One issue here is that you "3 2 is missing a parentheses in your tacit
sentence. (Throw in a ] placeholder to capture your y argument.)

But another issue is that the left argument to your explicit verb is rank 4
and you want to split it at rank 3 with a different x& binding for each of
these items.

Maybe try rephrasing this using explicit loops to illustrate how you would
need to build a tacit version? (Personally, I wouldn't bother--it would
work, but you'd introduce so much complexity that I think your efforts
might be better spent elsewhere...)

-- 
Raul

On Monday, May 20, 2019, Brian Schott <[email protected]> wrote:

> I am working with dyadic tessellations using `tesl` below. There are 2
> versions that give different results: "tacit" and "explicit" . The
> "explicit" version works well for me, but I cannot duplicate the results
> with the "tacit" format.
>
> So, I am wondering if anyone can tease the "tacit" format into the desired
> result?
>
> Notice that the shapes of both "filters" and "image" are prefixed with 3's
> and as a result the shape of the result reflects the 3 separate
> tessellations.  But my "tacit" version has an extra 3-shaped dimension in
> resulting shape.
>
>    tesl =: ;._3
>    link =: ;
>    cntrl =: 1 1,:2 2
>    filters =: 10+i.3 2 2 2
>    image =: 3 3 3$'abcdefghi','ABCDEFGHI',|.'abcdefghi'
>    $cntrl filters&(link"2) tesl"3 2 image NB. "tacit"
> 3 2 2 3 2 2
>    $filters(4 : 'cntrl x&(link"2) tesl y')"3 2 image NB. "explicit"
> 3 2 2 2 2
>
>
> (Btw, I have been using a trick to economize on the vertical space needed
> to screen-print the results in my experiments. An alternative to prefixing
> the expressions with "$" as in the expressions below, I have found that the
> prefix ",." can be quite effective.)
>
> Thanks,
>
> --
> (B=)
> ----------------------------------------------------------------------
> For information about J forums see http://www.jsoftware.com/forums.htm
>
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm

Reply via email to