🚀 go-pugleaf

On Jan 31, 11:03 am, Nico_Schlöm...@adobeforums.com wrote:
> hi!
>
> i'm dealing with matrices here and want to plot its value into postscript. in fact, i just need a black spot where a matrix entry is not zero, white everywhere else. easy, isn't it?
>
> particulary, my matrices have very few non-zero entries, so the ps file would only have a few black spots here and there.
>
> now, what's the best way to write this in postscript? the first thing i tried was the image environment: works well as soon as you got used to the syntax, but the big disadvantage for me was that i have to explicitly code all the white pixels, so the eps file gets a lot bigger than needed (it feels).
>
> the second thing i tried was defining a small "draw tiny box" command, then "move to" the non-zero spots, and evoke the function.
> that's better then the image way (the files are a lot smaller), but it still feels like a hack somehow.
>
> do you have any hints or suggestions for the issue? are there any commands or environment that would be better suited?
>
> all hints appreciated!
>
> cheers,
> nico

PostScript is, in some ways, a perfect language for sparse matrices.
You can simply use a dictionary.  A dictionary key could be anything,
including a number.  So you could store your matrix elements with
something like:

/mysparsemtx 100 dict def
/key x 1024 mul y add def
mysparsemtx key 1 put

Now wherever you have a "1" stored, you can treat that as a black
spot.  You can traverse the whole matrix with

mysparsemtx {
  pop % You can throw away the "1" since every entry has a "1" stored
  dup 1024 mod /y exch def 1024 div /x exch def  % unpack the original
x and y
  % do whatever you want with x and y
} forall

NOTE: I haven't tested this code, but this is the gist of the
approach.

-pd