OK - the problem appears to be fixed. Thanks to everyone! At a colleagues'
suggestion, I "typed" Matlab's intrinsic erfc. (This allows you to see the source
code if it's not compiled.) It turns out it is a translation of a Fortran 77 code in
netlib. I got that code, struggled trying to remember F77's annoying quirks (_seven_
columns... :-) ) and finally got the code to run on my platform. I then created a
module, put in the interface block for the f77 code, and now am finding erfc
correctly.
In retrospect, a great deal of my initial frustration was not having touched an F77
code in a long time. The error messages you get when you disobey fixed format rules
can be very strange if you've forgotten this quirk, and are not at all enlightening
at first.
Alvaro Fernandez
PS
Thanks to John Venier for coding up r1mach, and to Alan Miller for his derfc.f90,
even though I ended up not using their work. I prefer to encapsulate old code
whenever possible, rather than translate it.
A. F.
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|