AlphaNet

Optional dependency

AlphaNetCalculator requires the alphanet extra plus torch-scatter built for the same PyTorch version. See Extras Needing an Additional Install Step for the verified install sequence.

uvpip

uv add "materialsframework[alphanet]"
uv pip install torch==2.9.1 --index-url https://download.pytorch.org/whl/cpu
uv pip install torch-scatter -f https://data.pyg.org/whl/torch-2.9.1+cpu.html --reinstall

pip install "materialsframework[alphanet]"
pip install torch==2.9.1 --index-url https://download.pytorch.org/whl/cpu
pip install torch-scatter -f https://data.pyg.org/whl/torch-2.9.1+cpu.html --force-reinstall

AlphaNetCalculator

AlphaNetCalculator(
    model: str,
    config: str,
    device: Literal["cuda", "cpu", "mps"] = "cpu",
    precision: Literal["32", "64"] = "32",
    **kwargs: Any,
)

Calculator for material property calculations and structure relaxation using the AlphaNet potential.

Attributes:

Name	Type	Description
AVAILABLE_PROPERTIES	list[str]	A list of properties that this calculator can compute, including "energy", "forces", and "stresses".

References

• AlphaNet: https://doi.org/10.48550/arXiv.2501.07155

Initializes the AlphaNetCalculator with the specified model and calculation settings.

Parameters:

Name	Type	Description	Default
model	str	The path to the model checkpoint file.	required
config	str	The path to the configuration file for the AlphaNet model.	required
device	Literal['cuda', 'cpu', 'mps']	The device to use for calculations. Defaults to "cpu".	'cpu'
precision	Literal['32', '64']	The precision of the calculations. Defaults to "32".	'32'
**kwargs	Any	Additional keyword arguments passed to the BaseCalculator and BaseMDCalculator constructors.	{}

calculator

calculator() -> Calculator

Lazily builds the ASE Calculator object for the AlphaNet potential, using the settings from initialization.

Returns:

Name	Type	Description
Calculator	Calculator	The ASE Calculator object configured with the AlphaNet potential.