Hi,
The version I know from
https://github.com/kevinlawler/kona/wiki/Idioms
is
mss=: [: >./ (0 >. +)/\.
It fails on lists of length 0 or 1 but is very concise; the following works on
all list lengths but has a few extra characters:
mss1=: [: >./ [: (0 >. +)/\. ,&0
which is an almost literal translation of the K version.
It basically takes the maximum over i of subsequence sums starting at i (the K
version does the same with subsequences sums ending at i).
I believe this is the same as Kadane's algorithm (but don't take my word for
it).
While very cute, it doesn't include starting index and length, and things get a
bit uglier if you want those as well:
mssi=: [: ((] , {~) [: (i. >./) {:"1) [: 0"0^:(0 >: {:)@:+/\. 1&,. , 0: NB.
returns starting index, length, and value
While mssi performs the same steps as mss1, keeping track of indices defeats
the beauty of tacit J code and IMO is better done with explicit verbs.
Perhaps someone can find a prettier tacit version of mssi?
Cheers,
Louis
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
