[gmx-users] Normal Mode Analysis in Double-Precision

[Ran Friedman]

Inon Sharony wrote:
> This is exactly the point - I get only three significant digits.
> Although I performed the entire procedure in double-precision. The
> convergence to 5E-4 DID occur in single-precision, I just thought it
> wasn't good enough (the program recommends 1E-5 and when
> double-precision works for me I got 1E-6 with the files you sent me).
What did you have in the input?
In any case if you get 6 zero frequencies and the rest are positive
you're probably fine.
> I also wouldn't mind a clearer explanation of how you fixed bugzilla
> 168 (see my request at
> http://www.gromacs.org/pipermail/gmx-users/2008-August/035543.html).
> Which lines did you add, and which did you change to comments in
> eigio.c ?
The message from Tsjerk seems pretty clear. You can uncomment lines 110,
111 and 112 in eigio.c and recompile g_anaeig:
>make g_anaeig

If everything went fine, you'll get a new executable for g_anaeig, which
you can then copy to where you had the original version.

Ran.