pair_style agni command

pair_style agni/omp command


pair_style agni


pair_style agni pair_coeff * * Al.agni Al


Style agni style computes the manybody vectorial force components for an atom as


u labels the individual components, i.e. x, y or z, and V is the corresponding atomic fingerprint. d is the Euclidean distance between any two atomic fingerprints. A total of N_t reference atomic environments are considered to construct the force field file. alpha_t and l are the weight coefficients and length scale parameter of the non-linear regression model.

The method implements the recently proposed machine learning access to atomic forces as discussed extensively in the following publications - (Botu1) and (Botu2). The premise of the method is to map the atomic environment numerically into a fingerprint, and use machine learning methods to create a mapping to the vectorial atomic forces.

Only a single pair_coeff command is used with the agni style which specifies an AGNI potential file containing the parameters of the force field for the needed elements. These are mapped to LAMMPS atom types by specifying N additional arguments after the filename in the pair_coeff command, where N is the number of LAMMPS atom types:

  • filename
  • N element names = mapping of AGNI elements to atom types

See the pair_coeff doc page for alternate ways to specify the path for the force field file.

An AGNI force field is fully specified by the filename which contains the parameters of the force field, i.e., the reference training environments used to construct the machine learning force field. Example force field and input files are provided in the examples/USER/misc/agni directory.

Styles with omp suffix is functionally the same as the corresponding style without the suffix. They have been optimized to run faster, depending on your available hardware, as discussed in Section 5 of the manual. The accelerated style takes the same arguments and should produce the same results, except for round-off and precision issues.

The accelerated style is part of the USER-OMP. They are only enabled if LAMMPS was built with those packages. See the Making LAMMPS section for more info.

You can specify the accelerated style explicitly in your input script by including their suffix, or you can use the -suffix command-line switch when you invoke LAMMPS, or you can use the suffix command in your input script.

See Section 5 of the manual for more instructions on how to use the accelerated styles effectively.

Mixing, shift, table, tail correction, restart, rRESPA info:

This pair style does not support the pair_modify shift, table, and tail options.

This pair style does not write its information to binary restart files, since it is stored in potential files. Thus, you need to re-specify the pair_style and pair_coeff commands in an input script that reads a restart file.

This pair style can only be used via the pair keyword of the run_style respa command. It does not support the inner, middle, outer keywords.


Currently, only elemental systems are implemented. Also, the method only provides access to the forces and not energies or stresses. However, one can access the energy via thermodynamic integration of the forces as discussed in (Botu3). This pair style is part of the USER-MISC package. It is only enabled if LAMMPS was built with that package. See the Making LAMMPS section for more info.

The AGNI force field files provided with LAMMPS (see the potentials directory) are parameterized for metal units. You can use the AGNI potential with any LAMMPS units, but you would need to create your own AGNI potential file with coefficients listed in the appropriate units if your simulation doesn’t use “metal” units.