Here's a better illustration of the issue that J's implicit coercion of extended and rational number to floating point is not arithmetic:
n1=:{.100 0.01
n2=:<.n1
n1 - n2
0
n1 -&(^x:) n2
2.19254e186
Here, we have two numbers which are numerically equivalent,
but applying the same function to them gives very different results.
--
Raul
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
