[gmx-users] vdW parameters and atoms moving too fast

[Mark Abraham]

Nicolas SAPAY wrote:
> Hello everybody,
> 
> I'm still fighting with the CHARMM implementation in GROMACS. I have
> discover  discrepancies in some LJ/LJ1-4 energies and have perform some
> basic test to see if the problem comes from my parameters or the way I use
> Gromacs. So, I have built a system made of 1 OT and 1 HT non-bonded atoms
> seperated by 0.1 nm to test vdW parameters (the system is not directly
> relevent, it's just to make a test). Then, I have run 1 step of MD.
> Initial energy values are OK : I found exactly what I should find. But
> problems arise at step 1 where the atoms move too fast while kinetic
> energy remains null. I have check my mdp files and everything looks OK (I
> have used a small timestep of 0.0001 ps, a NVE ensemble, a Gromacs
> compiled with double precison, ...). Finally I don't understand why atoms
> move so fast with no kinetic energy. So, I wonder if my system is not "too
> simple" and leads to initialiation problems, or if there is some problem
> of accuracy with non bonded energy when atoms becomes too close 

I doubt it, but you can test for this readily enough... separate your atoms!

> and should
> relax, or I have made a too evident mistake that I cannot fix...

Attaching your .mdp and .top would probably be useful here.

Mark