nos.physics package

Submodules

nos.physics.helmholtz_residual module

class nos.physics.helmholtz_residual.HelmholtzDomainMSE

Bases: Module

forward(y: Tensor, v: Tensor, k: Tensor) → Tensor

Define the computation performed at every call.

Should be overridden by all subclasses.

Note

Although the recipe for forward pass needs to be defined within this function, one should call the Module instance afterwards instead of this since the former takes care of running the registered hooks while the latter silently ignores them.

class nos.physics.helmholtz_residual.HelmholtzDomainResidual

Bases: Module

forward(y: Tensor, v: Tensor, k: Tensor) → Tensor

Define the computation performed at every call.

Should be overridden by all subclasses.

Note

Although the recipe for forward pass needs to be defined within this function, one should call the Module instance afterwards instead of this since the former takes care of running the registered hooks while the latter silently ignores them.

nos.physics.laplace module

class nos.physics.laplace.Laplace(shapes: OperatorShapes | None = None, device: device | None = None)

Bases: Operator

forward(x: Tensor, u: Tensor, y: Tensor = None) → Tensor

Forward pass through the operator.

Parameters:

• x – Sensor positions of shape (batch_size, x_dim, num_sensors…).

• u – Input function values of shape (batch_size, u_dim, num_sensors…).

• y – Evaluation coordinates of shape (batch_size, y_dim, num_evaluations…).

Returns:

Evaluations of the mapped function with shape (batch_size, v_dim, num_evaluations…).

nos.physics.weight_scheduler_lin module

class nos.physics.weight_scheduler_lin.WeightSchedulerLinear(pde_loss_start: int, pde_loss_full: int)

Bases: Module

forward(epoch: float | tensor) → tensor

Define the computation performed at every call.

Should be overridden by all subclasses.

Note

Although the recipe for forward pass needs to be defined within this function, one should call the Module instance afterwards instead of this since the former takes care of running the registered hooks while the latter silently ignores them.

Module contents

class nos.physics.HelmholtzDomainMSE

Bases: Module

forward(y: Tensor, v: Tensor, k: Tensor) → Tensor

Define the computation performed at every call.

Should be overridden by all subclasses.

Note

Although the recipe for forward pass needs to be defined within this function, one should call the Module instance afterwards instead of this since the former takes care of running the registered hooks while the latter silently ignores them.

class nos.physics.HelmholtzDomainResidual

Bases: Module

forward(y: Tensor, v: Tensor, k: Tensor) → Tensor

Define the computation performed at every call.

Should be overridden by all subclasses.

Note

Although the recipe for forward pass needs to be defined within this function, one should call the Module instance afterwards instead of this since the former takes care of running the registered hooks while the latter silently ignores them.

class nos.physics.Laplace(shapes: OperatorShapes | None = None, device: device | None = None)

Bases: Operator

forward(x: Tensor, u: Tensor, y: Tensor = None) → Tensor

Forward pass through the operator.

Parameters:

• x – Sensor positions of shape (batch_size, x_dim, num_sensors…).

• u – Input function values of shape (batch_size, u_dim, num_sensors…).

• y – Evaluation coordinates of shape (batch_size, y_dim, num_evaluations…).

Returns:

Evaluations of the mapped function with shape (batch_size, v_dim, num_evaluations…).

class nos.physics.WeightSchedulerLinear(pde_loss_start: int, pde_loss_full: int)

Bases: Module

forward(epoch: float | tensor) → tensor

Define the computation performed at every call.

Should be overridden by all subclasses.

Note

Although the recipe for forward pass needs to be defined within this function, one should call the Module instance afterwards instead of this since the former takes care of running the registered hooks while the latter silently ignores them.