coloin wrote:If the suexrat9 works essentially by solving 100 isomorphs of the puzzle and averaging the time taken - I was wondering if it should be based on the time taken for the "easist" 25% ????
one nit -- its based on node counts rather than time
although time vs counts might correlate on one machine, they would not across different machines
node counts are only subject to the pesudo-random search order
BTW what is the command line of gsf's SX version
the exrate(n) function, n is the number of random search trials, default 100 if omitted
note that its suexrat9-like, but not exact, because exrate() is not DLX based, suexrat9 is
in -e filtering:
- Code: Select all
-e 'exrate()>2000'
or in -f output:
- Code: Select all
-f'%(extrate())x'
The "pencilmarks"
The GN puzzle has no bi-value cells and a few bi-locus boxes [I dont think a puzzle can be made without these ?].
nvalue/location is called degree in my solver
a degree 2 candidate is either part of a bivalue cell or a bilocation edge
this lists the total number of candidates (pencilmarks), the number of clues, and the number
of candidates or degree 2, 3, etc., separated by -
- Code: Select all
-f'%#d#-P
e.g., for the golden nugget
- Code: Select all
257-21-9-67-81-64-11-3-1
and this lists the tuples of degree 2 .. max degree
- Code: Select all
-f'%#tX
e.g., for the golden nugget
- Code: Select all
[2] [41]=2 [63]=2
[2] [74]=2 [96]=2
[2] [14]=3 [16]=3
[2] [42]=4 [53]=4
[2] [18]=6 [38]=6
[2] [25]=8 [65]=8
[2] [54]=8 [65]=8
[2] [81]=8 [92]=8
[2] [51]=9 [62]=9
...
[8] [11]=6 [12]=6 [13]=6 [21]=6 [22]=6 [23]=6 [31]=6 [32]=6
(the 9 tuples of degree 2 and the 1 tuple of degree 8, others omitted)
