MACEParameterOptions

Included in QATK.MLFF

class MACEParameterOptions

Deprecated since version 2026.03: MACEParameterOptions is deprecated and will be removed in a future release. Use MLParameterOptions instead.

MACE-specific parameter options container (deprecated). All attributes trigger deprecation warnings.

For new code, use MLParameterOptions which provides:

  • DEVICE (AUTOMATIC, CPU, GPU)

  • DTYPE (FLOAT32, FLOAT64)

  • TASK_TYPE (GENERAL, ATOM_WISE)

  • POOLING (SUM, MEAN, MAX)

class DEVICE

Device options for ML training and inference.

class DTYPE

Data type precision options.

class POOLING

Pooling methods for aggregation.

class TASK_TYPE

Task type for property prediction.

Notes

Deprecated since version 2026.03: MACEParameterOptions is deprecated and will be removed in a future release. Use MLParameterOptions instead, which provides the same functionality with a framework-agnostic design.

The MACEParameterOptions class provides predefined options for various MACE training parameters. These options are organized into nested classes, each representing a specific parameter category.

Migration Guide: Replace MACEParameterOptions with MLParameterOptions in your code. The API is identical for DEVICE, DTYPE, TASK_TYPE, and POOLING options.

Available Option Groups

DEVICE

Controls which hardware device is used for training:

  • MACEParameterOptions.DEVICE.AUTOMATIC - Automatically selects GPU if available, otherwise CPU

  • MACEParameterOptions.DEVICE.CPU - Use CPU for training

  • MACEParameterOptions.DEVICE.GPU - Use GPU (CUDA) for training

DTYPE

Specifies the floating-point precision for model parameters:

  • MACEParameterOptions.DTYPE.FLOAT32 - 32-bit floating point (faster, less memory)

  • MACEParameterOptions.DTYPE.FLOAT64 - 64-bit floating point (more accurate)

TASK_TYPE

Defines the type of prediction task:

  • MACEParameterOptions.TASK_TYPE.GENERAL - Global, per-configuration property prediction

  • MACEParameterOptions.TASK_TYPE.ATOM_WISE - Atom-wise property prediction

POOLING

Specifies the pooling method for aggregating atomic contributions:

  • MACEParameterOptions.POOLING.SUM - Sum atomic contributions

  • MACEParameterOptions.POOLING.MEAN - Average atomic contributions

  • MACEParameterOptions.POOLING.MAX - Take maximum of atomic contributions

Usage Example

training_parameters = TrainingParameters(
    experiment_name='my_experiment',
    default_dtype=MACEParameterOptions.DTYPE.FLOAT64,
    device=MACEParameterOptions.DEVICE.AUTOMATIC,
)