How to calculate the average volume of each Voronoi polyhedra type?

Quote from Constanze Kalcher on October 18, 2021, 10:01 pm

Hi Kingsley,

You can find the indices of atoms with a specific Voronoi Index (in this example the most frequent one) like this,

numpy.where((voro_indices == unique_indices[0]).all(axis=1))[0]

that means you can look up their corresponding Voronoi volumes like this:

voro_volumes = data.particles["Atomic Volume"]
voro_volumes[numpy.where((voro_indices == unique_indices[0]).all(axis=1))[0]])]

and simply use numpy.mean()to calculate the average value:

numpy.mean(voro_volumes[numpy.where((voro_indices == unique_indices[0]).all(axis=1))[0]])

 

Quote from Kingsley Carter on October 18, 2021, 11:55 pm

Dear Constanze,

 

Thank you very much for your tip. But I got some errors here.

I want to calculate 500 frame data and save all results of each frame into a file. But it has some errors in getting the average polyhedra volume of every index or something. Could you please have a look?

Thank you.

...
...

out1=open("CuZr.txt",'a')
for frame in range(pipeline.source.num_frames):
    data = pipeline.compute(frame)
    voro_indices = data.particles["Voronoi Index"]
    unique_indices, counts = row_histogram(voro_indices)

    # Find the average volume of every index
    voro_volumes = data.particles["Atomic Volume"]
    avol = numpy.empty([20], dtype=float)
    for v in range(20):
        vol = voro_volumes[numpy.where((voro_indices == unique_indices[v]).all(axis=1))[0]]
        avols = numpy.mean(voro_volumes[numpy.where((voro_indices == unique_indices[v]).all(axis=1))[0]])


    for i in range(20):
        print("%s\t= %i\t\t%.2f %%\t%.2f" % (tuple(unique_indices[i][2:]),counts[i],100.0*float(counts[i])/len(voro_indices),avol[i]),file=out1)
out1.close()

 

Quote from Constanze Kalcher on October 19, 2021, 3:36 pm

In your code snippet, avol is an empty numpy array of shape (20,), so using it in your print statement in line 56 doesn't make much sense.

If you would like us to send you a ready-to-use solution, please contact our Pro support via email and let us know your OVITO Pro license key.

Thanks.

Quote from Kingsley Carter on October 20, 2021, 4:45 am

Dear Consanze,

Thank you very much for your information.