[BUG] Cell file read incorrectly for 'qe/cp/traj' format
rashidrafeek opened this issue · 2 comments
When reading a qe/cp/traj output, the cells read from the [prefix].cel file is read incorrectly. The cells written in the [prefix].cel file produced in a cp.x calculation is transposed as mentioned here. But while reading them they are not transposed by dpdata and thereby a transposed cell instead of the original cell is read by dpdata. I noticed this when I tried to use dpdata to read a trajectory I ran previously and used the to_ase_structure function of LabeledSystem to write it into a different format and visualized it. I think the test file which tests this functionality was not created by a quantum espresso cp.x calculation.
I have modified dpdata locally to fix this and have confirmed that transposing the cell gives the desired result by visualizing. I can submit a pull request fixing this if it is okay.
@rashidrafeek Thank you very much for reporting the bug.
You are right the test file was synthetic data.
Could you please help us checking if the #520 fixes the issue and if the test file looks like one that is generated by cp.x?
Yeah, I have tested the PR and it does fix this issue. And yes, the transposed test file in the PR is similar to the one produced by QE.