Find calculations with alloys by elements

New to AiiDA
1

Lets say I want to inspect all calculations (variety of codes used) including gold and copper in my database.
First I tried something like this

query = QueryBuilder()
query.append(StructureData, filters = { 'attributes.kinds': {'contains': ["Au", "Cu"]} }, tag='structure')
query.append(CalcJobNode, with_incoming="structure")

But couldn’t get it to work since kinds is not a simple list of strings. So I did the following to get it to work quickly.

query = QueryBuilder()
query.append(CalcJobNode)
elements = ["Cu","Au"]
alloy_calcs = []
for calc, in query.iterall():
    try: 
        kind_names = calc.inputs.structure.get_kind_names()
        if all(el in kind_names for el in elements):
            alloy_calcs.append(calc)
    except Exception as e:
        print(e)

Is it possible to accomplish this with the QueryBuilder (maybe similar to my first approach)?

2

Hi @un1queu5er!

Yes, this is possible. The following should work for you:

from aiida import orm

# Generate the kinds that you want to query
kind1 = orm.Kind(symbols='Au', name='Au')
kind1 = kind1.get_raw()

kind2 = orm.Kind(symbols='Cu', name='Cu')
kind2 = kind2.get_raw()

print(kind1) # This is exactly the type of dict you are looking for in the kinds attribute
# {'symbols': ('Au',), 'weights': (1.0,), 'mass': 196.966569, 'name': 'Au'}

q = orm.QueryBuilder().append(
    orm.StructureData, filters={'attributes.kinds': {'contains': [kind1, kind2]}}
)

structures = q.all(flat=True)
# Those are structures that contain Au and Cu
# In my case: AuCu', 'AuCu4Gd', 'C6Au3Cu3N6', 'AuCu4Dy', 'AuCuZn2',

In principle, you don’t need to use the Kind objects and could simply define the necessary dicts yourself. I used them, as they automatically insert the mass in the dict resulting from .get_raw(), so that you don’t have to look it up.

One thing to note: the weights attribute is set to 1.0 per default. So in case you have structures\kinds with weight unequal to 1, you would need to take this into account.

1 Like