Handicap handling

[dsmic]

I think, this would be nice for the next kgs runs. I plan to run with 10k playouts to compare to last years results, and handicap makes it nicer for KGS.

I would have a look at this, just to avoid that we are both inwesting time.

[Francois van Niekerk]

I like the idea and have been planning on doing it for some time, but just never get the time.

If you haven't already thought of it, you should probably consider using dynamic komi to compensate for handicap stones.

Please keep me updated, I am curious about the progress.

[dsmic]

• initial support handicap stones

• missing: do not count the handicap stones for own area

• missing: dynamic komi

https://bitbucket.org/dsmic/oakfoam/commits/cbd0c9bbe5a9e165e40f292de2820e2a38989f1f

[dsmic]

next commit handles handicap stones counting.

Do you know of a sucessfull formula to calculate dynamic komi?

[dsmic]

I pushed minor changes. This version is running on kgs with 9600 playouts now.

[Francois van Niekerk]

The progress sounds good.

For dynamic komi, you probably want to look at "Balancing MCTS by dynamically adjusting komi value" by Petr Baudiš

[Francois van Niekerk]

This has been resolved