[ View Thread ] [ Post Response ] [ Return to Index ] [ Read Prev Msg ] [ Read Next Msg ]

BGonline.org Forums

53P-62S-54 — No Jacoby, No Beavers — 3-ply XG2 Rollout

Posted By: Taper_Mike
Date: Thursday, 23 January 2014, at 3:32 a.m.

In Response To: 53P-62S-54 (Michael Nielsen)





White is Player 2

score: 0
pip: 159
Unlimited Game
Jacoby Beaver
pip: 159
score: 0

Blue is Player 1
XGID=-a-B--DaB---dEa--c-e----B-:0:0:1:54:0:0:3:0:10
Blue to play 54

1.Rollout124/20 6/1* eq: +0.0723
Player:
Opponent:
50.71% (G:17.42% B:0.67%)
49.29% (G:12.72% B:0.64%)
Conf.: ±0.0035 (+0.0689...+0.0758) - [100.0%]
Duration: 4 hours 05 minutes
2.Rollout124/20 13/8eq: +0.0530 (-0.0193)
Player:
Opponent:
50.70% (G:14.84% B:0.61%)
49.30% (G:11.56% B:0.46%)
Conf.: ±0.0030 (+0.0500...+0.0560) - [0.0%]
Duration: 3 hours 45 minutes
 
1 31104 Games rolled with Variance Reduction.
Dice Seed: 83943449
Moves: 3-ply, cube decisions: XG Roller

Rollout by Taper_Mike
2013-Aug-12
eXtreme Gammon Version: 2.10.199.2658
53P-62S-54-tm.xgp
[X S19] "<=31
[X S13 H75 R87] "&e

eXtreme Gammon Version: 2.10

53P-62S-54 — Unlimited Games
Bot Seed Equity Nacbracs Link
GnuBg 0.xx 2-ply Trunc 723790475 +30.611 [S X3 R77 H80] /0.360t Taper_Mike
XG 2.10 XGR++ Eval [XGR++ seed] +68.4 [X S13 H75 R87] "&e Taper_Mike
XG 2.10 3-ply, Cubes: XGR 83943449 +72.3 [X S19] "<=31 Taper_Mike (above)
GnuBg 1.02 2-ply Trunc 734982821 +69.959 n[X S11 H68 R87] '/0.360t Taper_Mike
XG 2.10 3-ply, Cubes: XGR 5459943 +72 n[X S22] "<=1 Onur Alan

Key: H = Hit [13/9, 6/1*], R = Run [24/15], S = Split = Z = Reverse split [24/20, 13/8], X = Hit and split [24/20, 6/1*]

Explanation of nacbracs

Thanks for reminding me I had this, Michael!

The question still remains why hit-and-split is best for 53P-62S-54 while split-and-down tops the rollout of 53P-63S-53. As I am not clear on the theory here, I have to defer to my betters. I do like Maik Stiebler’s point regarding the duplications of 1s.

It is instructive to compare GnuBg v1 with v0. Philippe Michel, the creator of the neural nets used in v1, pointed out to me that v1 improves the absolute equity of the top play as well as fine tuning estimates of the differences in equity of the lesser plays. The data above are a case in point.

Mike

Messages In This Thread

 

Post Response

Your Name:
Your E-Mail Address:
Subject:
Message:

If necessary, enter your password below:

Password:

 

 

[ View Thread ] [ Post Response ] [ Return to Index ] [ Read Prev Msg ] [ Read Next Msg ]

BGonline.org Forums is maintained by Stick with WebBBS 5.12.