Source code for metatrain.soap_bpnn.documentation
"""
SOAP-BPNN
=========
This is a Behler-Parrinello type neural network :footcite:p:`behler_generalized_2007`,
which, instead of their original atom-centered symmetry functions, we use the Smooth
overlap of atomic positions (SOAP) :footcite:p:`bartok_representing_2013` as the atomic
descriptors, computed with `torch-spex <https://github.com/lab-cosmo/torch-spex>`_.
{{SECTION_INSTALLATION}}
{{SECTION_DEFAULT_HYPERS}}
{{SECTION_MODEL_HYPERS}}
with the following definitions needed to fully understand some of the parameters:
.. autoclass:: {{architecture_path}}.documentation.SOAPConfig
:members:
:undoc-members:
.. autoclass:: {{architecture_path}}.documentation.SOAPCutoffConfig
:members:
:undoc-members:
.. autoclass:: {{architecture_path}}.documentation.BPNNConfig
:members:
:undoc-members:
"""
from typing import Literal, Optional
from typing_extensions import TypedDict
from metatrain.utils.additive.composition import FixedCompositionWeights
from metatrain.utils.hypers import init_with_defaults
from metatrain.utils.long_range import LongRangeHypers
from metatrain.utils.loss import LossSpecification
from metatrain.utils.scaler.scaler import FixedScalerWeights
[docs]
class SOAPCutoffConfig(TypedDict):
"""Cutoff configuration for the SOAP descriptor."""
radius: float = 5.0
"""Should be set to a value after which most of interatomic is expected
to be negligible. Note that the values should be defined in the position
units of your dataset."""
width: float = 0.5
"""The radial cutoff of atomic environments is performed smoothly, over
another distance defined by this parameter."""
[docs]
class SOAPConfig(TypedDict):
"""Configuration for the SOAP descriptors."""
max_angular: int = 6
"""Maximum angular channels of the spherical harmonics when
computing the SOAP descriptors."""
max_radial: int = 7
"""Maximum radial channels of the spherical harmonics when
computing the SOAP descriptors."""
cutoff: SOAPCutoffConfig = init_with_defaults(SOAPCutoffConfig)
"""Determines the cutoff routine of the atomic environment."""
[docs]
class BPNNConfig(TypedDict):
"""Configuration for the BPNN architecture."""
num_hidden_layers: int = 2
"""Controls the depth of the neural network. Increasing this generally leads
to better accuracy from the increased descriptivity, but comes at the cost
of increased training and evaluation time."""
num_neurons_per_layer: int = 32
"""Controls the width of the neural network. Increasing this generally leads
to better accuracy from the increased descriptivity, but comes at the cost
of increased training and evaluation time."""
layernorm: bool = True
"""Whether to use layer normalization before the neural network. Setting this
hyperparameter to false will lead to slower convergence of training, but
might lead to better generalization outside of the training set distribution.
"""
class ModelHypers(TypedDict):
"""Hyperparameters for the SOAP + BPNN architecture."""
soap: SOAPConfig = init_with_defaults(SOAPConfig)
"""Configuration of the SOAP descriptors."""
legacy: bool = True
"""If true, uses the legacy implementation without chemical embedding and with one
MLP head per atomic species."""
bpnn: BPNNConfig = init_with_defaults(BPNNConfig)
"""Configuration of the neural network architecture."""
add_lambda_basis: bool = True
"""This boolean parameter controls whether or not to add a spherical
expansion term of the same angular order as the targets, when they are
tensorial."""
heads: dict[str, Literal["mlp", "linear"]] = {}
"""The type of head (“linear” or “mlp”) to use for each target
(e.g. heads: {"energy": "linear", "mtt::dipole": "mlp"}). All omitted
targets will use a MLP (multi-layer perceptron) head. MLP heads consists
of one hidden layer with as many neurons as the
SOAP-BPNN (see ``BPNNConfig.num_neurons_per_layer``)."""
zbl: bool = False
"""Whether to use the ZBL short-range repulsion as the baseline for the model.
May be needed to achieve better description at the close-contact, repulsive regime.
"""
long_range: LongRangeHypers = init_with_defaults(LongRangeHypers)
"""Parameters related to long-range interactions.
May be needed to describe important long-range effects not captured by
the short-range SOAP-BPNN model"""
class TrainerHypers(TypedDict):
"""Hyperparameters for training SOAP BPNN models."""
distributed: bool = False
"""Whether to use distributed training"""
distributed_port: int = 39591
"""Port for distributed communication among processes"""
batch_size: int = 8
"""The number of samples to use in each batch of training. This
hyperparameter controls the tradeoff between training speed and memory usage. In
general, larger batch sizes will lead to faster training, but might require more
memory."""
num_epochs: int = 100
"""Number of epochs."""
warmup_fraction: float = 0.01
"""Fraction of training steps used for learning rate warmup."""
learning_rate: float = 1e-3
"""Learning rate."""
log_interval: int = 5
"""Interval to log metrics."""
checkpoint_interval: int = 25
"""Interval to save checkpoints."""
atomic_baseline: FixedCompositionWeights = {}
"""The baselines for each target.
By default, ``metatrain`` will fit a linear model (:class:`CompositionModel
<metatrain.utils.additive.composition.CompositionModel>`) to compute the
least squares baseline for each atomic species for each target.
However, this hyperparameter allows you to provide your own baselines.
The value of the hyperparameter should be a dictionary where the keys are the
target names, and the values are either (1) a single baseline to be used for
all atomic types, or (2) a dictionary mapping atomic types to their baselines.
For example:
- ``atomic_baseline: {"energy": {1: -0.5, 6: -10.0}}`` will fix the energy
baseline for hydrogen (Z=1) to -0.5 and for carbon (Z=6) to -10.0, while
fitting the baselines for the energy of all other atomic types, as well
as fitting the baselines for all other targets.
- ``atomic_baseline: {"energy": -5.0}`` will fix the energy baseline for
all atomic types to -5.0.
- ``atomic_baseline: {"mtt:dos": 0.0}`` sets the baseline for the "mtt:dos"
target to 0.0, effectively disabling the atomic baseline for that target.
This atomic baseline is substracted from the targets during training, which
avoids the main model needing to learn atomic contributions, and likely makes
training easier. When the model is used in evaluation mode, the atomic baseline
is added on top of the model predictions automatically.
.. note::
This atomic baseline is a per-atom contribution. Therefore, if the property
you are predicting is a sum over all atoms (e.g., total energy), the
contribution of the atomic baseline to the total property will be the
atomic baseline multiplied by the number of atoms of that type in the
structure.
"""
scale_targets: bool = True
"""Normalize targets to unit std during training."""
fixed_scaling_weights: FixedScalerWeights = {}
"""Weights for target scaling.
This is passed to the ``fixed_weights`` argument of
:meth:`Scaler.train_model <metatrain.utils.scaler.scaler.Scaler.train_model>`,
see its documentation to understand exactly what to pass here.
"""
per_structure_targets: list[str] = []
"""Targets to calculate per-structure losses."""
num_workers: Optional[int] = None
"""Number of workers for data loading. If not provided, it is set
automatically."""
log_mae: bool = False
"""Log MAE alongside RMSE"""
log_separate_blocks: bool = False
"""Log per-block error."""
best_model_metric: Literal["rmse_prod", "mae_prod", "loss"] = "rmse_prod"
"""Metric used to select best checkpoint (e.g., ``rmse_prod``)"""
loss: str | dict[str, LossSpecification | str] = "mse"
"""This section describes the loss function to be used. See the
:ref:`loss-functions` for more details."""
batch_atom_bounds: list[Optional[int]] = [None, None]
"""Bounds for the number of atoms per batch as [min, max]. Batches with atom
counts outside these bounds will be skipped during training. Use ``None`` for
either value to disable that bound. This is useful for preventing out-of-memory
errors and ensuring consistent computational load. Default: ``[None, None]``."""