schrodinger.application.mlff.general_inference module

Functions for writing MLFF model output properties to Schrodinger Structure objects.

Property keys use r_mlff_* prefix with unit annotations following the matsci convention (e.g. r_mlff_Energy_(Hartree)). Jaguar-specific inference functions live in jaguar_inference.py.

schrodinger.application.mlff.general_inference.energy_property_key(energy_unit: MlffEnergyUnit) → str
Parameters:

energy_unit – The energy unit to embed in the property key.

Returns:

CT-level property key for the total energy.

schrodinger.application.mlff.general_inference.atomic_energy_property_key(energy_unit: MlffEnergyUnit) → str
Parameters:

energy_unit – The energy unit to embed in the property key.

Returns:

Atom-level property key for atomic energy.

schrodinger.application.mlff.general_inference.charge_property_key() → str
Returns:

Atom-level property key for predicted charge.

schrodinger.application.mlff.general_inference.force_property_key(dim: int, energy_unit: MlffEnergyUnit) → str
Parameters:

  • dim – Cartesian dimension index (0=X, 1=Y, 2=Z).

  • energy_unit – The energy unit to embed in the property key.

Returns:

Atom-level property key for force in the given dimension.

schrodinger.application.mlff.general_inference.virial_property_key(i: int, j: int, energy_unit: MlffEnergyUnit) → str
Parameters:

energy_unit – The energy unit to embed in the property key.

Returns:

CT-level property key for the (i, j) virial component.

schrodinger.application.mlff.general_inference.write_mlff_output_properties(model_output: MlffOutput, st: Structure, post_process: MlffPostProcessSteps) → None

Write MLFF model output properties to a Structure.

Modifies st in place, setting CT-level and atom-level properties with unit-annotated r_mlff_* keys. Forces and virial are written only when the corresponding flags in post_process are set.

Parameters:

  • model_output – Output from an MLFF model inference.

  • st – Structure to write properties to (modified in place).

  • post_process – The post-processing steps used for this inference, providing energy unit and flags for which outputs were computed.