[Computer-go] Operators for Frisbee Go Simulation

"Ingo Althöfer" 3-Hirn-Verlag at gmx.de
Thu Apr 14 06:06:45 PDT 2016


Hi Marc,

thanks for the infos. I try to give answers directly for each point.
(The problem is I have a stubborn cough, for weeks already. Each
day getting my work done is a challenge...)

> Goncalos were on 7th of April. Just copying them here:
> ---
> On frisbee Go itself I used the following definition:
> 1. An intended play must be legal -- no playing on top of a stone hoping
> it 'falls' to the neighbor positions.

Accepted.

> 2. Unintentional plays that are illegal are nulled and don't imply a
> desire to end the match.

Accepted.

> 3. The distribution of unintentional plays around the 4 neighbors is
> constant even at the border where there are never 4 neighbors; "You hit
> the target with prob. p, and its 4 neighbours with probability
> (1-p)/4.". The residual probability at the border is not reused for
> on-board plays.

Accepted.

> 4. Probability parameter p cannot be changed midgame, for simplification.

Of course. 
For the tournament I see only two natural choices for (1-p)/4, namely
= 1/6 or 1/8. For both cases I have appropriate dice for manual rollouts.

Likely the number of participants will notz be too large. So we might
play double round robin with 1/6 in one of the runs, and 1/8 in the other.

> 5. Technically, using the GTP, I assumed genmove_reg+play commands are
> used, instead of genmove+undo+play or something frisbee specific. This
> is probably stating the obvious.

No GTP please, but manual operation of the bots.


> Also when trying to theoretically solve even very simple L+D
> situations this game feels quite tricky for me and very different from
> Go. 

Of course. The specialties of Frisbee Go are most obvious in the very
final moves.


> Btw, for the theoretical question posed earlier (regarding the 3
> point eye) I assumed the p to be the chance of deviation. 

Right. That is the more natural thing to view things.

Cheers, Ingo.



More information about the Computer-go mailing list