[gmx-users] Error in NVT equilibration

[Justin Lemkul]

On 9/19/18 6:14 AM, Gonzalez Fernandez, Cristina wrote:
> Hi Justin,
>
>
> I’ve been looking for the non-bonded setting parameters for the GROMOS force field. I have found that for this force field coulmb_modifier and vdw_modifier = Potential-shift-Verlet. I have changed this in the .mdp file, but I don’t have seen more differences, Is it correct? Is it enough to change the columb and vdw modifier option?. I obtain the same error when running the equilibration.

I haven't used the GROMOS force fields with the new Verlet options, and 
it was parametrized for rather old algorithms, so I can't offer you the 
latest advice on this. Electrostatic forces are typically computed with 
PME, and a plain cutoff for vdW. Your cutoff values in the previous .mdp 
file were also incorrect and need to be fixed; it's not just a matter of 
the modifiers.

>
> I have tryied to use the default GMX values for dt (0,002) and nsteps (50000) because they are so different to the ones I have proposed (0,015 and 6667 respectively) and the nvt equilibration runs. My values derive from the

A time step of 15 fs is not going to be stable with an atomistic force 
field. A sensible dt is usually no larger than 2 fs without playing 
other tricks.

> annealing I am trying, and with the default values the annealing is not fulfilled since the temperature doesn’t start in the first annealing temperature.
>
>
> I need to increase the temperature from 278 to 298K with a rate of 0,298 K every time step for 1 ps. I checked your previous posts and use annealing for that purpose. From the rate 0,298K/step and knowing that I need to increase 20 K de temperature, I calculate the requied number of steps and from this value and 1 ps I found that dt should be 0,015 ps. From this dt, as I considered a total equilibration time of 100 ps, I calculated nsteps (6667).

Don't adjust dt, which is the simulation time step. Your simulation will 
never be stable with such a value. You can't simultaneously specify a 
heating rate, total change, and fixed time interval. You should consider 
dt fixed and determine how much time it will take to reach the total 
change in temperature at the desired rate. For your situation, to heat 
by 20 K with a rate of 0.298 K/2fs, you would need to warm over a period 
of about 134 ps.

>
> The annealing settings are:
>
> ;Simulated annealing
>
> annealing = single
>
> annealing_npoints = 2
>
> annealing_time = 0 1
>
> annealing_temp = 278 298
>
> How can I satisfy the 0,298K increase during 1 ps until 298K while running the equilibration?

Change

annealing_time = 0 134

and you'll get the rate of heating that you want. As stated above, you 
can't do it in 1 fs while also specifying the rate of heating. dt is not 
a dependent variable that is subject to such manipulations.

-Justin

-- 
==================================================

Justin A. Lemkul, Ph.D.
Assistant Professor
Virginia Tech Department of Biochemistry

303 Engel Hall
340 West Campus Dr.
Blacksburg, VA 24061

jalemkul at vt.edu | (540) 231-3129
http://www.thelemkullab.com

==================================================