scimba_torch.neural_nets.structure_preserving_nets.symplectic_nets

Symplectic neural networks composed of invertible layers.

Classes

GSymplecticNet(size, conditional_size, ...)	An invertible neural network composed of multiple invertible layers.
LASymplecticNet(size, conditional_size, ...)	An invertible neural network composed of multiple invertible layers.
PSymplecticNet(size, conditional_size, deg, ...)	An invertible neural network composed of multiple invertible layers.
PeriodicGSymplecticNet(size, ...)	An invertible neural network composed of multiple invertible layers.

class GSymplecticNet(size, conditional_size, width, nb_layers, **kwargs)

Bases: InvertibleNet

An invertible neural network composed of multiple invertible layers.

Parameters:

• size (int) – dimension of the input data

• conditional_size (int) – dimension of the conditional input data

• width (int) – width of the hidden layers in each layer

• nb_layers (int) – number of invertible layers to compose

• **kwargs – other arguments for the invertible layers

log_abs_det_jacobian(y, mu)

Computes the log absolute determinant of the Jacobian.

Parameters:

• y (Tensor) – Input tensor of shape (batch_size, size).

• mu (Tensor) – Conditional input of shape (batch_size, conditional_size).

Return type:

Tensor

Returns:

Log determinant of shape (batch_size,).

abs_det_jacobian(y, mu)

Computes the absolute determinant of the Jacobian.

Parameters:

• y (Tensor) – Input tensor of shape (batch_size, size).

• mu (Tensor) – Conditional input of shape (batch_size, conditional_size).

Return type:

Tensor

Returns:

Determinant of shape (batch_size,).

class PeriodicGSymplecticNet(size, conditional_size, width, nb_layers, **kwargs)

Bases: InvertibleNet

An invertible neural network composed of multiple invertible layers.

Parameters:

• size (int) – dimension of the input data

• conditional_size (int) – dimension of the conditional input data

• width (int) – width of the hidden layers in each layer

• nb_layers (int) – number of invertible layers to compose

• **kwargs – other arguments for the invertible layers

forward(inputs)

Applies all layers in sequence.

Parameters:

inputs (Tensor) – Input tensor of shape (2,batch_size, size).

Return type:

Tensor

Returns:

Transformed tensor of shape (batch_size, size).

backward(inputs)

Applies all layers in reverse order.

Parameters:

inputs (Tensor) – Input tensor of shape (2, batch_size, size).

Return type:

Tensor

Returns:

Inverse transformed tensor of shape (batch_size, size).

log_abs_det_jacobian(y, mu)

Computes the log absolute determinant of the Jacobian.

Parameters:

• y (Tensor) – Input tensor of shape (batch_size, size).

• mu (Tensor) – Conditional input of shape (batch_size, conditional_size).

Return type:

Tensor

Returns:

Log determinant of shape (batch_size,).

abs_det_jacobian(y, mu)

Computes the absolute determinant of the Jacobian.

Parameters:

• y (Tensor) – Input tensor of shape (batch_size, size).

• mu (Tensor) – Conditional input of shape (batch_size, conditional_size).

Return type:

Tensor

Returns:

Determinant of shape (batch_size,).

class LASymplecticNet(size, conditional_size, width, nb_layers, **kwargs)

Bases: InvertibleNet

An invertible neural network composed of multiple invertible layers.

Parameters:

• size (int) – dimension of the input data

• conditional_size (int) – dimension of the conditional input data

• width (int) – width of the hidden layers in each layer

• nb_layers (int) – number of invertible layers to compose

• **kwargs – other arguments for the invertible layers

log_abs_det_jacobian(y, mu)

Computes the log absolute determinant of the Jacobian.

Parameters:

• y (Tensor) – Input tensor of shape (batch_size, size).

• mu (Tensor) – Conditional input of shape (batch_size, conditional_size).

Return type:

Tensor

Returns:

Log determinant of shape (batch_size,).

abs_det_jacobian(y, mu)

Computes the absolute determinant of the Jacobian.

Parameters:

• y (Tensor) – Input tensor of shape (batch_size, size).

• mu (Tensor) – Conditional input of shape (batch_size, conditional_size).

Return type:

Tensor

Returns:

Determinant of shape (batch_size,).

class PSymplecticNet(size, conditional_size, deg, nb_layers, **kwargs)

Bases: InvertibleNet

An invertible neural network composed of multiple invertible layers.

Parameters:

• size (int) – dimension of the input data

• conditional_size (int) – dimension of the conditional input data

• deg (int) – degree of the polynomial transformation

• nb_layers (int) – number of invertible layers to compose

• **kwargs – other arguments for the invertible layers

log_abs_det_jacobian(y, mu)

Computes the log absolute determinant of the Jacobian.

Parameters:

• y (Tensor) – Input tensor of shape (batch_size, size).

• mu (Tensor) – Conditional input of shape (batch_size, conditional_size).

Return type:

Tensor

Returns:

Log determinant of shape (batch_size,).

abs_det_jacobian(y, mu)

Computes the absolute determinant of the Jacobian.

Parameters:

• y (Tensor) – Input tensor of shape (batch_size, size).

• mu (Tensor) – Conditional input of shape (batch_size, conditional_size).

Return type:

Tensor

Returns:

Determinant of shape (batch_size,).