[gmx-developers] problems about g_velacc

[Berk Hess]

Qiao Baofu wrote:

> Hi developers,
>
> The velocity autocorrelation function calculated from g_velacc 
> -nonormalize is wrong, which is 10^4 times bigger than the right value.
>
> I have a MD result of Cl- ions at T=425K. With it I calculated the 
> velocity autocorrelation function at t=0. g_velacc gives the values 
> 300 (without unit). But according to 3kT/2 = mv^2/2, the value should 
> be 0.3 (nm/ps)^2.
>
> From the following website, I found the problem has been talked about 
> in 2003. But unfortunately, it exists up to now.
> http://www.gromacs.org/pipermail/gmx-users/2003-August/006845.html

g_velacc reads an tpr file when you do not specify the option -n.
As described in the manual it will then determine a momentum 
autocorrelation function.
When you use -n the output should be the velocity autocorrelation function.

I will fix g_velacc, so it no longer automatically does a momentum 
autocorrelation
function.

Berk.