Re: [gnugo-devel] new vital attack pattern (revised)

[Martin Holters]

> > > Regarding pattern constraint: seems good, apart from
> > > `owl_maxeye(*)==1' part, which is too restrictive.  Consider e.g. this
> > > position:
> >
> > [...]
> >
> > Dropping that constraint altogether substantially increases the node
> > counts and moreover, produces more new failures than successes in the
> > regression tests. I couldn't pinpoint why some tests broke, but it
> > looked like GnuGO was dropping back to the exact same (wrong) moves it
> > made in the games on which the tests were based. So my _guess_ is that
> > this pattern without the maxeye-constraint matches too often and
> > prevents the correct moves from being tried.
> 
> Maybe you should use `owl_maxeye(*) >= 1' or `owl_mineye(*) >= 1' or
> `owl_proper_eye(*)'.  It would mean to play only in actual eyespaces.
> Most likely, the expenses you faced after dropping this part of
> constraint were caused by playing outside any eye, though i'm not really
> sure.

Yes, of course. I was on the wrong track there. I've changed it to
owl_proper_eye which resulted in a very minor increase in node counts,
but no regression failures (alas, no new passes as well). As this seems
more reasonable despite showing no direct effects, I've left it in my
revised patch.

The other problem is what Evan pointed, e.g.

|OOOOO
|OO.OO
|OOXOO
|XX.XO
|.X.XO
|.*XOO
|.OXO.
+-----

This wouldn't be matched with

?X?
X.X
*XO

but the respective move would still work. The alternative is to leave
the upper line out

X.X
*XO

and use a constraint. But after some experimenting, I have come to the
conclusion that in fact, constraining to one-space-eyes does not yield
any real advantages. So my new proposal is somewhat simplified, but
still the increase in node counts compared to the unpatched version is
negligible.

Regression delta:

ld_owl:413      PASS 0 [0]
ld_owl:414      PASS 1 S2 [1 S2]

(as before)


Index: patterns/owl_vital_apats.db
===================================================================
RCS file: /cvsroot/gnugo/gnugo/patterns/owl_vital_apats.db,v
retrieving revision 1.42
diff -u -r1.42 owl_vital_apats.db
--- patterns/owl_vital_apats.db 7 Jan 2004 10:00:20 -0000       1.42
+++ patterns/owl_vital_apats.db 15 Jan 2004 23:27:23 -0000
@@ -921,4 +921,19 @@
 ;lib(A)==2
 
 
+Pattern VA54
+# mh New pattern (3.5.4)
+# See ld_owl:413/414
+
+X.X    play atari inside one eye to destroy the other eye
+*XO
+
+:8,s,value(45)
+
+X.X
+*AO
+
+;lib(A)==2 && olib(*)>=2 && owl_proper_eye(*)
+
+
 # END OF FILE