Re: [gnugo-devel] Connection point not seen

[Arend Bayer]

On Wed, 4 Jun 2003, Arend Bayer wrote:

> On Tue, 3 Jun 2003, SP LEE wrote:
>
> >
> > Thanks for your comment. I'm afraid I'm not being able to touch the 
> > patterns, it seems to be difficult for me. How can I use Arend's escape 
> > rewrite?
>
> Unfortunately, it's not useable at the moment, as the patch is a bit
> out-of-date with the changes in readconnect.c.

Actually, while --experimental-break-in does not solve the owl problem,
it does successfully work around it. This seems to be a very nice side
effect of this new code:

   A B C D E F G H J K L M N O P Q R S T
19 . . . . . . . . . . . . . . . . . . . 19
18 . . . . . . . . . . X . X . . . . . . 18
17 . X X . . X . . . X O X O O X . . . . 17
16 . O X X . . . X . X O . O X X X . . . 16
15 . X . O . O . . . X . O O O . . . . . 15
14 . O O . . . . O . O O . X . X X . . . 14
13 . . . . . . . . . . . X . O O X . X . 13
12 . . . O O . . . X . . . . O X . O . . 12
11 . X X X . . O O X . . X . O . . . . . 11     WHITE (O) has captured 2 stones
10 X O O X . O X O O X . . . . . X . X . 10     BLACK (X) has captured 1 stones
 9 . O X X . . X O X X X . . . O . X . X 9
 8 . O X X O . X X O O . . O O X . O X . 8
 7 . O O O O . O O . O . O X X O O . O . 7
 6 . X O O X O . X O . . X X O X X O . . 6
 5 . X X X X X X . . . . . . . . X . . . 5
 4 . . . X O X O X X X . . X . . X O O . 4
 3 . . X X O O . O X O O . . X X O X O . 3
 2 . O X O . . O O O X X . . X O O X O . 2
 1 . . O . O X . . . . . . . . . . . . . 1
   A B C D E F G H J K L M N O P Q R S T

The problem here was that the K10 dragon is considered dead, because owl
does not find the escape move at O14. However, with --experimental-break-in
(well, if one applies a bug fix that I will post in a minute) the
valuation goes as follows:

The K10 dragon is considered dead, and so the whole area around L12 (up
to N14) is considered territory for white. However, after black plays
O14, this isn't changed at first. But then
block_off(O14, huge white territory) finds out that black cannot be
stopped from breaking in (as O14 is already connected to K10 etc.), and
hence marks this as non-territory for white:

  O14: 38.00 - change in territory
  O14:   -9.00 - total followup value, added -9.00 as territorial
followup
  O14: -4.50 - added due to followup (-9.00) and reverse followup values
(0.00)
  O14: 0.52 - connects strings (connect value 5, shape factor 0.104)
Move generation values O14 to 34.02
(...)
Move generation likes O14 with value 34.03
genmove() recommends O14 with value 34.03

(I will have to fix the negative followup value.)

I really hope that I can fix the performance issues so that we can
enable this code for 3.4.

Arend