Hello, hopefully someone has successfully done this:

I looking for help on converting lubksb and ludcmp to work with complex inputs in the f90 version of NR. I not quite sure what goes wrong, but at least it seems to me that just doing the obvious changes in the routines gives nonsense as output.

I have successfully implemented solution of complex systems of equations as a coupled set of real equations, but the 8x overhead is getting problematic for me ... also the roundoff error is probably larger that way.

Any help would be appreciated - If needed, I can attach here my suggested conversions of lubksb and ludcmp.

Regards,

Timo