gnugo-devel
[Top][All Lists]
Advanced

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

Re: [gnugo-devel] Eye pattern 304


From: Gunnar Farneback
Subject: Re: [gnugo-devel] Eye pattern 304
Date: Mon, 19 May 2003 21:17:19 +0200
User-agent: EMH/1.14.1 SEMI/1.14.3 (Ushinoya) FLIM/1.14.2 (Yagi-Nishiguchi) APEL/10.3 Emacs/20.7 (sparc-sun-solaris2.7) (with unibyte mode)

Paul wrote:
> i came up with inessential marginals to avoid revising pattern databases.
> e.g. in the SP Lee's position we had the following:
> 
>   ...O.
>   XXXOX
>   X....
> 
> which is considered as
> 
>   ..!
> 
> by optics code after domain building. when i made the topological analyser
> understand that the neighbor vertex is marginal too (as a false eye) i got
> this eye:
> 
>   .!!
> 

There seems to be more difficulties of this kind. Appended is a patch
adding the new test case owl1:313. Here the topological analysis finds
that D18 is a false eye, but add_false_eye() doesn't see that it
also should be removed from the eyespace:

Eyespace at A19: color=white border, esize=7, msize=2
A19
B19 (X)
C19
D19 (!)
A18
D18 (!)
A17

.X.!
.  !
.   

  guess_eye - 1111, pessimistic_min=0

With D18 removed, eye pattern 6003 would be matched, finding the vital
point at A18.

> since we don't allow such eyespaces it didn't match. i decided the easiest
> (and maybe the most logical) way would be to just ignore the right marginal,
> matching
> 
>   .!

I agree that it's correct to remove the superfluous margin from the
eyespace, but in light of the failure described above I think we
should make a slightly more drastic change to the code. Instead of
using inessential margins and patching up add_false_eye() to modify
the eyespace in various situations, we should simply rebuild the
eyespaces from scratch after the topological analysis. I believe this
will make the code both simpler and more robust.

/Gunnar

Index: regression/owl1.tst
===================================================================
RCS file: /cvsroot/gnugo/gnugo/regression/owl1.tst,v
retrieving revision 1.59
diff -u -r1.59 owl1.tst
--- regression/owl1.tst 13 May 2003 03:22:53 -0000      1.59
+++ regression/owl1.tst 19 May 2003 19:15:01 -0000
@@ -208,6 +208,10 @@
 312 owl_defend K6
 #? [1 C3]*
 
+loadsgf games/owl45.sgf
+313 owl_attack B18
+#? [1 (A18|C19)]
+
 
 ########### end of tests #####################
 
Index: regression/games/owl45.sgf
===================================================================
RCS file: regression/games/owl45.sgf
diff -N regression/games/owl45.sgf
--- /dev/null   1 Jan 1970 00:00:00 -0000
+++ regression/games/owl45.sgf  19 May 2003 19:15:02 -0000
@@ -0,0 +1,24 @@
+(;GM[1]FF[4]
+SZ[19]
+GN[GNU Go 3.3.19 load and print]
+DT[2003-05-19]
+KM[-4.0]AP[GNU Go 3.3.19]RU[Japanese]
+;AW[ja][ka][la][ma][bb][cb][eb][hb][ib][lb][bc][dc][ic][kc][mc]
+[ad][bd][dd][hd][jd][ld][md][nd][od][sd][fe][ge][he][je][ke][le]
+[me][pe][re][se][gf][kf][of][pf][qf][sf][dg][gg][hg][ig][kg][qg]
+[rg][dh][hh][jh][oh][rh][di][ei][hi][ji][oi][pi][qi][cj][dj][ej]
+[gj][hj][ij][lj][mj][oj][qj][ak][bk][ck][gk][hk][lk][mk][nk][ok]
+[qk][cl][hl][il][ml][nl][ql][rl][dm][em][im][lm][mm][rm][sm][cn]
+[en][fn][jn][kn][ln][mn][nn][sn][co][eo][go][jo][ko][mo][cp][ep]
+[fp][gp][ip][mp][np][op][dq][gq][hq][iq][lq][nq][dr][er][gr][nr]
+[or][pr][es][ps][qs]
+AB[ba][ea][ga][ha][ia][na][fb][gb][mb][nb][cc][ec][fc][gc][hc]
+[nc][oc][pc][qc][rc][sc][cd][fd][id][pd][qd][rd][be][ce][de][ee]
+[ie][ne][oe][qe][df][ff][hf][if][lf][mf][nf][cg][fg][lg][ng][og]
+[pg][bh][ch][fh][gh][kh][mh][ph][qh][ai][ci][fi][gi][ki][li][mi]
+[ni][aj][bj][fj][jj][kj][nj][pj][fk][ik][jk][kk][pk][dl][el][fl]
+[jl][kl][ll][ol][pl][fm][gm][hm][jm][nm][om][pm][gn][hn][in][on]
+[pn][qn][rn][do][ho][io][lo][oo][so][dp][hp][jp][kp][lp][pp][qp]
+[sp][eq][fq][jq][mq][oq][pq][qq][rq][fr][hr][ir][kr][lr][mr][qr]
+[sr][fs][gs][hs][is][ks][ms][ns][rs]
+PL[B])




reply via email to

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