In general, due to rank considerations, I would compare[: u v vs. u @: v, but
the point remains,
5 ts '(,[:+/_2{.])^: (<:`(0 1x"_)) 1111'
0.115203 271232
5 ts '(,+/@:(_2&{.))^: (<:`(0 1x"_)) 1111'
0.0813124 271232
also,
p=. @:]
5 ts '(,(+/p _2&{.))^: (<:`(0 1x"_)) 1111'
0.0802102 271552
Personally, I stay away from [: and its ad-hoc parsing rule but [: has many
fans...
----- Original Message ----
From: Arie Groeneveld <[EMAIL PROTECTED]>
Playing with fibonacci sequences I ran up against the following:
10 ts '(,[:+/_2&{.)^: (<:`(0 1x"_)) 111'
0.0011442 12416
10 ts '(,+/@(_2&{.))^: (<:`(0 1x"_)) 111'
0.0011354 12416
5 ts '(,[:+/_2{.])^: (<:`(0 1x"_)) 1111'
0.1276228 271232
5 ts '(,+/@(_2&{.))^: (<:`(0 1x"_)) 1111'
0.0878322 271232
Does this example suggest it's worth it to reformulate a J-sentence with
caps to one with at(op)'s if we are dealing with large(r) quantities?
=@@i
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
