Vibrational phase selection for apolyatomic molecules using ReaxFF

Dear Lammps users,

I am about to run Classical MD simulations for a system of polyatomic molecules using NVE ensemble and ReaxFF as the forcefield. All molecules are of a same type.

After I used the minimization style which makes system to reach a local minima (or a stable configuration), all of the plyatomic molecules would be at the minimum of the well in other words at the point of zero potential vibrational energy. After randomly selecting the initial velocities of the atoms from the Gaussian distribution at the target temperature, trajectory begins. However, due to the transfer of energy from kinetic modes of energy to vibrational potential energy, the kinetic temperature will drop from the thermal target temperature.

I believe that after the minimization step I have to somehow assigning vibrational phase (distribution) to the molecules such that the average vibrational potential energy won’t be zero but be the value predicted by equipartitioning of the energy. This is where I have no idea how to do the vibrational phase selection. Can anyone kindly help me how to do it?

Specially at the temperature I am working with (4,000 K), the anharmonic effects are significant and for the multi variable potential I have for my polyatomic, it is vague to me how i have to do so.

Thank you so much for assisting me.


You would see the same effect with any system or potential,

e.g. unbonded atoms with a simple LJ potential. Run the

bench/in.lj script as an example. It’s just repartitioning
of kinetic vs potential energy. You can either

initialize the system with bigger velocities (higher temp)

so the temp quickly drops to what you want. Or use

a thermostat to bring the temp up to where you want.