LightningODE_FixedPotential

class scdiffeq.core.lightning_models._lightning_ode_fixed_potential.LightningODE_FixedPotential(latent_dim: int = 50, name: str | None = None, dt: float = 0.1, mu_hidden: ~typing.List[int] | int = [400, 400], mu_activation: str | ~typing.List[str] = 'LeakyReLU', mu_dropout: float | ~typing.List[float] = 0.2, mu_bias: bool = True, mu_output_bias: bool = True, mu_n_augment: int = 0, train_lr=0.0001, train_optimizer=<class 'torch.optim.rmsprop.RMSprop'>, train_scheduler=<class 'torch.optim.lr_scheduler.StepLR'>, train_step_size: int = 10, backend: str = 'auto', adjoint=False, loading_existing: bool = False, *args, **kwargs)

Bases: BaseForwardMixIn, PotentialMixIn, BaseLightningDiffEq

LightningODE-FixedPotential

Parameters:

latent_dimint, optional

Dimensionality of the latent space, by default 50

namestr, optional

Name of the model, by default None

dtfloat, optional

Time step for the ODE solver, by default 0.1

mu_hiddenUnion[List[int], int], optional

Hidden layer sizes for the neural network, by default [400, 400]

mu_activationUnion[str, List[str]], optional

Activation function(s) for the neural network, by default “LeakyReLU”

mu_dropoutUnion[float, List[float]], optional

Dropout rate(s) for the neural network, by default 0.2

mu_biasbool, optional

Whether to use bias in the neural network, by default True

mu_output_biasbool, optional

Whether to use bias in the output layer of the neural network, by default True

mu_n_augmentint, optional

Number of augmentations for the neural network, by default 0

train_lrfloat, optional

Learning rate for training, by default 1e-4

train_optimizertorch.optim.Optimizer, optional

Optimizer for training, by default torch.optim.RMSprop

train_schedulertorch.optim.lr_scheduler._LRScheduler, optional

Learning rate scheduler for training, by default torch.optim.lr_scheduler.StepLR

train_step_sizeint, optional

Step size for the learning rate scheduler, by default 10

backendstr, optional

Backend for the ODE solver, by default “auto”

adjointbool, optional

Whether to use the adjoint method for the ODE solver, by default False

loading_existingbool, optional

Whether to load an existing model, by default False

Returns:

None

Notes:

This class implements a fixed potential ODE using PyTorch Lightning.

Examples:

>>> model = LightningODE_FixedPotential(latent_dim=20, dt=0.05)
>>> model.fit(data)

__init__(latent_dim: int = 50, name: str | None = None, dt: float = 0.1, mu_hidden: ~typing.List[int] | int = [400, 400], mu_activation: str | ~typing.List[str] = 'LeakyReLU', mu_dropout: float | ~typing.List[float] = 0.2, mu_bias: bool = True, mu_output_bias: bool = True, mu_n_augment: int = 0, train_lr=0.0001, train_optimizer=<class 'torch.optim.rmsprop.RMSprop'>, train_scheduler=<class 'torch.optim.lr_scheduler.StepLR'>, train_step_size: int = 10, backend: str = 'auto', adjoint=False, loading_existing: bool = False, *args, **kwargs) → None

LightningODE_FixedPotential

Parameters:

latent_dimint, optional

Dimensionality of the latent space, by default 50

namestr, optional

Name of the model, by default None

dtfloat, optional

Time step for the ODE solver, by default 0.1

mu_hiddenUnion[List[int], int], optional

Hidden layer sizes for the neural network, by default [400, 400]

mu_activationUnion[str, List[str]], optional

Activation function(s) for the neural network, by default “LeakyReLU”

mu_dropoutUnion[float, List[float]], optional

Dropout rate(s) for the neural network, by default 0.2

mu_biasbool, optional

Whether to use bias in the neural network, by default True

mu_output_biasbool, optional

Whether to use bias in the output layer of the neural network, by default True

mu_n_augmentint, optional

Number of augmentations for the neural network, by default 0

train_lrfloat, optional

Learning rate for training, by default 1e-4

train_optimizertorch.optim.Optimizer, optional

Optimizer for training, by default torch.optim.RMSprop

train_schedulertorch.optim.lr_scheduler._LRScheduler, optional

Learning rate scheduler for training, by default torch.optim.lr_scheduler.StepLR

train_step_sizeint, optional

Step size for the learning rate scheduler, by default 10

backendstr, optional

Backend for the ODE solver, by default “auto”

adjointbool, optional

Whether to use the adjoint method for the ODE solver, by default False

loading_existingbool, optional

Whether to load an existing model, by default False

Returns:

None

Notes:

This class implements a fixed potential ODE using PyTorch Lightning.

Examples:

>>> model = LightningODE_FixedPotential(latent_dim=20, dt=0.05)
>>> model.fit(data)