svetlanna.elements package

Submodules

svetlanna.elements.aperture module

class svetlanna.elements.aperture.Aperture(simulation_parameters: SimulationParameters, mask: Tensor | Parameter | Parameter)

Bases: MulElement

Aperture of the optical element with transmission function, which takes the value 0 or 1

get_transmission_function() → Tensor

Method which returns the transmission function of the element

Returns

torch.Tensor: transmission function

to_specs() → Iterable[ParameterSpecs]: Create specs

property transmission_function_axes: tuple[str, ...]

Axes of the transmission function. For example, (‘H’, ‘W’)

Returns

tuple[str, …]: Axes

class svetlanna.elements.aperture.MulElement(simulation_parameters: SimulationParameters)

Bases: Element, ABC

Class that generalize all elements with E->T@E like forward function, where T is transmission function

forward(incident_wavefront: Wavefront) → Wavefront

Calculate the field after propagating through the element

Parameters

input_fieldWavefront: Field incident on the aperture

Returns

Wavefront: The field after propagating through the element

abstract get_transmission_function() → Tensor

Method which returns the transmission function of the element

Returns

torch.Tensor: transmission function

abstract property transmission_function_axes: tuple[str, ...]

Axes of the transmission function. For example, (‘H’, ‘W’)

Returns

tuple[str, …]: Axes

class svetlanna.elements.aperture.RectangularAperture(simulation_parameters: SimulationParameters, height: float, width: float)

Bases: MulElement

A rectangle-shaped aperture with a transmission function taking either a value of 0 or 1

get_transmission_function() → Tensor

Method which returns the transmission function of the element

Returns

torch.Tensor: transmission function

to_specs() → Iterable[ParameterSpecs]: Create specs

property transmission_function_axes: tuple[str, ...]

Axes of the transmission function. For example, (‘H’, ‘W’)

Returns

tuple[str, …]: Axes

class svetlanna.elements.aperture.RoundAperture(simulation_parameters: SimulationParameters, radius: float)

Bases: MulElement

A round-shaped aperture with a transmission function taking either a value of 0 or 1

get_transmission_function() → Tensor

Method which returns the transmission function of the element

Returns

torch.Tensor: transmission function

to_specs() → Iterable[ParameterSpecs]: Create specs

property transmission_function_axes: tuple[str, ...]

Axes of the transmission function. For example, (‘H’, ‘W’)

Returns

tuple[str, …]: Axes

svetlanna.elements.diffractive_layer module

class svetlanna.elements.diffractive_layer.DiffractiveLayer(simulation_parameters: SimulationParameters, mask: Tensor | Parameter | Parameter, mask_norm: float = 6.283185307179586)

Bases: Element

A class that described the field after propagating through the passive diffractive layer with a given phase mask

forward(incident_wavefront: Wavefront) → Wavefront

Method that calculates the field after propagating through the SLM

Parameters

input_fieldWavefront: Field incident on the SLM

Returns

Wavefront: The field after propagating through the SLM

reverse(transmission_wavefront: Wavefront) → Wavefront

Method that calculates the field after passing the SLM in back propagation

Parameters

transmitted_fieldWavefront: Field incident on the SLM in back propagation (transmitted field in forward propagation)

Returns

Wavefront: Field transmitted on the SLM in back propagation (incident field in forward propagation)

to_specs() → Iterable[ParameterSpecs]: Create specs

property transmission_function: Tensor

svetlanna.elements.element module

class svetlanna.elements.element.Element(simulation_parameters: SimulationParameters)

Bases: Module

A class that describes each element of the system

Parameters

nn_type_: _description_
metaclass_type_, optional: _description_, by default ABCMeta

abstract forward(incident_wavefront: Wavefront) → Wavefront: Forward propagation through the optical element

make_buffer(name: str, value: _T, persistent: bool = False) → _T

Make buffer for internal use.

Use case: ` self.mask = make_buffer('mask', some_tensor) ` This allow torch to properly process .to method on Element by marking that mask should be transferred to required device.

Parameters

namestr: name of the new buffer (it is more convenient to use name of new attribute)
value_T: tensor to be buffered
persistentbool, optional: see torch docs on buffers, by default False

Returns

_T: the value passed to the method

process_parameter(name: str, value: _V) → _V

Process element parameter passed by user. Automatically registers buffer for non-parametric tensors.

Use case: ` self.mask = process_parameter('mask', some_tensor) `

Parameters

namestr: name of the new buffer (it is more convenient to use name of new attribute)
value_V: the value of the element parameter

Returns

_V: the value passed to the method

to_specs() → Iterable[ParameterSpecs | SubelementSpecs]: Create specs

svetlanna.elements.free_space module

class svetlanna.elements.free_space.FreeSpace(simulation_parameters: SimulationParameters, distance: float | Tensor | Parameter | Parameter, method: Literal['fresnel', 'AS'])

Bases: Element

A class that describes a propagation of the field in free space between two optical elements

forward(incident_wavefront: Wavefront) → Wavefront

Calculates the field after propagating in the free space

Parameters

input_fieldWavefront: Field before propagation in free space

Returns

Wavefront: Field after propagation in free space

Raises

ValueError: Occurs when a non-existent direct distribution method is chosen

impulse_response_angular_spectrum() → Tensor

Creates the impulse response function for angular spectrum method

Returns

torch.Tensor: 2d impulse response function for angular spectrum method

impulse_response_fresnel() → Tensor

Creates the impulse response function for fresnel approximation

Returns

torch.Tensor: 2d impulse response function for fresnel approximation

reverse(transmission_wavefront: Wavefront) → Wavefront

Calculate the field after it propagates in the free space in the backward direction.

Parameters

transmission_fieldWavefront: Field to be propagated in the backward direction

Returns

Wavefront: Propagated in the backward direction field

to_specs() → Iterable[ParameterSpecs]: Create specs

svetlanna.elements.lens module

class svetlanna.elements.lens.ThinLens(simulation_parameters: SimulationParameters, focal_length: float | Tensor | Parameter | Parameter, radius: float = inf)

Bases: Element

A class that described the field after propagating through the thin lens.

forward(incident_wavefront: Wavefront) → Wavefront

Calculates the field after propagation through the thin lens.

Parameters

input_fieldWavefront: The field incident on the thin lens.

Returns

Wavefront: The field after propagation through the thin lens.

get_transmission_function() → Tensor

Returns the transmission function of the thin lens.

Returns

torch.Tensor: The transmission function of the thin lens.

reverse(transmission_wavefront: Wavefront) → Wavefront

Calculates the field after passing through the lens during back propagation.

Parameters

transmission_fieldWavefront: The field incident on the lens during back propagation. This corresponds to the transmitted field in forward propagation.

Returns

Wavefront: The field transmitted through the lens during back propagation. This corresponds to the incident field in forward propagation.

to_specs() → Iterable[ParameterSpecs]: Create specs

property transmission_function: Tensor

svetlanna.elements.nonlinear_element module

class svetlanna.elements.nonlinear_element.FunctionModule(function: Callable[[Tensor], Tensor], function_parameters: Dict | None)

Bases: Module

A class for transforming an arbitrary function with multiple parameters. Allows training function parameters

forward(function_argument: Tensor) → Tensor

forward method for a class inherited from torch.nn.Module

Parameters

function_argumenttorch.Tensor: Argument of the function

Returns

torch.Tensor: Function with trainable parameters

class svetlanna.elements.nonlinear_element.NonlinearElement(simulation_parameters: SimulationParameters, response_function: Callable[[Tensor], Tensor], response_parameters: Dict | None = None)

Bases: Element

A class representing a nonlinear optical element with a given amplitude response function. Preserves the phase distribution of the incident wavefront

forward(incident_wavefront: Wavefront) → Wavefront

Method calculating the wavefront after passing a nonlinear optical element

Parameters

incident_wavefrontWavefront: Wavefront before the nonlinear optical element

Returns

Wavefront: Wavefront passing through a nonlinear optical element

svetlanna.elements.reservoir module

Bases: Element

Reservoir element.

append_feedback_queue(field: Wavefront)

Append a new wavefront to the feedback queue.

Parameters

fieldWavefront: The new wavefront to be added to the end of the queue.

drop_feedback_queue() → None: Clear all elements from the feedback queue.

forward(incident_wavefront: Wavefront) → Wavefront: Forward propagation through the optical element

pop_feedback_queue() → None | Wavefront

Retrieve and remove the first element from the feedback queue if available.

Parameters

fieldWavefront: The first wavefront in the queue, or None if the queue is empty or not full yet.

to_specs() → Iterable[ParameterSpecs | SubelementSpecs]: Create specs

svetlanna.elements.slm module

class svetlanna.elements.slm.SpatialLightModulator(simulation_parameters: ~svetlanna.simulation_parameters.SimulationParameters, mask: ~torch.Tensor, height: float | ~torch.Tensor | ~torch.nn.parameter.Parameter | ~svetlanna.parameters.Parameter, width: float | ~torch.Tensor | ~torch.nn.parameter.Parameter | ~svetlanna.parameters.Parameter, location: ~typing.Tuple = (0.0, 0.0), number_of_levels: int = 256, step_function: ~typing.Callable[[~torch.Tensor], ~torch.Tensor] = <function relu>, mode: ~typing.Literal['nearest', 'bilinear', 'bicubic', 'area', 'nearest-exact'] = 'nearest')

Bases: Element

A class that described the field after propagating through the Spatial Light Modulator with a given phase mask

Parameters

Element_type_: _description_

forward(incident_wavefront: Wavefront) → Wavefront: Forward propagation through the optical element

property get_aperture: Tensor

property resized_mask: Tensor

reverse(transmission_wavefront: Wavefront) → Wavefront

property transmission_function: Tensor

Module contents

class svetlanna.elements.Aperture(simulation_parameters: SimulationParameters, mask: Tensor | Parameter | Parameter)

Bases: MulElement

Aperture of the optical element with transmission function, which takes the value 0 or 1

get_transmission_function() → Tensor

Method which returns the transmission function of the element

Returns

torch.Tensor: transmission function

to_specs() → Iterable[ParameterSpecs]: Create specs

property transmission_function_axes: tuple[str, ...]

Axes of the transmission function. For example, (‘H’, ‘W’)

Returns

tuple[str, …]: Axes

class svetlanna.elements.DiffractiveLayer(simulation_parameters: SimulationParameters, mask: Tensor | Parameter | Parameter, mask_norm: float = 6.283185307179586)

Bases: Element

A class that described the field after propagating through the passive diffractive layer with a given phase mask

forward(incident_wavefront: Wavefront) → Wavefront

Method that calculates the field after propagating through the SLM

Parameters

input_fieldWavefront: Field incident on the SLM

Returns

Wavefront: The field after propagating through the SLM

reverse(transmission_wavefront: Wavefront) → Wavefront

Method that calculates the field after passing the SLM in back propagation

Parameters

transmitted_fieldWavefront: Field incident on the SLM in back propagation (transmitted field in forward propagation)

Returns

Wavefront: Field transmitted on the SLM in back propagation (incident field in forward propagation)

to_specs() → Iterable[ParameterSpecs]: Create specs

property transmission_function: Tensor

class svetlanna.elements.Element(simulation_parameters: SimulationParameters)

Bases: Module

A class that describes each element of the system

Parameters

nn_type_: _description_
metaclass_type_, optional: _description_, by default ABCMeta

abstract forward(incident_wavefront: Wavefront) → Wavefront: Forward propagation through the optical element

make_buffer(name: str, value: _T, persistent: bool = False) → _T

Make buffer for internal use.

Use case: ` self.mask = make_buffer('mask', some_tensor) ` This allow torch to properly process .to method on Element by marking that mask should be transferred to required device.

Parameters

namestr: name of the new buffer (it is more convenient to use name of new attribute)
value_T: tensor to be buffered
persistentbool, optional: see torch docs on buffers, by default False

Returns

_T: the value passed to the method

process_parameter(name: str, value: _V) → _V

Process element parameter passed by user. Automatically registers buffer for non-parametric tensors.

Use case: ` self.mask = process_parameter('mask', some_tensor) `

Parameters

namestr: name of the new buffer (it is more convenient to use name of new attribute)
value_V: the value of the element parameter

Returns

_V: the value passed to the method

to_specs() → Iterable[ParameterSpecs | SubelementSpecs]: Create specs

class svetlanna.elements.FreeSpace(simulation_parameters: SimulationParameters, distance: float | Tensor | Parameter | Parameter, method: Literal['fresnel', 'AS'])

Bases: Element

A class that describes a propagation of the field in free space between two optical elements

forward(incident_wavefront: Wavefront) → Wavefront

Calculates the field after propagating in the free space

Parameters

input_fieldWavefront: Field before propagation in free space

Returns

Wavefront: Field after propagation in free space

Raises

ValueError: Occurs when a non-existent direct distribution method is chosen

impulse_response_angular_spectrum() → Tensor

Creates the impulse response function for angular spectrum method

Returns

torch.Tensor: 2d impulse response function for angular spectrum method

impulse_response_fresnel() → Tensor

Creates the impulse response function for fresnel approximation

Returns

torch.Tensor: 2d impulse response function for fresnel approximation

reverse(transmission_wavefront: Wavefront) → Wavefront

Calculate the field after it propagates in the free space in the backward direction.

Parameters

transmission_fieldWavefront: Field to be propagated in the backward direction

Returns

Wavefront: Propagated in the backward direction field

to_specs() → Iterable[ParameterSpecs]: Create specs

class svetlanna.elements.FunctionModule(function: Callable[[Tensor], Tensor], function_parameters: Dict | None)

Bases: Module

A class for transforming an arbitrary function with multiple parameters. Allows training function parameters

forward(function_argument: Tensor) → Tensor

forward method for a class inherited from torch.nn.Module

Parameters

function_argumenttorch.Tensor: Argument of the function

Returns

torch.Tensor: Function with trainable parameters

class svetlanna.elements.NonlinearElement(simulation_parameters: SimulationParameters, response_function: Callable[[Tensor], Tensor], response_parameters: Dict | None = None)

Bases: Element

A class representing a nonlinear optical element with a given amplitude response function. Preserves the phase distribution of the incident wavefront

forward(incident_wavefront: Wavefront) → Wavefront

Method calculating the wavefront after passing a nonlinear optical element

Parameters

incident_wavefrontWavefront: Wavefront before the nonlinear optical element

Returns

Wavefront: Wavefront passing through a nonlinear optical element

class svetlanna.elements.RectangularAperture(simulation_parameters: SimulationParameters, height: float, width: float)

Bases: MulElement

A rectangle-shaped aperture with a transmission function taking either a value of 0 or 1

get_transmission_function() → Tensor

Method which returns the transmission function of the element

Returns

torch.Tensor: transmission function

to_specs() → Iterable[ParameterSpecs]: Create specs

property transmission_function_axes: tuple[str, ...]

Axes of the transmission function. For example, (‘H’, ‘W’)

Returns

tuple[str, …]: Axes

class svetlanna.elements.RoundAperture(simulation_parameters: SimulationParameters, radius: float)

Bases: MulElement

A round-shaped aperture with a transmission function taking either a value of 0 or 1

get_transmission_function() → Tensor

Method which returns the transmission function of the element

Returns

torch.Tensor: transmission function

to_specs() → Iterable[ParameterSpecs]: Create specs

property transmission_function_axes: tuple[str, ...]

Axes of the transmission function. For example, (‘H’, ‘W’)

Returns

tuple[str, …]: Axes

Bases: Element

Reservoir element.

append_feedback_queue(field: Wavefront)

Append a new wavefront to the feedback queue.

Parameters

fieldWavefront: The new wavefront to be added to the end of the queue.

drop_feedback_queue() → None: Clear all elements from the feedback queue.

forward(incident_wavefront: Wavefront) → Wavefront: Forward propagation through the optical element

pop_feedback_queue() → None | Wavefront

Retrieve and remove the first element from the feedback queue if available.

Parameters

fieldWavefront: The first wavefront in the queue, or None if the queue is empty or not full yet.

to_specs() → Iterable[ParameterSpecs | SubelementSpecs]: Create specs

class svetlanna.elements.SpatialLightModulator(simulation_parameters: ~svetlanna.simulation_parameters.SimulationParameters, mask: ~torch.Tensor, height: float | ~torch.Tensor | ~torch.nn.parameter.Parameter | ~svetlanna.parameters.Parameter, width: float | ~torch.Tensor | ~torch.nn.parameter.Parameter | ~svetlanna.parameters.Parameter, location: ~typing.Tuple = (0.0, 0.0), number_of_levels: int = 256, step_function: ~typing.Callable[[~torch.Tensor], ~torch.Tensor] = <function relu>, mode: ~typing.Literal['nearest', 'bilinear', 'bicubic', 'area', 'nearest-exact'] = 'nearest')

Bases: Element

A class that described the field after propagating through the Spatial Light Modulator with a given phase mask

Parameters

Element_type_: _description_

forward(incident_wavefront: Wavefront) → Wavefront: Forward propagation through the optical element

property get_aperture: Tensor

property resized_mask: Tensor

reverse(transmission_wavefront: Wavefront) → Wavefront

property transmission_function: Tensor

class svetlanna.elements.ThinLens(simulation_parameters: SimulationParameters, focal_length: float | Tensor | Parameter | Parameter, radius: float = inf)

Bases: Element

A class that described the field after propagating through the thin lens.

forward(incident_wavefront: Wavefront) → Wavefront

Calculates the field after propagation through the thin lens.

Parameters

input_fieldWavefront: The field incident on the thin lens.

Returns

Wavefront: The field after propagation through the thin lens.

get_transmission_function() → Tensor

Returns the transmission function of the thin lens.

Returns

torch.Tensor: The transmission function of the thin lens.

reverse(transmission_wavefront: Wavefront) → Wavefront

Calculates the field after passing through the lens during back propagation.

Parameters

transmission_fieldWavefront: The field incident on the lens during back propagation. This corresponds to the transmitted field in forward propagation.

Returns

Wavefront: The field transmitted through the lens during back propagation. This corresponds to the incident field in forward propagation.

to_specs() → Iterable[ParameterSpecs]: Create specs

property transmission_function: Tensor