gnugo-devel
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

[gnugo-devel] A thrashing dragon


From: bump
Subject: [gnugo-devel] A thrashing dragon
Date: Wed, 5 Feb 2003 07:10:37 -0800

I added a test in trash.tst (the last one) that looks quite
interesting. Look at the position at move 204 in the game
uno-gnugo-3.3.16-200302050206.sgf.

There is a semeai going on at the bottom but GNU doesn't
recognize it. GNU believes that the W dragon at M8 is
dead and therefore the B dragon at L8 is alive.

The reason for this is that after B fills one liberty
at M5 during do_owl_attack, the M8 string has four liberties
while the L8 string has eight. The M8 string is classified
as a lunch and is considered not owl lively.

Yet the M8 string has a lot of life in it. It takes 7 moves 
to take it off the board and there is a ko to fight.

At stackp==0, the M8 string has 5 liberties and no
attack point is found though some attack threat points
are given. The owl_attack status is DEAD and the owl_attack
point is PASS.

If only GNU recognized the semeai,  --decide-semeai L8/M8 -o vars.sgf
finds:

After black at O4, L8 is ALIVE, M8 is DEAD (11 nodes)
After white at H8, L8 is DEAD, M8 is ALIVE (12 nodes)

with an acceptable variation set.

Dan





reply via email to

[Prev in Thread] Current Thread [Next in Thread]