| |
BGonline.org Forums
72-game rollouts in XG
Posted By: Timothy Chow In Response To: 72-game rollouts in XG (eXtreme Gammon)
Date: Monday, 23 May 2011, at 2:41 p.m.
Xavier wrote:
I did not see much value in rotating on the 2nd roll.
Also i did not see easy way to make it so that Rollout shorter than 1296 would not be biased. I am sure i can put in place some quasi-randomness (randomizing the 2nd roll while making sure each combination is hit at the end).
I thought about this a bit and unless I'm missing something, it shouldn't be hard to rotate on both the first and second rolls without causing biases in shorter rollouts. Create a random permutation of the 36 possible first rolls, and for each first roll, create a random permutation of the 36 possible second rolls, and interleave them.
For example, with coins instead of dice, first create a random permutation of HH HT TH TT, and then for each of those four possibilities, create a fresh random permutation of HH HT TH TT. Say your first random permutation is TH HH TT HT, and your four second-roll permutations are:
first roll TH: TT TH HH HT
first roll HH: HH HT TH TT
first roll TT: TH TT HH HT
first roll HT: TT HH TH HT
Then you should run through the 16 possibilities for the two rolls as follows:
TH TT
HH HH
TT TH
HT TT
TH TH
HH HT
TT TT
HT HH
TH HH
HH TH
TT HH
HT TH
TH HT
HH TT
TT HT
HT HT
If you're worried about the effect of rotating through the first-roll possibilities in the same order every time, then you can further randomize by running through the first-roll possibilities in a different random order each time (but keeping them associated with the same second roll).
Generating random permutations is easy using the Moses-Oakford algorithm (called "Shuffle" on page 2 of this article).
| |
BGonline.org Forums is maintained by Stick with WebBBS 5.12.