DiffusionAnalyzer: initial and final structures do not match.

[thienbinh92]

Hello, I made a question in the matsci.org community discussion forum last month, but I have not received any answer yet. I hope when I post a question here, someone may offer a favor.
I use DiffusionAnalyzer to analyze AIMD's results at 800, 1000, and 1200K:
import matplotlib.pyplot as plt import collections from pymatgen.core import Structure from pymatgen.analysis.diffusion_analyzer import DiffusionAnalyzer file1 = ["test-400/vasprun.xml", "test-600/vasprun.xml", "test-800/vasprun.xml"] analyzer1 = DiffusionAnalyzer.from_files(file1, specie="Li", smoothed=False)
I get ValueError: initial and final structures do not match.
I do not really understand this error. Not sure if I make something wrong?

++++++++++++++++++++++++++++++++++++ My procedure to perform AIMD is like this: 1. Run with the NPT ensemble with INCAR: ISIF = 3 TEBEG = 1000 TEEND = 1000 IBRION = 0 ISYM = 0 NSW = 5000 NBLOCK = 1 MDALGO = 3 LANGEVIN_GAMMA = 10 10 10 10 LANGEVIN_GAMMA_L = 10 SMASS = 1.0 POTIM = 2.0 2. Take CONTCAR with average volume, then copy it to POSCAR. Run with NVT ensemble: ISIF = 3 and NSW = 25000. Then I use vasprun.xml and analyze the data with DiffusionAnalyzer.

[shyuep]

I am not sure what you are attempting to do, but you cannot analyze diffusivity using AIMD runs at three different temperatures. To calculate diffusivity at one temperature, you need a run with the same parameters, possibly spread across multiple start/stop runs (during which the initial structure of the next run must be identical to the final structure of the previous run).

After you get the diffusivity at different temperatures, then you can create an Arrhenius plot. Please review the examples in the matgenb.materialsvirtuallab.org site for information.

[thienbinh92]

Thank you so much! I am clear now.
I found in the example, that the MDS plot looks "spiky". Is it typical for raw data without a smooth option?

[shyuep]

Yes.