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

BGonline.org Forums

why is this still haunting us?

Posted By: Chuck Bower
Date: Monday, 8 June 2009, at 7:52 p.m.

In Response To: why is this still haunting us? (Bob Koca)

I'm not sure which bot you refer to. Here is my understanding of both bots' rollout methods:

1) SW -- if "live cube", two 'simultaneous' rollouts are performed, one purely cubeless $-play (from which cube-adjustment is later applied) and one with a truly live cube. If you don't run with a live-cube, only one rollout is performed. (This contributes to why few people want to run SW live-cube rollouts -- it takes much longer than a cubeless = cube-adjusted rollout, which already takes a lot of time.)

2) GNU-bg -- if "cubeful", gnu-bg plays with a live cube but ends each trial when a cube is passed. It then lists the breakdowns based upon its evaluation of what would happen if the game continued. If "cubeless", gnu-bg freezes the cube location but then still plays out the position "according to score". To get the equivalent of SW's cubeless rollout you must make sure that the score is money-game-like (i.e., matchlength = 0). In this latter case the game is played to completion every time, just as happens with SW's cubeless rollout.

My objection is with cube-adjustment with the contention (quite possibly wrong) that both bots handle the cube well enough to trust live-cube rollouts more than the cube-adjustment algorithm. I.e. Don't trust cube-adjustment but rather always use a live cube.

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.