pymatgen | Python Materials Genomics | Dataset library

The issue was indeed that I was using the wrong python version to run the pmg-script (It was a work computer with many different python versions and I was using python 3.7.6 but there was an old python 2.7 set as default which was used when jus ran "python pmg"). I ran conda activate base in my /home/username/anaconda3 and then python pmg etc. and it worked.

Judging from your build IDs, you're working with an osx-64 platform. The dependency requirements for SymTopo look weird for all the osx-64 builds, specifically, they require pymatgen=2019.4.11.*, which from searching Anaconda Cloud just doesn't exist. In the linux-64 builds, however, this requirement is pymatgen>=2019.4.11, so I'm thinking the author has done this incorrectly, hence why you can't resolve a coherent package configuration.

Other than contacting the author and asking them to post some working builds, I think you have two possible courses of action:

1. Force Install
2. Using PyPI Builds

Either way, assume the resulting env will be fragile, so try to avoid any further unnecessary installations into it.

If we assume that the pymatgen=2019.4.11.* really should be pymatgen>=2019.4.11 then you could just install all the prereqs (see YAML in next section), then use

You don't need to query for each individual mpid. Your first code block already queries for the "cif" information of all the materials!

All you need to do is to convert the cif strings to structures using PyMatGen:

You can use a regular expression to look for _cell_length_a (or any other key), followed by some spaces, and then capture whatever comes after that until the end of that line.

The second set of co-ordinates in square brackets are the fractional co-ordinates of that site with respect to the Structure's Lattice. Fractional co-ordinates wrap around periodic boundaries, so:

[-0.3863, -0.2759, 1.3863]

is equivalent to these co-ordinates in the [0,1) range:

[0.6137, 0.7241, 0.3863]

For technical reasons, fractional co-ordinates are not wrapped to [0, 1) by default however, since this information is meaningful for some simulation code. For practical purposes however, the easiest way to interpret them is in the [0, 1) range.

You can use getattr in order to dynamically select attributes from objects (the Orbital package in your case; for example getattr(Orbital, 's')).

So your loop would be rewritten to:

To do this, use the SpacegroupAnalyzer object.

Update:

So, I've figured out a way of doing this. I actually converted each neighboring atom into their Cartesian coordinates. Then, I subtracted the coordinates of the central atom to which the neighboring atoms are attached. Lastly, I found the angles by calculating the dot product of each two vectors divided by the product of their magnitudes, then taking the arc cosine of these values. The code which I used is below. This may not be the most elegant way of doing things, but it does get the job done. If anyone has improvements, feel free to comment.