On Thu, Aug 23, 2012 at 5:07 AM, Skip Cave <[email protected]> wrote:
> I would like to generate a vector sequence of numbers starting from s and
> going to e, with increment i
...
> what would the tacit form of the sequence verb look like, if we defined the
> dyadic verb seq thus:
> (s,e) seq i
> here is an explicit function for that:
> s + i * i. 1 + (e-s) % i
Note that if we do not bother with the increment, a vector going from
s to e could be:
(+ i.@>:)/ -~/\ s,e
(+ i.@>:)/ -~/\ 5 8
5 6 7 8
or maybe a variation such as
0 1 (+ i.)/@:+ -~/\ s,e
0 1 (+ i.)/@:+ -~/\ 5 8
5 6 7 8
But that doesn't give us any really great mechanisms for inserting the
"increment". We could use
(+ i.@>:&.(%&i))/ -~/\ s,e
(+ i.@>:&.(%&3))/ -~/\ 5 8
5 8
but I do not know of any good ways to make i be a tacit argument for that case.
We could compute this more directly if instead of s,e,i we had s,i,1+(e-s)%i)
+`(*i.)/s,i,1+(e-s)%i
+`(*i.)/5 3 2
5 8
So if we can figure out how to build the list s,i,1+(e-s)%i we're
done. One approach here could be:
(s,e) ([: +`(*i.)/ ,&{., 1 - %&(-/)~) i
5 8 ([: +`(*i.)/ ,&{., 1 - %&(-/)~) 3
5 8
That said, note that most of the work here has to do with dealing with
the arguments in the form you specified. And, also, note that you
have a dependency between s,e and i -- if (e-s)%i is not an integer
you get an error.
So, personally, I would ask why you need to be doing this this way --
what are you really trying to accomplish?
Thanks,
--
Raul
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
