Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

Divergence of the power spectra with the larger supercell #17

Closed
ballal09 opened this issue Jul 4, 2022 · 10 comments
Closed

Divergence of the power spectra with the larger supercell #17

ballal09 opened this issue Jul 4, 2022 · 10 comments

Comments

@ballal09
Copy link

ballal09 commented Jul 4, 2022

Dear Abel Carreras

I hope you are doing well. I have run into an issue when I was trying to compute the full power spectrum for a larger MD supercell. When I have calculated the full power spectrum for 1 1 4 supercell (120 atoms), I have found a good match with the harmonic dos (of course 1 1 4 supercell as well). But, When I increased the supercell to 2 2 4 (480 atoms), I have found that the high frequency regime of the power spectrum does not match with the harmonic phonon dos for 2 2 4 supercell although there is some similarity. I have attached the two plots here for your reference. Please let me know your thoughts on this. For your information, I have used 30 ps of trajectories (1 fs timestep) to calculate the power spectrum for both supercell. Therefore, I am thinking the length of the simulation may not be an issue. Thank you very much for your time.

FPS_T100K_S114
FPS_T100K_S224_N30000

Best
Ballal
@abelcarreras
Copy link
Owner

I think the issue could be an effect of attaching MD trajectories.
The joint between trajectories can introduce some high frequency artifacts. I have not tested this extensively so I don't know the magnitude of this effect. In general I think this effect should be small but it may depend on the system.
Things you can check:

  1. Compute the full power spectrum of the individual MD and check if the issue is also present.
  2. Make sure the individual MD are properly converged. The artifacts introduced in the trajectory joints may be amplified by not completely converged MD. Maybe you can use "--read_from" option to test different equilibration times and check if this issue is minimized or not.

@ballal09
Copy link
Author

ballal09 commented Jul 5, 2022

Dear Abel Carreras

Thank you for your quick response. I have some confusion understanding your suggestion.

  1. I have computed the full power spectrum for 1 1 4 and 2 2 4 supercell above individually. Would you please explain what do you mean by full power spectrum of the individual MD?
  2. I have carefully jointed the trajectories sequentially, so I hope that won't be an issue. I am working on to check the convergence of the MD run.

Again Thank you for your time.

Best
Ballal

@abelcarreras
Copy link
Owner

  1. If I understood correctly you computed the full power spectrum joining several MD trajectories right? The idea is to check the full power spectrum using a single MD trajectory.
  2. But these trajectories are completely independent right? If that is the case at the joints it should be a discontinuity in the positions and velocities.

@ballal09
Copy link
Author

ballal09 commented Jul 5, 2022

Dear Abel Carreras

Thanks for the quick response. Now it make sense to me. First of all, the jointed trajectories are not independent, I have resumed the simulation where it was left. I am pretty sure there is no discontinuity in the position and velocity data. I have also calculated the full power spectrum using a single MD trajectories but no luck. I would appreciate any valuable suggestion. Thank you for your time!

Best
Ballal

@abelcarreras
Copy link
Owner

Ok, then if the trajectory is correct maybe there is some bug in the code.
Can you send me a minimal example that shows the issue so I can reproduce it in my computer?

@ballal09
Copy link
Author

ballal09 commented Jul 7, 2022

Sure, I would be happy to provide you the data to reproduce the problem. Please let know how do I send the data. Thanks.

@abelcarreras
Copy link
Owner

You can just attach the data in an e-mail or share a dropbox/Gdrive link here.
As you prefer.

@ballal09
Copy link
Author

ballal09 commented Jul 13, 2022 via email

@abelcarreras
Copy link
Owner

Hi @ballal09,

I found the issue. The point is that the order of atoms in the supercell is not correct. In order to compute the atomic displacements Dyanphopy requieres the atoms of the supercell to be in a particular order respect to the unit cell. This order is the same as the supercells generated using phonopy.

You can generate a "proper" supercell from a unitcell using -c_poscar option in dynaphopy. It is recommended to first generate the supercell using dynaphopy and then run the MD using this supercell.

However if the simulation is already done, as an alternative, you can run Dynaphopy using the tag --auto_order to try to guess the atoms order from the MD data. This may not work in all cases but fortunately, in your case, it worked and I was able to generate the proper full power spectrum.

The way to check that the atoms order has been taken correctly is to do a Maxwell-Boltzmann analysis. This plots the distribution of atomic velocities and fits it to a MB distribution. If you observe that the distribution is well fitted to the MB curve with the temperature you set in your simulation then the order is most likely right.

@ballal09
Copy link
Author

Dear Abel Carreras

It was really helpful! I have fixed the problem. Thank you so much for your time and help!

Best
Ballal

@gabkrenzer gabkrenzer mentioned this issue Feb 2, 2023
Closed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Assignees
No one assigned
Labels
None yet
Projects
None yet
Milestone
No milestone
Development

No branches or pull requests
2 participants
@abelcarreras @ballal09