Re: [gnugo-devel] new vital attack pattern

[Martin Holters]

ons 2004-01-14 klockan 21.43 skrev Paul Pogonyshev:
> Martin wrote:
> > The following patch fixes ld_owl:413 and ld_owl:414. Effect on the node
> > counts is negligible.
> 
> Are there any more changes in regressions?

No, sorry, forgot to mention it, but these two are the only changes.

> 
> >
> >  ;lib(A)==2
> >
> > +Pattern VA54
> > +
> > +X.X     play atari inside one eye to destroy the other eye
> > +*XO
> > +
> > +:8,s,value(15)
> > +
> > +XaX
> > +*BO
> > +
> > +;lib(B)==2 && olib(*)>=2 && owl_eye_size(a)==1 && owl_maxeye(*)==1
> >
> >  # END OF FILE
> 
> The very low value looks artificial, since the move is very logical.
> I assume that you used so low value, because of positions like this,
> where the move doesn't work as intended:
[...]

No, admittedly, choosing a proper value is a black art to me. I just
lowered the value and realised that this reduced the node counts in the
regression suite while it still fixed the two tests I was aiming at. So
I left it in place.

> Maybe you should instead provide more context in the pattern diagram,
> like this (correct me if it doesn't work for all cases):
> 
>       Pattern VA54
> 
>       ?X?
>       X.X     play atari inside one eye to destroy the other eye
>       *XO

Ah, yes, this indeed looks promising.

> 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. 

> And two little bits regarding decorations.  It is a common practice to
> add short "change log" in pattern comments (in this case, just a
> mention of who and when created the file, see e.g. VA53).  

Ok, err... Should the version indicate the version I'm basing the patch
or the one I hope it to be included in (i.e. 3.5.3 or 3.5.4)?


> And please,
> do keep two empty lines between patterns, as they help to visually
> distinguish different patterns.  You can see that this convention is
> used consistently in all our databases.

Right, sorry.


> If you use Emacs, you might be interested in Emacs major mode for
> editing pattern databases.

Too bad, I prefer vim.


I have changed the pattern to the "more context version" you have
proposed and increased the value to 45 (just because it is so commonly
used - it's still a black art to me). Regression tests are running and
looking good up to now; I'll post the revised patch once the tests are
through and the results are ok.

// Martin