As Philip Viton noted in the General forum, D. gives strange results. 
This is  true even for rank 0 functions.

   f=:%
   (f d. (i.6)) 1
1 _1 2 _6 24 _120
   (f D. (i.6)) 1
1 _1 1.9984 111022 _3.33067e12 7.77156e19

This is not a new problem: the same occurs with J504.  I assume there is
some instability in the numerical differentiation algorithm.

Best wishes,

John

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

Reply via email to