RE: computer-go: A problem with understanding lookahead

["TMCooper" <tmcooper@xxxxxxxxxxxxxxx>]

> -----Original Message-----
> From: owner-computer-go@xxxxxxxxxxxxxxxxx
> [mailto:owner-computer-go@xxxxxxxxxxxxxxxxx]On Behalf Of Vlad Dumitrescu
> Sent: Thursday, January 18, 2001 9:04 PM
> To: computer-go@xxxxxxxxxxxxxxxxx
> Subject: Re: computer-go: A problem with understanding lookahead
>
>
> > Sorry, but I think Vincent's right.  Assuming that the random
> scoring function has a normal distribution about some mean, then
> the best chance of getting a good score at an odd ply is when
> there are many moves available.  Conversely, the best chance of
> the opponent _not_ getting a good score (for the opponent) at an
> even ply is if the opponent has few moves to choose from.  So a
> random searches tends to return moves that result in many choices
> for the program and few for the opponent.
>
> And if the distribution is different? Uniform, for example?
>
> /Vlad
>
The shape of the distribution makes no difference, as long as no single
random
value has finite probability (For finite state computers, each value should
have
a small probability).

The random values are only compared, no arithmetic operations are performed
on them,
so if you cube all random values, you get the same programme behaviour but a
different
PDF.  The cubing can be replaced by the application of any monotonically
increasing
function.

Tom.



The information contained in this e-mail and any subsequent correspondence is private and is intended solely for the intended recipient(s). For those other than the intended recipient any disclosure, copying, distribution, or any action taken or omitted to be taken in reliance on such information is prohibited and may be unlawful.