Phonon Optimized Potentials

An example of this file is shown below:

THIS LINE IS IGNORED
  generations: 1
  extrainds: 0
  population_size: 1
  mutation_rate: 0.2
  elite_percentage: 0.05
  tabulated_parameters: 0
  nontabulated_parameters: 10
  M: 5
  N: 0
  force_weight: 0.2
  energy_weight: 0
  energy_shape_weight: 0.4
  stress_weight: 0.4
  pressure_weight: 0
  pressure_shape_weight: 0
  intervals:                   0 4    0 1        0 50    0.01 10     0.01 10      0 50      0 50       0.01 50    1e-3 5    1e-3 5 
  decimal_place_accuracies:     2      2         2        2             2           2       2            2         2         2          
----------------------------------------LAMMPS SETUP COMMANDS
    log none  
    neighbor 1.0 multi  
    boundary p p p  
    units metal  
    atom_style hybrid atomic charge
    atom_modify map array  
    ### Declare the box
    lattice diamond 5.46
    region box prism 0 1 0 1 0 1 0 0 0 units lattice  
    create_box 1 box  
    ### Create atom type 1 in the box
    create_atoms 1 random 17 1919191 NULL # Positions don't matter... just declare the memory
    mass 1 69.7 #Ge
    compute P all pe  
    compute S all pressure thermo_temp  
----------------------------------------LAMMPS PAIR STYLE COMMANDS

  pair_style hybrid/overlay morse [] born 12 coul/dsf 0.25 12
  set type 1 charge []
  pair_coeff 1 1 coul/dsf
  pair_coeff 1 1 born [] [] [] [] []
  pair_coeff 1 1 morse [] [] []

----------------------------------------END OF FILE

The format of this file must be exactly as shown. All the fields in the first section are as follows:

generations: number of generations for the genetic algorithm

extrainds: number of extra individuals to add to the population, for refitting purposes

population_size: number of individuals for the genetic algorithm

mutation_rate: decimal probability to perform mutations

elite_percentage: decimal percent of individuals to have in the elite portion of a random selection algorithm

tabulated_parameters: number of parameters in a tabulated potential file

nontabulated_parameters: number of parameters in a LAMMPS input script, that are not in a tabulated file

M: number of configurations

N: number of atoms

force_weight: weighting factor for fitting forces

energy_weight: weighting factor for fitting energies

energy_shape_weight: weighting factor for fitting energy shape, which is total energy minus the energy of the lowest energy configuration. This ensures that the lowest energy has a value of zero.

stress_weight: weighting factor for fitting stresses

pressure_weight: weighting factor for fitting pressure

pressure_shape_weight: same as energy shape, except applied to pressure

—————————————-LAMMPS SETUP COMMANDS

This section includes all LAMMPS commands. Declare your neighbor skin distance, units, atom styles, and simply declare your box. You also tneed to declare the memory for atom types - at this point it doesn't mater where they are placed so just use the create_atoms random command (the positions will be updated as configurations are read). Finally, you need to include computes for potential energy and stress (see the above example).

—————————————-LAMMPS PAIR STYLE COMMANDS These are normal LAMMPS pair style commands, except you replace a fitting parameter with a bracketed term. Leave the terms that you don't want to fit as numbers.

Make sure the sum of all the bracketed terms corresponds with the sum of nontabulated_parameters and tabulated_parameters. In this example, we don't have any tabulated parameters.

If you are fitting a tabulated potential, fill the potential file with the bracketed terms ([]) in the necessary places and name the file “REFPOT”. Then label the potential file in this section as “TEMPPOT{}”. The “{}” symbols will ensure that this potential is written and evaluated in a parallel fashion.