It isn't a question - just an observation.
In the context that I wrote it- I have considered a model of a
situation where this can come up. In some parts of an array a 0%0 term
shows up- but only in situations where the correct result is 0.
Convenient in that I don't need to set these terms to 0
This is related to the increment of magnetic flux density at one part of
a current loop due to an Idl increment elsewhere in the loop. This can
easily be found for all points in the loop.-including the self terms
which should be 0. 0%0 being 0 in this case is quite applicable.
Otherwise I would have to set these points to 0 by changing the
diagonal of an array to zeros.
OK?
Don Kelly.
On 20/05/2013 7:17 AM, Raul Miller wrote:
I don't understand this question.
Can you restate it?
In other words are you asking whether floating point numbers are valid
in your code? Or are you asking why someone would divide epsilon by
zero? Or are you trying to express a problem? Or is it something
else?
That said, my generic reaction is that if you have practical issues
where the distinction between 0%0 and 1e_256%0 matters that probably
means your model is unstable.
Thanks,
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
