jarvis.ai.descriptors package

Submodules

jarvis.ai.descriptors.cfid module

Classical Force-field Inspired Descriptors (CFID).

Find details in: https://journals.aps.org/prmaterials/abstract/10.1103/PhysRevMaterials.2.083801

class jarvis.ai.descriptors.cfid.CFID(atoms)

Bases: object

Convert Atoms class into 1557 descriptors.

get_comp_descp(jcell=True, jmean_chem=True, jmean_chg=True, jrdf=False, jrdf_adf=True, print_names=False)

Get chemo-structural CFID decriptors.

Args:

struct: Structure object

jcell: whether to use cell-size descriptors

jmean_chem: whether to use average chemical descriptors

jmean_chg: whether to use average charge distribution descriptors

jmean_rdf: whether to use radial distribution descriptors

jrdf_adf: whether to use radial and angle distribution descriptors

print_names: whether to print names of descriptors

Returns:

cat: catenated final descriptors

jarvis.ai.descriptors.cfid.feat_names()

Names of the 1557 descriptors.

jarvis.ai.descriptors.coulomb module

Coulomb matrix for Atoms.

Refer to: 10.1103/PhysRevLett.108.058301

jarvis.ai.descriptors.coulomb.coulomb_matrix(atoms='', max_dim=100)

Convert Atoms class to max_dim x max_dim matrix.

Args:

atoms: atoms object

max_dim: maximum number of atoms=sqrt(max_dim)

Returns:

z: numpy array of 1 x max_dim dimension

Module contents

Modules to obtain representations for ML.