[gmx-users] pullx/pullf and COM energy

[Alex]

Ah, nevermind -- problem solved. The proper integral is of course wrt 
d(vt-x), which gives a fairly noisy version of the COM energy.


On 5/13/2017 1:26 AM, Alex wrote:
> Hi all,
>
> I am slowly pulling a single ion (restrained in-plane) along the Z 
> direction at a constant rate via a spring (what you call umbrella). At 
> the moment, the physics probably doesn't matter, but my question will 
> be clear in a moment and yes, I am aware of threads like 
> http://thread.gmane.org/gmane.science.biology.gromacs.user/77762.
>
> It is my understanding that pullf.xvg reports the force exerted by the 
> spring upon the ion, i.e. the k(vt-x(t)), where v is the pulling rate 
> and k is the puller spring constant. It is also my understanding that 
> the COM pulling energy (if you apply 'gmx energy' to your energy 
> output) reports the energy stored in that spring. Seems elementary. It 
> would follow that if I were to take the force data from pullf.xvg and 
> integrate it, I would indeed obtain the COM energy.
>
> The dumb integral is f(x)vdt, while the more appropriate approach is 
> f(x)dx, where dx is obtained from pullx.xvg for every t. Makes little 
> difference -- in my case vt and x(t) are always very close and I get 
> very similar results from the two integrals, aside from noise. 
> However, both of them are also entirely different from what I get for 
> the COM pull energy from the energy utility.
>
> Can someone help solve this mystery?
>
> Thank you,
>
> Alex
>