[computer-go] 10k UCT bots

[Don Dailey]

Mark Boon wrote:
>
> On 13-mei-08, at 14:15, Don Dailey wrote:
>
>> Yes, it's not random at all.   The points near the end of the list 
>> are much less likely to be chosen for instance.
>
> OK, I'm not very good at statistics, but I don't see how the last 
> points are "much" less likely to be picked. At best they are a 
> "little" less likely to be picked but I actually don't see that 
> either. I would like to see some logical argument supporting that claim.
I already retracted my statement in a previous message and I agree with you.

In fact, I don't think there is any bias based on where the points are 
in the list,  it has more to do with where the non-usable points are in 
the list.

I think it's pretty complicated too.   Assuming the list is static for 
each game,  you are going to get some ugly clustering effects similar to 
hash tables that use open addressing for conflict resolution.    All of 
the point after an un-usable point has a locally increased chance of 
being selected "soon", depending on how you gather up the empty points.  

I think it may not be so bad if your algorithm is to compress the list 
by moving a random point into the newly occupied slot.   It won't be 
uniformly random, but it's hard for me to predict if this is a serious 
crime or not without thorough testing.

- Don



>
> Mark
>
>
>
> ------------------------------------------------------------------------
>
> _______________________________________________
> computer-go mailing list
> computer-go at computer-go.org
> http://www.computer-go.org/mailman/listinfo/computer-go/