6.4. DeepMDIO Input/Output interfaces

Warning

DeepMD kit is installed by pharmaforge; however, additonal libraries may be required for specific applications.

class pharmaforge.interfaces.deepmdio.DeepMDData(e, f)[source]

Bases: object

This class is used to store the energy and forces calculated by DeepMD.

This is a simple data class that holds the energy and forces of the system. It is used to store the results of the DeepMD calculations and is returned by the DeepMDInterface class, to have similar attributes to outcomes of ASE calculators.

Parameters:

e (float) – The energy of the system.
f (list) – The forces acting on the atoms in the system.

e

The energy of the system.

Type:: float

f

The forces acting on the atoms in the system.

Type:: list

Methods

get_forces
get_potential_energy

get_forces()[source]

get_potential_energy()[source]

class pharmaforge.interfaces.deepmdio.DeepMDInterface(model=None)[source]

Bases: AbstractIO

This class is used to interface with DeepMD-kit for running machine learning potentials, and will be used to calculate the energies and forces of the system.

Note

This interface only takes a model as the option, and thus does not have default options.

Parameters:: model (str) – The path to the DeepMD model file. This should be a .pb file.

calculator

The DeepMD calculator object.

Type:: object

model

The path to the DeepMD model file.

Type:: str

allow_parallel

Whether to allow parallel calculations. Default is False, ans currently cannot pass the threads to DeepMD.

Type:: bool

See also

pharmaforge.interfaces.abstractio.AbstractIO: The abstract interface class for the Psi4Interface.
deepmd.infer.DeepPot: The deepmd-kit inference calculator object.

Methods

`ObtainDeepData`(filename)	Obtain the system data from the DeepIO interface and converts each configuration to an ASE atoms object.
`ObtainQueryData`(queryresults)	Obtain the system data from a query
`calculate`([limit, verbose, num_workers])	Calculate the energies and forces of the system using the interface.
`find_clusters`(structure[, cutoff])	Identify clusters of atoms in a structure based on a distance cutoff.

process_structure