ply,” should I be using these lower numbers? In order to have GnuBg use 
weaker setting for its ply 2, should I tell GnuBg to “Change eval after ply 1?” 
 
Once again, the answer may be no. When Daniel enters a 2 in this field, the 
output contains the message, “Different evaluations after 2 plies...” This 
seems to indicate that the first two plies, ply 0 and ply 1, use the stronger 
settings. Subsequent plies, beginning with ply 2, use the weaker settings. 
 
But I am not sure. 
 
I tried to explain this in my last post, but I feel I did not make my concerns 
clear. In fact, if you now understand what I am driving at, I may have done the 
disservice of introducing doubt about something you already had clear. It 
wouldn’t be the first time. 
 
It’s a programmer thing. We are way too literal. 
 
----------------------------------------------------------------------
Also, be sure to read this message at the (current) end of the thread. It 
includes several rollouts that suggest I should enter the number 4 in the field 
“Truncate at ply” in order get the 5 plies used by XG.
 
GnuBg settings to mimic XGR++ 
Posted By: Taper_Mike <address@hidden>
Date: Tuesday, 31 July 2012, at 11:58 a.m.
http://www.bgonline.org/forums/webbbs_config.pl?noframes;read=127178
 
Please reply here, or better yet, by posting a message in the thread at BGO.
 
Thanks,
Mike Mannon