[gmx-users] g_tcaf use issue

Jones de Andrade johannesrs at gmail.com
Wed Nov 11 03:16:33 CET 2015


We are using gromacs in order to explore how well/not so well different
methods to calculate dynamical properties of liquids perform.

At this moment, we are using g_tcaf (already seen g_energy -visc results)
to evaluate how it would perform in a homogeneous mixture of two organic
molecules (one with 5 atoms, the other with 19 atoms).

We are aware of possible convergence and noise issues, so the simulation
ran for 25 million steps, saved at each 500 (full .trr file). We chose a
long run + sparse frame recording combination due to disk space limitations.

First thing we noticed, is that -mol flag will lead to a segmentation
fault. Anyone has any idea on why this?

Second, without -mol flag, we get results (we tried both with and without
the -k34 flag), but we are having some difficulties in sorting them out in
the output files.

For instance, the *tcaf_all* file show 64 different curves for calculations
performed without -k34 flag, and 96 with it.I find hard to relate those
numbers to 3 or 5 (planes), 2 (molecules), 5, 19, 24 (atoms).

Same problems with the *tcaf_fit* files: 16 curves without -k34 flag, and
24 with it.

The *visc_k* files seems to be more clear: without -k34 flag it yields 4
"groups of data" (3 planes + 1), while with it I get 6 (5 + 1). Make sense,
but then again, it seems to "automagicaly" divide the whole simulation in
time intervals, over which I don't have any control: is that correct? How
does it choose how to divide it?

Thanks a lot in advance for anything,


More information about the gromacs.org_gmx-users mailing list