[gmx-users] Re: Pressure and All-bonds

[Dimitris Dellis]

Problem resolved.
All-Angles result in over-constrain if the number bonds+angles 
potentials are more than the number
of constraints needed to keep molecule rigid.

System configuration becomes semi-freezed - large clusters and a lot of 
empty space.
Using any thermostat it gives the "correct" temperature and almost 
double pressure, while
in NVE, it gives almost double temperature and "correct" pressure, for 
my case
(18 bond+angle potentials, 15 constraints needed).

D.



Dimitris Dellis wrote:
> Hi All.
> I try to use different potential models with an Oh symmetry molecule.
> I created a topol with nonbond_params, bonds and angles sections, no 
> include etc.
> Initial configuration comes from another program that deals only rigid 
> bodies,
> potential parameters come from literature, tested with other progs.
>
> This flexible model works fine with gromacs (3.3.3 and 3.3.2).
>
> When I change constrains from none to all-angles (shake/0.0001), 
> according to manual
> all bonds and angles are converted to constrains => rigid molecule.
> Simulation, run, molecules remain rigid (verified with various tools) 
> but I get erroneous pressure.
> Especially for very low densities (<0.1 g/ml), where intermolecular 
> interactions are close to zero,
> and pressure is almost independent on  intermolecular interactions,
> the predicted pressure with all-angles is almost half (9.95 instead of 
> 20, 4.96 instead of 10 etc) of the predicted using
> the same topol without constrains or the predicted using the same 
> rigid model
> with other programs that deal rigid bodies.
> At higher densities still I get lower pressure but this could come 
> from potential.
> Any idea why this ?
>
> D.D.
>
>