I think I worded that wrong. I don't want the algorithm re-expressed, I want
an interesting new way of doing it. The code I
posted was thrown together in 20 minutes to elucidate the spec. So, don't
suggest trivial improvements like
scale =. ] % (<0 0) { ]
or
ap =. [: >@:; _2 <@:,@:{\ 2 # ]
cxn_mats =. arcs2mat @: ap
etc. OTOH, thoughts on different representation of the data, like
((* _1 ^ ])&:; {.&1@:#&.>)&> jjss
0 1 _2 _3
0 1 2 _3
0 3 _1 2
0 1 3 _2
0 1 _2 3
are encouraged.
-Dan
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
