Related to the "NEW-merge-to-MASTER" question I had a while back, I'm
trying to write a generalized way of adding a variable number of rows
(columns eventually, via tranposing):
j=. <array of boxed cells, where each row is a literal cell followed
by numeric cells>
n=. (1{($j))-1 NB. n is the number of data columns AFTER the first
k=. 'a';'b';'c';'d';'e' NB. "toy" example
kk=. n$<0
What I'd like to be able to do is to create "n" rows (based on array
"j") of zero-filled boxes that are connected to list "k" (and then
transpose the resulting array to be used with array "j"). In other
words, I want code that automatically adjusts itself to an associated
array without having to write different versions for every type of
array we currently use. (After all, that's part of the point of
programming, isn't it?) What I've found through experimentation so far
is this:
if n=1: k1=. (k,:kk)
if n=2: k2=. (k,:kk),kk
if n=3: k3=. (k,:kk),(kk,:kk)
if n=4: k4=. (k,:kk),(kk,:kk),kk
if n=5: k5=. (k,:kk),(kk,:kk),(kk,:kk)
if n=6: k6=. (k,:kk),(kk,:kk),(kk,:kk),kk
if n=7: k7=. (k,:kk),(kk,:kk),(kk,:kk),(kk,:kk)
etc.
Although I see the pattern, I can't figure out how to express that
alternating coding pattern in J. Is there any way to express this (or
some other algorithm that achieves the same results) through some sort
of implicit loop or whatever, rather than through an explicit "for."
loop or something?
What I'm looking for is some expression (or explicit verb definition)
that will automatically create "n" rows of boxed zeros connected to
list "k" of boxed data. (Additionally, perhaps a similar expression
could add "n" vectors of zeros to an existing numeric vector.) It
would seem that something like that ought to be possible, right?
Thanks in advance!
Harvey
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
