[gmx-users] pcoupl Berendsen

Gonzalez Fernandez, Cristina cristina.gonzalezfdez at unican.es
Wed Nov 14 11:28:17 CET 2018


Hi Justin,

I have taken a few days in answering you because I was trying to reduce the discrepancies between the pressure I obtain after simulation and the one I set in the .mdp file. However, I have no achieve very good results. As I indicated in the previous email, in my simulations ref_p=1bar and ref_t=298K.
According to the output of gmx energy, the pressure average after the simulation is 0.19bar , Err.est= 0.59 and RMSD=204,98; and for the temperature average=298.003, Err.est=0,0032; RMSD=2,76.

>From this results and your previous email, as the error (Err.est) in the pressure is of the same magnitude as the pressure value after simulation, the pressure significantly differs from the reference value (1bar), so for example, more simulation time will be required. This is also supported by the high RMSD, which is in the order of hundreds. For the temperature, the error is 5 orders of magnitude lower than the obtained value and the RMSD is very low. This suggests that the system has reach the equilibrium temperature. Are these reasons correct?

Another thing that makes me think that the pressure I obtain is not correct is that the pressure I obtained after simulation and the one I obtain after analysis also differ significantly (0.19 and 3.9 bar respectively).

I have used long both NPT equilibration and simulation times (50 ns) and the results are similar to the ones I have indicated above, which apparently means that the system is stable.


>From these discrepancies, do you think the differences are not as important as I am considering?, what could I do to obtain more accurate pressure values?

Regarding the Parrinello-Rahnman is "not stable for low pressures", I understood that when using low pressures obtaining the reference pressure is sometimes difficult by using Parrinello-Rahman. I was trying to use this article in order to explain why my simulation pressures differ from the ref_p, but as you say, I have also read papers that use Parrinello-Rahman for simulating 1bar systems.

Thank you very much for all your help,

C.


-----Mensaje original-----
De: gromacs.org_gmx-users-bounces at maillist.sys.kth.se <gromacs.org_gmx-users-bounces at maillist.sys.kth.se> En nombre de Justin Lemkul
Enviado el: jueves, 8 de noviembre de 2018 14:01
Para: gmx-users at gromacs.org
Asunto: Re: [gmx-users] pcoupl Berendsen



On 11/8/18 7:50 AM, Gonzalez Fernandez, Cristina wrote:
> Dear Gromacs users,
>
> In my simulations, I have specified ref_p= 1bar but after MD 
> simulation I obtain pressures equal to 0.19 bar (even

A pressure without an error bar is a meaningless value. The fluctuations of pressure in most systems are on the order of tens or hundreds of bar, meaning your result is indistinguishable from the target value.

> with long simulation times) when using pcoupl=Parrinello-Rahman. I 
> know that Parrinello-Rahman is recommend for production runs and 
> Berendsen for NPT equilibration. However, I have read in an article 
> that Parrinello-Rahman is not stable for low pressures, so in such 
> situations its better to use Berendsen. I have tried to use Berendsen 
> for

I would be interested to know how this "not stable for low pressures" 
was determined, because it seems completely unlikely to be true. Most MD simulations nowadays use Parrinello-Rahman for pressure coupling at 1
bar/1 atm without any issue if the system is properly equilibrated (and if not, the problem is with preparation, not the barostat itself).

> MD simulation but I obtain this Warning and I cannot remove it with the -maxwarn option.
>
> "Using Berendsen pressure coupling invalidates the true ensemble for the thermostat"
>
>
> How can I use Berendsen for MD simulation?

Simply, you can't, and you shouldn't. The Berendsen method produces an invalid statistical mechanical ensemble. It relaxes systems quickly and is therefore still useful for equilibration, but should never be employed during data collection. Full stop.

-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

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

--
Gromacs Users mailing list

* Please search the archive at http://www.gromacs.org/Support/Mailing_Lists/GMX-Users_List before posting!

* Can't post? Read http://www.gromacs.org/Support/Mailing_Lists

* For (un)subscribe requests visit
https://maillist.sys.kth.se/mailman/listinfo/gromacs.org_gmx-users or send a mail to gmx-users-request at gromacs.org.


More information about the gromacs.org_gmx-users mailing list