Re: [CP2K:15029] Gentle Heating

[Josip Lovrić]

There is no way to do 800K over 2 ps.

On Mon, Mar 29, 2021, 15:03 ASSIDUO Network <lenardc...@gmail.com> wrote:

Good day, I trust you are doing well.

I'm in the process of getting my hands dirty with AIMD using CP2K. I am familiar with MD simulations from AMBER, but not quite yet CP2K.

I would like set up a gentle heating in my AIMD simulation, gently heating the box up from 0 K to 800 K. I have read on this group that annealing should be used for this, but my issues here are the following:
- The heating phase has the volume and number of molecules constant, but allows the energy, the temperature and the pressure to change. None of the ensembles seem to fit this. So, I decided to go with NVT since that is what I have used with the heating simulation with AMBER (many tutorials available on it), but there you could specify the initial and reference temperature.

- How do I properly use annealing to heat the temperature up from 0 K to 800 K over 2 ps? Annealing just scales the velocities of the particles, so I'm not sure how to best choose this scaling factor.

- How do I print the temperature throughout the AIMD run?

I have included my input file. I would appreciate if you could go through the &MOTION section and tell me if it all makes sense.

Kind regards,
Lenard

--
You received this message because you are subscribed to the Google Groups "cp2k" group.
To unsubscribe from this group and stop receiving emails from it, send an email to cp2k+uns...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/cp2k/e2da3b1a-39d8-45aa-912f-531eb540f12en%40googlegroups.com.

[Marcella Iannuzzi]

Dear Lenard, 

Heating the system by constantly rescaling the velocity generates a dynamics that does not correspond to any meaningful thermodynamic ensemble.

For sure, to set up a thermostat while forcing the temperature to change is not a good idea.

If you want to use the ANNEALING in cp2k just set NVE, obviously the energy is not going to be conserved.

Activating the annealing, at each step the actual temperature is multiplied by the scaling factor and the velocities are rescaled accordingly.

Probably a better way to bring the system from a lower to a higher temperature would be to run a set of NVT simulations at increasing temperature, say 150K, 300K, ...800K

Each simulation should be several ps long, such that the system has the time to sort of equilibrate at each temperature before ramping up.

best

Marcella

[Lenard Carroll]

But adding the kinetic energy all at once would result in the system acting non-physically. I know this as I've done classical MD simulations for years. I like the approach of multiple NVT simulations, gradually increasing the temperature. I think I will also be using the previous temperature's final trajectory/velocities for the new temperature NVT simulation.

Thanks

To view this discussion on the web visit https://groups.google.com/d/msgid/cp2k/5f2e3349-9440-4bde-b87f-cdde5a90b9d3n%40googlegroups.com.

[Thomas Kühne]

Dear Lenard, 

alternatively you may try using TEMEPRATURE_ANNEALING with values slightly larger than 1. 

However, this corresponds to an exponential not linear increase in temperature. 

Cheers, 

Thomas 

To view this discussion on the web visit https://groups.google.com/d/msgid/cp2k/CAKur15-3ngs%3D1_ZxjM9FVNvmSj_Q9oBVwG0OYF%2BV%2BX53oru9BQ%40mail.gmail.com.

[Marcella Iannuzzi]

Dear Lenard, dear Thomas, 

I am still of the idea that defining these as NVT or NVE simulations is misleading, 

since the generated configurations do not belong to any of these ensembles. 

Moreover, adding some kinetic energy stepwise and running 10/20 ps to equilibrate end explore, at least partially, the corresponding ensemble, is for sure much better than increasing the temperature from 0 to 800K in 2 ps, as was the original intention. 

Best

Marcella

[Lenard Carroll]

I believe now too that 2 ps is too short, it's not that it's impossible to achieve, I just don't think it is enough time for the system to produce sensible results in the process.

To view this discussion on the web visit https://groups.google.com/d/msgid/cp2k/39b9eb8c-2f28-4624-9064-7258fce8a040n%40googlegroups.com.