Donna:

Let me try to summarize some of this discussion and suggest a direction to
move forward.

Any computer can represent only finitely many numbers.  Whether they be
integer, floating point, or rational, there are infinitely many quantities
than cannot be represented exactly.  There is nothing special about J in
this regard: any language has similar expressiveness and limitations,
although the details will vary.  As a side issue, a programmer can
represent a rational number as a numerator/denominator pair without
canonicalization: this is an implementation decision, and has nothing to
do with the language.   Questions about representations of numbers can be
answered through reference to any familiar language.

There are techniques that can be used to obtain accurate results despite
these limitations: this is what numerical analysis is about.

If you are interested in linear programming (I'm not sure if you are, or
whether you were just citing an example), you might want to take a look at

http://www.jsoftware.com/jwiki/Stories/RichardBrown

which contains an annotated description of using the simplex algorithm in J.

More generally, you will probably learn more from trying a solve a
familiar problem in J than from any amount of discussion here.  Otherwise
it is like trying to understand swimming by talking about it rather than
getting your feet wet.

Best wishes,

John

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

Reply via email to