Re: [Bug-gnubg] Confidence intervals from rollouts

[Joern Thyssen]

On Wed, Sep 04, 2002 at 11:31:05AM +0200, address@hidden wrote

> Is it possible to do unstratified rollouts with gnu? 

Not unless you do some hacking, which, I assume, you're able to do :-)

Look in QuasiRandomDice in rollout.c and change:

  else {

    /* normal rollout: doubles allow on first roll */

    if( !iTurn && !( cGames % 36 ) ) {
      anDice[ 0 ] = ( iGame % 6 ) + 1;
      anDice[ 1 ] = ( ( iGame / 6 ) % 6 ) + 1;
      return 0;
    } else if( iTurn == 1 && !( cGames % 1296 ) ) {
      anDice[ 0 ] = ( ( iGame / 36 ) % 6 ) + 1;
      anDice[ 1 ] = ( ( iGame / 216 ) % 6 ) + 1;
      return 0;
    } else
      return RollDice( anDice, rngx );

  }

to

  else 
     return RollDice ( anDice, rngx );

It should be quite simple to add this as an option, so if I find some
time I'll do that. 


> And can anyone tell me 
> what the "Rollout as initial position" does?

Same as a normal rollout, except that doubles are not permitted on the
first roll.

This can be used to calculate the equity for the person on roll in the
initial position (hence the name).

You can also use the option to calculate your own match equity table if
you average the two rollouts with either play on roll.

For example, rolling out 1-away, 2-away Crawford gives:

player 0 on roll: 70.314%
player 1 on roll: 65.801%

indicating that the MWC at 1-away, 2-away is  68.06%

Jørn

-- 
Joern Thyssen, PhD
Vendsysselgade 3, 3., DK-9000 Aalborg, Denmark
+45 9813 2791 (private) / +45 2077 2689 (mobile) / +45 9633 7036 (work)