Lightly tested and non-optimized and ugly, but seems to work!
genseq=:({: + i.@:>:@:-/)@]
seq =: (] (>/@[ \:~@]^:[ genseq) \:~) f.
] (>/@[ \:~@]^:[ ({: + i.@:>:@:-/)@]) \:~
seq 3 7 3 4 5 6 7 seq 7 3 7 6 5 4 3
On Fri, Apr 6, 2018 at 3:23 PM, Devon McCormick <devon...@gmail.com> wrote:
> Hi -
>
> I'm working with selecting sequential items from a vector given a starting
> and an ending point as a pair of integers. If we assume we always get the
> pair as (lesser, greater), something like this suffices:
>
> seq=: {. + [: i. [: >: -~/
> seq 3 7
> 3 4 5 6 7
>
> However, if our assumption about the order of our points is violated, we
> get nonsense:
>
> seq 7 3
> 7 8 9
>
> Just for robustness, how could we write a version of "seq" that would
> return the sequence in reverse order in this latter case, i.e. give the
> proper sequence from seven to three: 7 6 5 4 3 ?
>
> I generalized the "seq" verb to remove the ordering assumption: my idea was
> to generate the ascending sequence, then flip it if >/y is true. I tried
> this with "agenda" but am not sure how to get it to work on one thing - the
> ascending vector - on the basis of comparison of another thing - the pair
> of integers.
>
> I did achieve this using the "power" conjunction:
>
> seq=: 3 : '|.^:(>/y)](<./ + [: i. [: >: >./ - <./) y'
> seq 3 7
> 3 4 5 6 7
> seq 7 3
> 7 6 5 4 3
>
> Does anyone have any ideas for a more elegant, preferably tacit, solution?
>
> Thanks,
>
> Devon
> --
>
> Devon McCormick, CFA
>
> Quantitative Consultant
> ----------------------------------------------------------------------
> For information about J forums see http://www.jsoftware.com/forums.htm
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
