Re: [gnugo-devel] arend_1_25.3: less (??) experimental influence

[Arend Bayer]

On Wed, 13 Feb 2002, Daniel Bump wrote:
> 
> It appears that the experimental influence option is definitely stronger
> but now the question is why it is so slow, and whether something
> can be done about this.

A friend of mine played a 40 game-match, guess it was the first he could
make full use of the performance of his recently bought PC. (9 h on
an Athlon 1,4 GHz). 10 results got lost; together with my results
and Dan's match we definitely sufficient statistical evidence.

Arend

Total:
Experimental wins 35 out of 44 games, 29 of them with >10 pts.

Experimental influence as Black:

30 games, Black wins 25 games, 20 of them with >10 pts

Result according to B: B+4.5
Result according to B: B+56.5
Result according to B: W+2.5
Result according to B: B+23.5
Result according to B: B+57.5
Result according to B: B+2.5
Result according to B: B+33.5
Result according to B: B+36.5
Result according to B: W+14.5
Result according to B: B+5.5

RE[B+35.5]
RE[B+14.5]
RE[B+10.5]
RE[B+42.5]
RE[B+24.5]
RE[B+47.5]
RE[B+39.5]
RE[B+20.5]
RE[W+11.5]
RE[B+54.5]
RE[B+47.5]
RE[B+9.5]
RE[W+3.5]
RE[B+19.5]
RE[B+24.5]
RE[B+25.5]
RE[B+20.5]
RE[B+0.5]
RE[W+0.5]
RE[B+46.5]

Experimental influence playing white:
14 games, White won 10, 9 of them with >10 pts

Result according to W: B+21.5
Result according to W: W+10.5
Result according to W: W+24.5
Result according to W: B+4.5


RE[B+0.5]
RE[B+2.5]
RE[W+3.5]
RE[W+13.5]
RE[W+14.5]
RE[W+45.5]
RE[W+24.5]
RE[W+45.5]
RE[W+21.5]
RE[W+16.5]