set_sigma

Contents

set_sigma#

autoplex.fitting.common.regularization.set_sigma(atoms, reg_minmax, isolated_atoms_energies=None, scheme='linear-hull', energy_name='REF_energy', force_name='REF_forces', virial_name='REF_virial', element_order=None, max_energy=20.0, config_type_override=None)[source]#

Handle automatic regularisation based on distance to convex hull, amongst other things.

TODO: Need to make sure this works for full multi-stoichiometry systems.

Parameters:

  • atoms ((list of ase.Atoms)) – list of atoms objects to set reg. for. Usually fitting database

  • reg_minmax ((list of tuples)) – list of tuples of (min, max) values for energy, force, virial sigmas

  • scheme ((str)) – method to use for regularization. Options are: linear_hull, volume-stoichiometry linear_hull: for single-composition system, use 2D convex hull (E, V) volume-stoichiometry: for multi-composition system, use 3D convex hull of (E, V, mole-fraction)

  • energy_name ((str)) – name of energy key in atoms.info

  • force_name ((str)) – name of force key in atoms.arrays

  • virial_name ((str)) – name of virial key in atoms.info

  • isolated_atoms_energies ((dict)) – dictionary of isolated energies for each atomic number. Only needed for volume-x scheme e.g. {14: ‘-163.0’, 8:’-75.0’} for SiO2

  • element_order ((list)) – list of atomic numbers in order of choice (e.g. [42, 16] for MoS2)

  • max_energy ((float)) – ignore any structures with energy above hull greater than this value (eV)

  • config_type_override ((dict)) – give custom regularization for specific config types

  • [(0.1 (e.g. reg_minmax =) –

  • 1) –

  • (0.001 –

  • 0.1) –

  • (0.0316 –

  • 0.316) –

  • (0.0632 –

  • 0.632)] –

  • [(emin –

  • emax) –

  • (semin –

  • semax) –

  • (sfmin –

  • sfmax) –

  • (ssvmin –

  • svmax)] –

Return type:

list[Atoms]

Contents