Topic: big numbers

What is the recommended approach to handling big numbers. I need to evaluate alternating series (that don't simplify) containing terms of magnitude 1e25 but with resultants near 1 and a required precision of 1e-4. Suggestions please.

Re: big numbers

While you could try using REAL(KIND=16) quad precision, you'll be right on the edge of its precision.  I would alternatively suggest using a multi-precision library that's simple, like FMLIB that should allow arbitrary precision.  The link above leads to the project page, and you should be able to simply add the necessary files to a Simply Fortran project.  However, I'm having some difficulty downloading the source files due to an issue with line endings...  I'll contact the author.

Jeff Armstrong
Approximatrix, LLC