BUG: accuracy of output structure

[materialsproject]

BUG: accuracy of output structure

[utf]

Hi @JaGeo, thanks for noticing this. It is a very important point! I hadn't realised the structure in the vasprun was written to such low precision.

Let's read it from the CONTCAR as you suggest!

[JaGeo]

I currently think it is related to this line here:

atomate2/src/atomate2/vasp/schemas/calculation.py

Line 496 in 73b9b3e

structure = vasprun.final_structure

The final structure is taken from the vasprun.xml ... And, the vasprun.xml has fewer digits than the CONTCAR. @utf , should we instead read in the CONTCAR? I will in any case try it locally.

[JaGeo]

Yeah, I was actually also a bit surprised to see this low precision... (it's also far from optimal in my opinion)

I will try to implement it next week. I think there are at least two points in the code where this is done but I might be overlooking points. I think also the initial structure is read from the vasprun.xml ...

[utf]

Fixed in #167