[gmx-users] Relationship between temperature, molecular velocity, and intramolecular potential

[Seunghyun Chung]

Hello all,

Recently, I got weird results from my simulations. My NVT system 
contains about 300 molecules, which are looks similar to the C60 
fullerenes but a little bit modified.

I simulated them with two conditions:
1) constrained bonds
2) harmonic bonds

The problem is this:
When I measured the average 'molecular' velocity, case 1 is much faster 
than case 2.

I thought the molecular translational energy is independent to the 
molecular internal energy. But this result is completely opposite to 
this reasoning. It seems like internal vibration increase the system 
temperature and, as a result, it lowers the molecular translational 
energy of case 2 when it compare to that of case 1.

Their 'atomistic' average velocity are same (therefore, their measured 
temperatures are the same because, to my knowledge, GROMACS uses 
atomistic vel. for temperature) but 'molecular velocity' is not same. Is 
this right conclusion?

I'm asking this because my previous thought was they were independent 
and the molecular velocities of case 1 and case 2 were the same.
I tried both nose-hoover and berendensen and got the same results.

Thank you in advance,
Seung-Hyun