On Tue, Nov 18, 2008 at 1:59 PM, Roger Hui <[EMAIL PROTECTED]> wrote:
> start1=: 0 1 1 ,: 0 0 1
> step1=: ,. (-/,+/)@:({:"1) + 1 _1 * |.@:(|[EMAIL PROTECTED]:"1)
> plot ;/ step1^:12 start1

I was thinking about transposing the data,
like that, but decided that the other approach
looked simpler.

I had not thought about re-organizing the rotation
calculations, but I am not sure if the 4x speedup
is worthwhile.  For now, at least, I feel that my
dumber approach is easier to understand (and,
perhaps more usefully generic -- for example,
replacing the rotation matrix with an alternative
yields a rather broad range of related fractals).

-- 
Raul
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm

Reply via email to