[gmx-users] g_dielectric problem

[Guilherme Duarte Ramos Matos]

Dear GROMACS users community,

I'm working with water solutions of small organic molecules and I am
interested in their dielectric spectra. After submitting the trajectory
file to g_dipoles and getting the dipcorr.xvg file, I'm facing some issues
when using g_dielectric. I've been using:

g_dielectric -f dipcorr.xvg -ffn exp_exp

the program runs but I get this worrisome message:


"Read data set containing 2 colums and 12501 rows
Assuming (from data) that timestep is 0.1, nxtail = 5000
Creating standard deviation numbers ...
nbegin = 50, x[nbegin] = 5, tbegin = 5
Step       chi^2      Lambda          A1          A2          A3         A4
irow = 2, icol = 2
         nan         nan         nan
         nan       0.000         nan
         nan         nan         nan
Numerical Recipes run-time error...
GAUSSJ: Singular Matrix-2
Fit failed!
DATA INTEGRAL:   3.5, tauD(old) =   5.2 ps, tau_slope =   0.0, tau_slope,D
=   0.0 ps
tau_D from tau1 = 1.77e-26 , eps(Infty) =   80.000
nbegin = 50, x[nbegin] = 5, tbegin = 5
Making smooth transition from 50 through 53

SMOOTH integral = 3.61438e+00
FIT INTEGRAL (tau_M):   3.6, tau_D =   5.4

Doing FFT of 32768 points

MAXEPS = 5.18410e+01 at frequency 3.60107e+01 GHz (tauD =      4.4 ps)"


Can someone help me understand what should I do if I want g_dielectric to
do this fit for me? The other problem I'm facing is that the dielectric
spectrum looks like a NMR FID instead of the usual shape. What did I miss
there?

I tried to attach dipcorr.xvg, derive.xvg and epsw.xvg but unfortunately
the email was too big.

Sorry for the inconvenience


~ Guilherme

*****************************************************
Guilherme D. R. Matos
Graduate Student at UC Irvine

*****************************************************