BGonline.org Forums

21$-41U-31H-52H-51H-41H-31

Posted By: Philippe Michel
Date: Wednesday, 26 June 2013, at 9:06 p.m.

In Response To: 21$-41U-31H-52H-51H-41H-31 (Taper_Mike)

Still, I think the trunc does a better job finding the top plays than Tim’s 15k rollout.

No way. The worse than -0.6 equities, the enormous differences between plays that make both anchors, are extremely suspect.

But in any case this doesn't look like a position where a few plies of lookahead would help. After these, the positions will be mostly the same, there is no obvious tactical quandary where the truncated rollout would somehow look over the obstacle.

It would be interesting, however, to see how well the new neural nets in GnuBg version 1.00 handle this position.

The score (after 0 games) is: GNUbg 0, You 0

Move number 4: You to play 31

GNUbg	160
You	214
Position ID: 4HPkARFgjjwgWQ Match ID: cIkFAAAAAAAA

• You moves bar/22 8/7

#		Ply	Move	Equity
•	1	R	bar/22 8/7	-0.240
			0.473 0.093 0.005 - 0.527 0.257 0.025 -0.238 -0.240 0.002 0.002 0.001 - 0.002 0.002 0.001 0.007 0.011
			Full cubeful rollout with var.redn.
			7776 games, Mersenne Twister dice gen. with seed 686358663 and quasi-random dice
			Play: world class 2-ply cubeful prune [world class]
			keep the first 2 0-ply moves and up to 6 more moves within equity 0.16
			Skip pruning for 1-ply moves.
			Cube: 2-ply cubeful prune [world class]
	2	R	bar/22 6/5*	-0.256 ( -0.016)
			0.474 0.095 0.006 - 0.526 0.269 0.028 -0.248 -0.256 0.003 0.003 0.001 - 0.003 0.003 0.001 0.011 0.015
			Full cubeful rollout with var.redn.
			7776 games, Mersenne Twister dice gen. with seed 686358663 and quasi-random dice
			Play: world class 2-ply cubeful prune [world class]
			keep the first 2 0-ply moves and up to 6 more moves within equity 0.16
			Skip pruning for 1-ply moves.
			Cube: 2-ply cubeful prune [world class]
	3	R	bar/24 8/5*	-0.271 ( -0.031)
			0.463 0.102 0.004 - 0.537 0.251 0.032 -0.250 -0.271 0.001 0.002 0.000 - 0.001 0.002 0.001 0.005 0.009
			Full cubeful rollout with var.redn.
			7776 games, Mersenne Twister dice gen. with seed 686358663 and quasi-random dice
			Play: world class 2-ply cubeful prune [world class]
			keep the first 2 0-ply moves and up to 6 more moves within equity 0.16
			Skip pruning for 1-ply moves.
			Cube: 2-ply cubeful prune [world class]
	4	R	bar/22 24/23	-0.273 ( -0.033)
			0.461 0.097 0.004 - 0.539 0.238 0.015 -0.230 -0.273 0.001 0.001 0.000 - 0.001 0.001 0.001 0.004 0.008
			Full cubeful rollout with var.redn.
			7776 games, Mersenne Twister dice gen. with seed 686358663 and quasi-random dice
			Play: world class 2-ply cubeful prune [world class]
			keep the first 2 0-ply moves and up to 6 more moves within equity 0.16
			Skip pruning for 1-ply moves.
			Cube: 2-ply cubeful prune [world class]
	5	R	bar/22 11/10	-0.282 ( -0.042)
			0.462 0.093 0.004 - 0.538 0.240 0.018 -0.237 -0.282 0.001 0.001 0.000 - 0.001 0.001 0.001 0.004 0.008
			Full cubeful rollout with var.redn.
			7776 games, Mersenne Twister dice gen. with seed 686358663 and quasi-random dice
			Play: world class 2-ply cubeful prune [world class]
			keep the first 2 0-ply moves and up to 6 more moves within equity 0.16
			Skip pruning for 1-ply moves.
			Cube: 2-ply cubeful prune [world class]

Compared to that, the 2 ply evaluations have the slotting plays seriously underestimated but there is no ridiculous discrepancy like between Timothy's rollout and your truncated one (and most probably 0.90 evaluations as well).

Alert: bad move ( -0.067)

#		Ply	Move	Equity
	1	2	bar/22 24/23	-0.251
			0.469 0.103 0.005 - 0.531 0.243 0.023
	2	2	bar/22 11/10	-0.297 ( -0.046)
			0.461 0.098 0.004 - 0.539 0.248 0.030
•	3	2	bar/22 8/7	-0.319 ( -0.067)
			0.464 0.098 0.004 - 0.536 0.267 0.037
	4	2	bar/22 6/5*	-0.335 ( -0.084)
			0.469 0.121 0.005 - 0.531 0.301 0.051
	5	2	bar/24 8/5*	-0.342 ( -0.091)
			0.455 0.105 0.006 - 0.545 0.265 0.043

Messages In This Thread

• Inchoate backgame
  Timothy Chow -- Thursday, 30 December 2010, at 6:57 p.m.
  • Inchoate backgame
    Casper van der Tak -- Thursday, 30 December 2010, at 7:03 p.m.
  • Inchoate backgame
    Bob Koca -- Thursday, 30 December 2010, at 8:48 p.m.
  • Inchoate backgame
    phil simborg -- Thursday, 30 December 2010, at 11:42 p.m.
  • Inchoate backgame
    Matt Cohn-Geier -- Friday, 31 December 2010, at 3:45 a.m.
    • Inchoate backgame
      mamabear -- Saturday, 1 January 2011, at 10:02 p.m.
  • Inchoate backgame
    neilkaz -- Friday, 31 December 2010, at 5:12 a.m.
    • Inchoate backgame
      christian munk-christensen -- Sunday, 2 January 2011, at 5:22 p.m.
  • Contradictory rollouts! Read all about it here
    Timothy Chow -- Friday, 31 December 2010, at 8:32 p.m.
    • Contradictory rollouts! Read all about it here
      Stick -- Friday, 31 December 2010, at 10:53 p.m.
      • Contradictory rollouts! Read all about it here
        Timothy Chow -- Saturday, 1 January 2011, at 12:41 a.m.
        • Contradictory rollouts! Read all about it here
          Stick -- Saturday, 1 January 2011, at 12:57 a.m.
  • 21$-41U-31H-52H-51H-41H-31 — Unlimited Games — 2-ply GnuBg Trunc
    Taper_Mike -- Monday, 24 June 2013, at 12:59 a.m.
    • 21$-41U-31H-52H-51H-41H-31 — Unlimited Games — 2-ply GnuBg Trunc
      Stick -- Monday, 24 June 2013, at 1:58 a.m.
      • 21$-41U-31H-52H-51H-41H-31
        Taper_Mike -- Monday, 24 June 2013, at 3:41 a.m.
        • GNU version 1.00
          Neil Robins -- Tuesday, 25 June 2013, at 11:44 p.m.
          • GNU version 1.00
            Michael Petch -- Wednesday, 26 June 2013, at 4:21 a.m.
            • GNU version 1.00
              Taper_Mike -- Wednesday, 26 June 2013, at 5:36 a.m.
            • GNU version 1.00
              Tom Keith -- Wednesday, 26 June 2013, at 1:03 p.m.
              • GNU version 1.00
                Michael Petch -- Wednesday, 26 June 2013, at 3:49 p.m.
                • GNU version 1.00
                  Tom Keith -- Wednesday, 26 June 2013, at 4:50 p.m.
            • GNU version 1.00
              Seth -- Friday, 28 June 2013, at 10:05 p.m.
        • 21$-41U-31H-52H-51H-41H-31
          Philippe Michel -- Wednesday, 26 June 2013, at 9:06 p.m.
        • 21$-41U-31H-52H-51H-41H-31
          Philippe Michel -- Wednesday, 26 June 2013, at 9:12 p.m.