Ehrenfest dynamics: Glycine

[Victor Volkov]

Good day.

I am exploring molecular dynamics under field.

I set glycine at the center of a box,

and suggested Ehrenfest dynamics with a time step of about 0.04 fs,

while the custom field file sets the external field to ZERO.

The simulation proceeds alright, till the time-frame 500, after which

there is a peculiar structural "expansion".

Q1: Would you comment this?

Also, I attach a fragment of the momentum file:

CP2K prints reference point position to be "chaotic",

while I select REFERENCE COAC: since the atomic charges of the glycine,

the reference point should be next to the C-alpha atom.

Q2:  would you comment on the usage of the setting I selected? 

Thank you.

Victor

[Matt Watkins]

Hi Victor,

from your setup the main things are:

• the cutoff is too small - should be set like a standard DFT calculation - something around 400 Ry and the REL_CUTOFF to 60 Ry or similar.
• the timestep is very aggressive. If you plot your energy as function of time I think you will see the conserved quantity drift badly at some point in your simulation and then it explodes. Try something like
  
  • stepsize [au_t] 0.1

Matt

[Victor Volkov]

Dear Matthew:

good afternoon.

Thank you for the kind reply.

Indeed, the cut-of is too small.

Thank you.

> the timestep is very aggressive.

This is curious.

I always thought that under-sampling (using a longer time-step to speed up simulated results)

is a problem.

I selected the time-step to over-sample (a bit) a period which would correspond

to oscillation of a field in near UV - about 230 nm.

Oversampling would be necessary to model circular polarized perturbation using two fields,

when the second one (0 1 0) is Pi/2 shifted in respect to the first (1 0 0), if "propagation"

is considered along Z (0 0 1).

In principle, I am curious to see two things:

1) if the external field could be such that molecule would break apart;

2) if, then what kind of fragments and how they would fly away according to the driving conditions.

It is a sort sense if a "destruction of the Hamiltonian" may mimic certain event upon ionisation,

not all events but some.

Thank you again.

With best regards.

Victor

.

 

--
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/c389ccfe-3201-4f0a-83e7-e931000e2366n%40googlegroups.com.

[Victor Volkov]

Dear Matthew:

good afternoon.

Sorry to bother once more.

I took Gly and Ala oriented the way that their COO-Ca-N plane is parallel to XZ plane

and modeled charge dynamics propagating under such 5 field settings

{Ex = 0, Ey = 0} no field

{Ex=0, Ey} polarized along Y

{Ex, Ey = 0} polarized along X

{Ex, Ey + P/2 shifted} left circular polarized

{Ex, Ey - P/2 shifted} right circular polarized.

The initial geometries are proximal to the global minimum optimized under DFT.

I attach an image where I sum-up the results.

When there are no fields, apparently, there is just the equillibration: energy flows into kinetic populating vibrations,

and there is the intra-molecular H dynamics between the terminals.

Field polarized along X (parallel to the COO-Ca-N plane ) makes thermalization slowe.

Field polarized along Y (perpendicular to the COO-Ca-N plane) introduce strong anticorrelation of side groups charges for Gly,

but it does not slow down the thermalization for Gly.

Instead, in the case of Ala, the temperature and the kinetic energy explode at a certain point - the shake up start to break the molecule.

At the same time, the shake up for the potential energy is about the same as when field is parallel to X.

This is peculiar since one could think that shaking between the terminals should have larger effect. 

Using CD polarized light, phase of charge modulations of the side groups, only, reflect "chirality" of the drive.

In the case of Ala, cp2k simulates both, the temperature and the kinetic energy explode at a certain point.

I simulate using energy constraining NVE and instructing T=300K.

Potential energy does change smooth, always.

I wish to ask your opinion: the increase of temperature as cp2k simulate is it a signature

of investing energy into molecule that it start to break it apart "properly",

or it is a "proper" mis-performance of the NVE part of the equations,

or it is a signature of the equations collapse?

I doubt about the third because I tested increasing the amplitude of the drive: at certain point

the potential energy start to show a ripple.

The data I attach here does show potential energy to vary smoothly.

Therefore, I think it is likely that the molecule is getting very hot.

Thank you.

With best,

Victor 

  

 

On Mon, May 22, 2023 at 4:56 PM Matt Watkins <mattwa...@gmail.com> wrote:

[Natalia K]

Dear Victor,

I am curious, in these calculations, is the total energy ("Cons Qty" in the systemlabel.ener file) conserved? What about the case when the temperature grows high? What version of CP2K did you use to get the results attached to your last email?

Thank you and best regards,

Natalia

[Victor Volkov]

Dear Natalia:

> in these calculations, is the total energy ("Cons Qty" in the systemlabel.ener file) conserved?

With an external field you invest energy into the system, if it has a susceptibility, a degree of freedom to accept the

external drive.

According to the nature of the system, part of the external field would stimulate electron dynamics and correlations,

which would affect structural dynamics and, hence, temperature.

At the same time, when we define a type of ensemble, a thermostat, or a barostat,

then there would be an exchange with the bath or not.

To secure quality of a numeric solution, I monitor dynamics of dipole moment:

it should follow the external field smooth.

Once its motion is choppy-jerking, the numerical solution may not be adequate.

Not stable and not reliable numerical solutions are possible.

This is up to the quality of the algorithm.

I managed reasonable simulations but not longer than 8 femtosecond with the step of 38.8 atosecond.

I focused on the thermodynamic response to UV periodic radiation.

Several atosecond step was necessary to over-sample UV periodicity well.

 

> What about the case when the temperature grows high? 

In my studies I used NVE ensemble.

Using NVE ensemble I observe temperature to increase and this to be consistent with the 

kinetic energy to increase.

If an effect of the external field onto electron dynamic, then, there should be an impact on structural dynamics, and, therefore, temperature.

Accordingly, "Cons Qty" = Pot.Energy + Kin.Energy.

> What version of CP2K

CP2K version 2023.1 (Development Version)

I hope these comments may help.

Victor

To view this discussion on the web visit https://groups.google.com/d/msgid/cp2k/2ba789bb-9ef4-4b8e-9fd6-a06c53f39720n%40googlegroups.com.