Getting structure using material id in for loop not

pymatgen
1

Hi,

I am trying the get structures using MPRester()
I have around 4000 materials id I am interested and following is the code.

for mat_id in df['id_discharge']:
        structure = mpr.get_structure_by_material_id(mat_id, final=False)
        IStructure.add_oxidation_state_by_guess(structure)
        structures.append(structure)

Even after overnight running, there is no error and not finished. Is it supposed to be this long? or is there something wrong with the code?

Please help.
Thanks,

2

Hi @Chae-Ho_Yim,

For this number of materials, it would probably be easier to run:

with MPRester() as mpr:
    docs = mpr.query({'material_id': {'$in': all_mat_ids}, ['structure', 'material_id'])

Where all_mat_ids is a list of material ids of interest.

However, the line IStructure.add_oxidation_state_by_guess(structure) is probably what’s taking the most time, not the API retrieval.

Best,

Matt

3

Hi, Matt.

I tried as you recommended, but the docs is returning empty array. Would it be because there is no properties such as ‘structure’?

I was going through the MPRester().supported_properties and supported_task_properties , they do not include structure.

Many thanks.
Chae-Ho

4

Hi @Chae-Ho_Yim, “structure” is definitely a supported property, could you paste your code?

5

Thanks, figured out.

But I do have another problem.
I figured, MPRester also has option for Initial and Final structure to obtain.
Whatever option I use, I get 2900 structures, where the list of matID is 4400. When I was using the for loops I could obtain all 4400 structures.

Is there any reason why there is the difference. I have checked whether there was any duplicate in mat ID and there wasn’t.

Thanks.

6

This is difficult to diagnose without a code example, both methods should give equivalent results.

7

Thanks, Matt.

Following is the code.

df = pd.read_csv("MP_battery_info.csv")
df = df.rename({'id_discharge': 'material_id'}, axis=1)
all_mat_id = []
for matid in df['material_id']:
    all_mat_id.append(matid)
# properties = ['material_id', 'initial_structure', 'energy_per_atom', 'spacegroup', 'e_above_hull',
#               'band_gap']

properties = ['material_id', 'initial_structure']


with MPRester() as mpr:
    docs = mpr.query({"material_id": {"$in": all_mat_id}}, properties)

For the for loop:

mpr=MPRester()
structure=[]
for matid in df['material_id']:
    stru=mpr.get_structure_by_material_id(matid, final=False)
    structure.append(stru)

Please let me know if you need the csv file, the csv files are all the information from the battery id.

Thanks,

8

Looks like it should be fine, but why are you asking for the initial structure and not the final structure (e.g. just structure, or final=True)?

9

It was the only way to get structures, I believe some of the material id in the csv didn’t have the final structure, which caused error. Putting final=False was the only way I could run and get the information.

The only think I don’t understand is, when I use MPRester, putting properties with initial_structure, shouldn’t I get the same number as you said?
Thanks,

10

Try properties=["structure", "material_id"]