[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
[Bug-gnubg] Effective Pip Count Bug
From: |
Ned Cross |
Subject: |
[Bug-gnubg] Effective Pip Count Bug |
Date: |
Tue, 2 Dec 2003 18:39:07 -0800 |
I reported this minor bug back on 9/1, but I did not explain it very well.
I ran across the bug again today, so it still persists.
Running win32, old GUI, on Win2k, build 031122 (1.1457).
In some bearoff situations, the One-Checker Race window is getting the
pipcount for the 2 players swapped. As a result the calculations for Avg
rolls to bearoff, Effective Pip Count, and Estimated Cubeless GWC (blah
on roll): are all reported incorrectly.
In the example I previously provided, the actual GWC for player on roll was
74.2% by 2-ply analysis, but GNU reported 47.2% in the theory window.
Details below.
I think Joern presented output from the no-gui version that showed correct
information, but the two examples below show an issue with the windows GUI
version.
The position I encountered today was similar, with the theory window giving
53.5% GWC instead of 71.3% (see below)
--Ned
The score (after 0 games) is: black 0, white 0 (match to 3 points)
Move number 41: X to play 63
GNU Backgammon Position ID: 1A0AAMkNAAAAAA
Match ID : cAlvAAAAAAAA
+13-14-15-16-17-18------19-20-21-22-23-24-+ O: black
| | | O O O O | OO 0 points
| | | O O | OO
| | | O | OO
| | | | O
| | | | O
v| |BAR| | 3 point match (Cube: 1)
| | | | X
| | | | X
| | | X | XX
| | | X X | XX Rolled 63
| | | X X X X | XX 0 points
+12-11-10--9--8--7-------6--5--4--3--2--1-+ X: white
Pip counts: O 34, X 31
* white moves 6/off 3/off
Alert: missed double ( -0.249)! [very bad]
Cube analysis
Rollout cubeless equity +0.427 (Money: +0.427)
Cubeful equities:
1. Double, take +0.979
2. Double, pass +1.000 ( +0.021)
3. No double +0.730 ( -0.249)
Proper cube action: Double, take
Rollout details:
Centered 1-cube:
0.713 0.000 0.000 - 0.287 0.000 0.000 CL +0.427 CF +0.730
[0.000 0.000 0.000 - 0.000 0.000 0.000 CL 0.001 CF 0.002]
Player black owns 2-cube:
0.717 0.000 0.000 - 0.283 0.000 0.000 CL +1.144 CF +0.979
[0.000 0.000 0.000 - 0.000 0.000 0.000 CL 0.001 CF 0.002]
Truncated cubeful rollout (depth 11) with var.redn.
Output generated Tue Dec 02 14:15:37 2003
by GNU Backgammon 0.14-dev 1.1452 031119 (Text Export version 1.62)
Yet, under Analyse/One-Checker Race/One Checker Bearoff GNUBG gives:
Number of rolls to bear off, assuming each player has one chequer only.
Pips Avg. rolls Std.dev.
black 31 4.370 1.019
white 34 4.742 1.058
Estimated cubeless gwc (white on roll): 53.5332%
Effective pip count:
EPC Wastage
black 35.685 4.685
white 38.725 4.725
EPC = Effective pip count = Avg. rolls * 8.167
Wastage = EPC - Pips
_______________________________________________________
black (O, 0 pts) vs. white (X, 0 pts) (Match to 5)
Game number 1
Move number 41: X to play 21
GNU Backgammon Position ID: 6B4AALJ1AAAAAA
Match ID : UQmlAAAAAAAA
+13-14-15-16-17-18------19-20-21-22-23-24-+ O: black
| | | O O O | OO 0 points
| | | O O | OO
| | | O O | O
| | | O | O
| | | | O
v| |BAR| | 5 point match
| | | | X
| | | | X
| | | X | X
| | | X X X | X Rolled 21
| | | X X X X X | XX 0 points
+12-11-10--9--8--7-------6--5--4--3--2--1-+ X: white (Cube: 2)
Pip counts: O 43, X 38
* white moves 6/4 1/off
Alert: missed double ( -0.170)! [very bad]
Cube analysis
2-ply cubeless equity +0.483 (Money: +0.483)
0.742 0.000 0.000 - 0.258 0.000 0.000
Cubeful equities:
2-ply cubeful 100% speed [world class]
1. Double, take +0.981
2. Double, pass +1.000 ( +0.019)
3. No double +0.811 ( -0.170)
Proper cube action: Redouble, take
Yet, under Analyse/One-Checker Race/One Checker Bearoff GNUBG gives:
Number of rolls to bear off, assuming each player has one chequer only.
Pips Avg. rolls Std.dev.
black 38 5.230 1.123
white 43 5.842 1.197
Estimated cubeless gwc (white on roll): 47.2289%
Effective pip count:
EPC Wastage
black 42.712 4.712
white 47.712 4.712
EPC = Effective pip count = Avg. rolls * 8.167
Wastage = EPC - Pips
[Prev in Thread] |
Current Thread |
[Next in Thread] |
- [Bug-gnubg] Effective Pip Count Bug,
Ned Cross <=