Phase Retrieval Problem

Алгоритмы восстановления фазы


from svetlanna.phase_retrieval_problem import ...

Классы

SetupLike

Наследует: Protocol

A class for phase_retrieval_problem with personal realizations of forward and reverse methods instead of methods in svetlanna.setup.LinearOpticalSetup

📥 Параметры

Параметр	Тип	Описание
`Protocol`	`_type_`	description

Методы

forward


forward(self, input_field: torch.Tensor) -> torch.Tensor

reverse


reverse(self, transmission_field: torch.Tensor) -> torch.Tensor

AlgorithmOptions

Наследует: TypedDict

PhaseRetrievalResult

Represents the phase retrieval result

Функции

`gerchberg_saxton_algorithm(target_intensity: torch.Tensor, source_intensity: torch.Tensor, forward: Callable, reverse: Callable, initial_approximation: torch.Tensor, tol: float, maxiter: int, target_phase: torch.Tensor | None = None, target_region: torch.Tensor | None = None) -> prr.PhaseRetrievalResult`

Gerchberg-Saxton algorithm(GS) for solving the phase retrieval problem

📥 Параметры

Параметр	Тип	Описание
`target_intensity`	`torch.Tensor`	Intensity profile in the Fourier plane
`source_intensity`	`torch.Tensor`	Intensity distribution before the optical setup(in the image plane)
`forward`	`Callable`	Function which describes forward propagation through the optical system
`reverse`	`Callable`	Function which describes reverse propagation through the optical system
`initial_approximation`	`torch.Tensor`	Initial approximation for the phase profile
`tol`	`float`	Accuracy for the algorithm
`maxiter`	`int`	Maximum number of iterations
`target_phase`	`torch.Tensor	None, optional`
`target_region`	`torch.Tensor	None, optional`

📤 Возвращает

prr.PhaseRetrievalResult

Exemplar of class PhaseRetrievalResult which presents result of optimization

`hybrid_input_output(target_intensity: torch.Tensor, source_intensity: torch.Tensor, forward: Callable, reverse: Callable, initial_approximation: torch.Tensor, tol: float, maxiter: int, target_phase: torch.Tensor | None = None, target_region: torch.Tensor | None = None, constant_factor: float = 0.9) -> prr.PhaseRetrievalResult`

Hybrid Input-Output(HIO) algorithm for for solving the phase retrieval problem

📥 Параметры

Параметр	Тип	Описание
`target_intensity`	`torch.Tensor`	Intensity profile in the Fourier plane
`source_intensity`	`torch.Tensor`	Intensity distribution before the optical setup(in the image plane)
`forward`	`Callable`	Function which describes forward propagation through the optical system
`reverse`	`Callable`	Function which describes reverse propagation through the optical system
`initial_approximation`	`torch.Tensor`	Initial approximation for the phase profile
`tol`	`float`	Accuracy for the algorithm
`maxiter`	`int`	Maximum number of iterations
`target_phase`	`torch.Tensor	None, optional`
`target_region`	`torch.Tensor	None, optional`
`None`	`constant_factor: float`	Learning rate value for the HIO algorithm, by default 0.9

📤 Возвращает

prr.PhaseRetrievalResult

Exemplar of class PhaseRetrievalResult which presents result of optimization

`retrieve_phase(source_intensity: torch.Tensor, optical_setup: LinearOpticalSetup | SetupLike, target_intensity: torch.Tensor) -> prr.PhaseRetrievalResult`

`retrieve_phase(source_intensity: torch.Tensor, optical_setup: LinearOpticalSetup | SetupLike, target_intensity: torch.Tensor, target_phase: torch.Tensor, target_region: torch.Tensor) -> prr.PhaseRetrievalResult`

`retrieve_phase(source_intensity: torch.Tensor, optical_setup: LinearOpticalSetup | SetupLike, target_intensity: torch.Tensor, target_phase: torch.Tensor | None = None, target_region: torch.Tensor | None = None) -> prr.PhaseRetrievalResult`

Function for solving phase retrieval problem: generating target intensity profile or reconstructing the phase profile of the field

📥 Параметры

Параметр	Тип	Описание
`source_intensity`	`torch.Tensor`	Intensity distribution before the optical setup
`optical_setup`	`LinearOpticalSetup	SetupLike`
`target_intensity`	`torch.Tensor`	Intensity profile in the Fourier plane
`target_phase`	`torch.Tensor	None, optional`
`target_region`	`torch.Tensor, optional`	Region to preserve phase and amplitude profiles in the Fourier plane( optional for the generating target intensity profile problem)
`initial_phase`	`torch.Tensor, optional`	Initial approximation for the phase profile, by default None
`method`	`Literal['GS', 'HIO'], optional`	Algorithms for phase retrieval problem, by default ‘GS’
`options`	`dict, optional`	Dictionary with optimization parameters, by default { ‘tol’: 1e-16, # criteria for stop optimization ‘maxiter’: 100, # maximum number of iterations ‘constant_factor’: 0.9, # convergence parameter for HIO ‘disp’: False # show result of optimization }

📤 Возвращает

prr.PhaseRetrievalResult

Exemplar of class PhaseRetrievalResult which presents result of optimization

⚠️ Исключения

ValueError — Unknown optimization method