[gmx-users] Velocity autocorrelation

[Ignacio Fernández Galván]

--- On Thu, 29/3/12, Krzysztof Kuczera <kkuczera at ku.edu> wrote:

> You probably need to save velocities
> at least every 10 fs;
> look at your plot on a fine enough time scale  - say
> 0-1 ps range
> probably running in NVE ensemble would help as well

Thanks for your suggestion. I was already saving velocities every 10 fs. I tried with a NVE ensemble, but the behaviour is the same.

I have uploaded a sample velocity ACF here:
http://djelibeibi.unex.es/files/vac_NVE.xvg
That's the autocorrelation file for CN atoms in acetonitrile.

I have also uploaded the corresponding mdp file:
http://djelibeibi.unex.es/files/dinamica_NVE.mdp

I still don't know how to properly analyze this, I would say that I want the amplitude to drop to almost zero, and that would mean a relaxation time of ~5-6 ps. But is this the correct/expected behaviour?

Thank you,
Ignacio