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

BGonline.org Forums

N-ply Algorithm

Posted By: Rich Munitz
Date: Sunday, 23 January 2011, at 5:32 a.m.

In Response To: N-ply Algorithm (Jason Lee)

Well, I was on board with Rich until this... I wanted 7-ply to be stronger than XGR+ too. Alas...

Agreed. All bets are off with this.

Relative to an exhaustive 7-ply evaluation of all roll/move combinations employing the minimax algorithm, the XG 7-ply does no pruning on the roll combinations, but nearly the most severe possible pruning on the moves - selecting only one move for deeper consideration on each roll out of the many legal moves possible. XG 7-ply will give us a truncated 1-ply rollout result of depth 7 that has 100% confidence (i.e. an infinitely long rollout of depth 7 will give the same result). However, if the rollout is starting off with some donkey plays on the first roll at 1-ply that it doesn't make at 2-ply, the accuracy of the entire result could be thrown off, even though there is absolute statistical confidence in the outcome. XGR+ meanwhile potentially starts off with better plays but then produces very low statistical confidence in its final number.

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.