MD running for Pyxaid

[shudong wang]

Dear developers,

The algorithm in NAMD is to do the adiabatic MD firstly, here the MD is Born-Oppenheimer always(am I right?). My questions is could we perform the CPMD (Car-Parrinello) instead of 

BOMD in the adiabatic simulations? And then we can sunbstract the trajectry to do the NAMD.

Thanks!

Shudong

[Alexey Akimov]

Hi Shudong,

Yes, one can use CP dynamics or BO (with DFT or classical or any other force fields) to produce the trajectories. 

I don't understand how you are going to subtract trajectories (what minus what?), but it is likely not needed anyway. All one needs for NBRA-NA-MD are somehow generated trajectories.

Of course, the quality of this "somehow" defines the quality of the subsequent NA-MD simulations.

Best,

Alexey

[shudong wang]

Dear Alexey,

Thank you for your quik reply.

I am running the different level of the adiabatic MD just to get the optimal trajectories for the subsequent NAMD. 

I found the different ion relaxitions give the various NAMD results with QE and VASP interfaces. Even in QE, the

Andersen and Berendsen ion_temperature are different for the NMDA results. Do you have an idea about the MD

setting for QE?

Thanks!

Shudong

[Alexey Akimov]

Dear Shudong,

Yes, of course the electronic dynamics will depend on nuclear one and on the details of thermostat. The couplings depend on the vibrational frequencies and types of motions of nuclei. 

You may find it useful to look at the discussions at:

Akimov, A. V. Libra: An Open-Source “Methodology Discovery” Library for Quantum and Classical Dynamics Simulations. J. Comput. Chem. 2016, 37, 1626–1649. In particular, see the section on the azobenzene simulations with and without thermostat (Figure 9).

Also, take a look at this work: 

Akimov, A. V. Stochastic and Quasi-Stochastic Hamiltonians for Long-Time Nonadiabatic Molecular Dynamics. J. Phys. Chem. Lett. 2018, 9, 248–257.

where I have played with a simple toy model and explored the consequences of nuclear dynamics modulation (NVE ensemble vs. NVT with different types of noise). This is exactly what one has in the atomistic simulations with different nuclear relaxations or thermostats.

Hope this helps,

Alexey