Generate a list of SMACT-allowed compositions

In [1]:

    

### Imports
from smact import Element, element_dictionary, ordered_elements
from smact.screening import smact_filter
from datetime import datetime
import itertools
import multiprocessing


    

In [2]:

    

all_el = element_dictionary()   # A dictionary of all element objects

# Say we are just interested in first row transition metals
els = [all_el[symbol] for symbol in ordered_elements(21,30)]

# We can print the symbols 
print([i.symbol for i in els])


    

    




['Sc', 'Ti', 'V', 'Cr', 'Mn', 'Fe', 'Co', 'Ni', 'Cu', 'Zn']

In [3]:

    

# Generate all M1-M2 combinations
metal_pairs = itertools.combinations(els, 2)
# Add O to each pair 
ternary_systems = [[*m, Element('O')] for m in metal_pairs]
# Prove to ourselves that we have all unique chemical systems
for i in ternary_systems:
    print(i[0].symbol, i[1].symbol, i[2].symbol)


    

    




Sc Ti O
Sc V O
Sc Cr O
Sc Mn O
Sc Fe O
Sc Co O
Sc Ni O
Sc Cu O
Sc Zn O
Ti V O
Ti Cr O
Ti Mn O
Ti Fe O
Ti Co O
Ti Ni O
Ti Cu O
Ti Zn O
V Cr O
V Mn O
V Fe O
V Co O
V Ni O
V Cu O
V Zn O
Cr Mn O
Cr Fe O
Cr Co O
Cr Ni O
Cr Cu O
Cr Zn O
Mn Fe O
Mn Co O
Mn Ni O
Mn Cu O
Mn Zn O
Fe Co O
Fe Ni O
Fe Cu O
Fe Zn O
Co Ni O
Co Cu O
Co Zn O
Ni Cu O
Ni Zn O
Cu Zn O

In [4]:

    

# Use multiprocessing and smact_filter to quickly generate our list of compositions
start = datetime.now()
if __name__ == '__main__':   # Always use pool protected in an if statement 
    with multiprocessing.Pool(processes=4) as p:    # start 4 worker processes
        result = p.map(smact_filter, ternary_systems)
print('Time taken to generate list:  {0}'.format(datetime.now()-start))


    

    




Time taken to generate list:  0:00:00.448839

In [5]:

    

# Flatten the list of lists
flat_list = [item for sublist in result for item in sublist]
print('Number of compositions: --> {0} <--'.format(len(flat_list)))
print('Each list entry looks like this:\n  elements, oxidation states, stoichiometries')
for i in flat_list[:5]:
    print(i)


    

    




Number of compositions: --> 16615 <--
Each list entry looks like this:
  elements, oxidation states, stoichiometries
(('Sc', 'Ti', 'O'), (1, -1, -2), (3, 1, 1))
(('Sc', 'Ti', 'O'), (1, -1, -2), (4, 2, 1))
(('Sc', 'Ti', 'O'), (1, -1, -2), (5, 1, 2))
(('Sc', 'Ti', 'O'), (1, -1, -2), (5, 3, 1))
(('Sc', 'Ti', 'O'), (1, -1, -2), (6, 4, 1))

In [6]:

    

from pymatgen import Composition
def comp_maker(comp):
    form = []
    for el, ammt in zip(comp[0], comp[2]):
        form.append(el)
        form.append(ammt)
    form = ''.join(str(e) for e in form)
    pmg_form = Composition(form).reduced_formula
    return pmg_form

if __name__ == '__main__':  
    with multiprocessing.Pool(processes=4) as p:
        pretty_formulas = p.map(comp_maker, flat_list)

print('Each list entry now looks like this: ')
for i in pretty_formulas[:5]:
    print(i)


    

    




Each list entry now looks like this: 
Sc3TiO
Sc4Ti2O
Sc5TiO2
Sc5Ti3O
Sc6Ti4O

In [7]:

    

import pandas as pd
new_data = pd.DataFrame({'pretty_formula': pretty_formulas})
# Drop any duplicate compositions
new_data = new_data.drop_duplicates(subset = 'pretty_formula')
new_data.describe()


    

    
Out[7]:







  

    

      

      
pretty_formula
    
  
  

    

      
count
      
9353
    
    

      
unique
      
9353
    
    

      
top
      
Ti3V4O7
    
    

      
freq
      
1