Maxwell-Boltzmann distribution analysis does not fit well about dynaphopy HOT 4 CLOSED

That's strange. How many atoms did you use in the supercell?
Have you tried reading the XDATCAR instead? (remember to set the time step using -ts flag)
What do you obtain for the atomic displacements distribution calculation?

from dynaphopy.

@abelcarreras I used nearly the same configuration as the Si example in the doc of DynaPhoPy. I just tested using the XDATCAR, and found the velocity fitted well. Then I tried using different number of steps: -n 5000 and ``-n 2000`, giving slightly different temperatures fluctuating around 800 K. That means XDATCAR gave the correct results for Maxwell-Boltzmann distribution fitting. The same tests using OUTCAR always give 2022.4 K, no matter how many MD steps I specified.

Except the difference in Maxwell-Boltzmann distribution analysis,

• the renormalized phonon dispersion looks the same in the two cases
• "Peak analysis" at Gamma point using -n 8000 gives the following results,
  
• "Atomic Displacement" analysis at direction (1,0,0) gives,
  

I used VASP 5.3.3 for the MD calculation.

from dynaphopy.

Ok, I see. I will check the OUTCAR parser. I use a slightly different version of VASP so maybe something changed in the OUTCAR file. The differences may also be due to the precision. The XDATCAR has more precision than the OUTCAR. I recommend to read the trajectory always from the XDATCAR file. In the future I plan to drop the OUTCAR parser support.
Thanks a lot for the feedback I will revise the OUTCAR parser and the online manual.

from dynaphopy.

Thank you for your reply. Your code is really nice.

from dynaphopy.