[gmx-users] question about tabulated force field

[Liu, Y.]

Hi there,
       I have a question about tabulated force field: what is the
difference when the tabulated potential and build-in potential are used in
Gromacs simulations? why they give different results?
       I simulate a DPPC membrane with Gromos force field. I use standard
gromos input parameters in the simulation, but the area per lipid (0.630
(0.008) nm2) is different from when I compute it with tabulated force
field(0.613 (0.007)nm2). This difference also exist when I use MARTINI
force field.
       I have use gmx dump -s *.tpr > comparison.xvg in two cases(build-in
and tabulated potential) to make sure that the only difference in two
scenarios is the tabulated potential. Here is the difference:
<    coulombtype                    = Reaction-Field
---
>    coulombtype                    = User
40,41c40,41
<    epsilon-rf                     = 62
<    vdw-type                       = Cut-off
---
>    epsilon-rf                     = 1
>    vdw-type                       = User
164c164
<    tau-t:         0.1         0.1         0.1
---
>    tau-t:        0.01         0.1           1
169c169
<    energygrp-flags[  0]: 0 1
---
>    energygrp-flags[  0]: 2 1
       Then you may think my tabulated potential is problematic. However, I
have used Gromacs 4.* with -debug flag to output the tabulated potential.
Therefore, the tabulated potential should be correct.
       In the output .edr file, I also found that the tabulated potential
simulation cannot offer correct temperature for lipids as indicated in the
thermal stat. At least, with the same parameters, it works well in the
build-in potential simulation.
       I do not understand where is the problem and what is the difference
between the tabulated potential and build-in potential in Gromacs. Maybe it
use different engine to compute the force?
        Looking forward to your reply. Thank you.
regards
Liuyang