[gmx-users] Errors with g_dielectric

[Jennifer Casey]

Hello,

I am trying to calculate the dielectric constant for pure tetrahydrofuran
(THF) at 298K.  I keep running into problems though.  I have looked through
the gmx user list to see if others have had these problems, but I didn't see
any mention of them (although I did see that others were asked to report
issues with g_dipoles to bugzilla).

The first thing I do is run g_dipoles using the command (I do this in order
to get the ACF to use in g_dielectric):
g_dipoles -f nvt10ns.xtc -s previous.tpr -corr mol -mu 1.75
*I would have liked to attach the tpr and xtc file, but the message was too
big.  I can send them if they will help*
When I do this, I get the following output to terminal:


There are 255 molecules in the selection
Using volume from topology:  34.3361 nm^3
Last Frame 30000 time 12000.001
t0 0, t 12000, teller 30001
**then there is a long pause (approx 5 minutes)**
Dipole Moment (Debye)
______________________
Average = 1.9194     Std. Dev.  = 0.0085     Error = 0.0000

**Then it lists the different dipole moments, kirkwood factors, and finally
an epsilon = 4.47756**
(I cannot attach the output files since they too make this e-mail too big.
I will send them if requested.)

It seems that the g_dipoles is working fine for me.


Once I have the autocorrelation function (dipcorr.xvg), I want to use
g_dielectric.  Before I talk about the problems I have here, I wanted to
verify a few things about the various options:
epsRF - the default here is 78.5, even though the default in g_dipoles is 0
(infinity).  I wanted it to be infinity, so I assume I change it.
eps0 - this is the epsilon of my liquid - but is it the epsilon that was
calculated from g_dipoles (4.47756)?

When I run the command:
g_dielectric -f dipcorr.xvg -epsRF 0 -d -o -c
I get a segmentation fault before anything happens:
Read data set containing 2 columns and 15001 rows
Assuming (from data) that timestep is 0.4, nxtail = 1249
Creating standard deviation numbers ...
nbegin = 13, x[nbegin] = 5.2, tbegin = 5
Segmentation Fault

If I leave out the -epsRF, I still get the same error.  If I include eps0, I
still get a segmentation fault.  It seems strange to me since GROMACS
generates the input and yet has an issue with it.

I would like to point out that the manual states to use dipcorr.xvg to get
the dielectric constant, but after reading the paper GROMACS references, it
seems that <Mtot^2> is more appropriate. I tried running the command
g_dielectric -f  Mtot.xvg, and the segmentation fault went away.  Instead
lambda went to infinity and there was a fatal error (nparm = 0 in the file
../../../../src/tools/exptfit.
c, line 466.

I am probably missing something obvious, but I am having a hard time
figuring out what it is.  I appreciate any help.

Thank you for your time,
Jenny
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://maillist.sys.kth.se/pipermail/gromacs.org_gmx-users/attachments/20100726/40834a54/attachment.html>