[gmx-users] increase of temperature or pressure

[Mark Abraham]

On Wed, Dec 19, 2012 at 11:27 PM, Ahmet yıldırım <ahmedo047 at gmail.com>wrote:

> Dear users,
>
> When analyzing some of the simulations;
> I get 312 K, 310 K or 308 K instead of 300K (the t_ref value for the whole
> system).
> 1. What is the reason for the increase of temperature? For example it can
> be cutoff artifacts. What else can be?
>

Any kind of integration inaccuracy could lead to heating, e.g. too large a
time step. The time period after equilibration over which you observed the
average temperature, and the size of the system is a necessary part of
reporting a temperature, else it's meaningless. Too little time or an
average over too few atoms leads to large fluctuations and thus low
expectations that the average of the sample will be correct.

2. To get the the error small tau_t can be reduced. What is the minimum
> value for tau_t?
>

That's probably treating the symptoms, not the disease.

Sometimes, the pressure in the output files is high ( 7 bar, 6 bar or 1.5
> bar instead of 1 bar).
> I.The reason for this?
> II. tau_p can be reduced as temperature. And what is the minimum value for
> tau_p?
>

See GROMACS webpage on pressure coupling and measuring pressure. Manual
discussion also.

Briefly, what could be the reasons for the increase in temperature or
> pressure? What are your suggestions for the solution?
>

Follow protocols in publications that discussed how and why they chose
their settings for the force field of interest, or which demonstrated their
choices were good. Making things up, or copying others blindly is a recipe
for wasting computer time.

Mark