Ctrl+K

exitwave

exitwave#

Functions related to creating and manipulating ptychographic exitwave arrays.

Ptychographic exitwaves are stored as a single complex array which represent the wavefield after any and all interaction with the sample and thus there’s just free space propagation to the detector.

class tike.ptycho.exitwave.ExitWaveOptions[source]#

Bases: object

Manage data and setting related to exitwave updates.

New in version 0.25.0.

__init__(measured_pixels, noise_model='gaussian', step_length_weight=0.5, step_length_usemodes='all_modes', step_length_start=0.5, unmeasured_pixels_scaling=1.0, propagation_normalization='ortho')#
Parameters

  • measured_pixels (numpy.typing.NDArray.numpy.bool_) –

  • noise_model (str) –

  • step_length_weight (float) –

  • step_length_usemodes (str) –

  • step_length_start (float) –

  • unmeasured_pixels_scaling (float) –

  • propagation_normalization (str) –

Return type

None

copy_to_device(comm)[source]#

Copy to the current GPU memory.

Return type

ExitWaveOptions

copy_to_host()[source]#

Copy to the host CPU memory.

Return type

ExitWaveOptions

measured_pixels: numpy.typing.NDArray.numpy.bool_#

A boolean array that defines spatial regions on the detector where we have measured pixels. False for bad pixels and True for good pixels.

noise_model: str = 'gaussian'#

‘gaussian’ OR ‘poisson’ The noise model for the exitwave updates

propagation_normalization: str = 'ortho'#

Choose the scaling of the FFT in the forward model:

“ortho” - the forward and adjoint operations are divided by sqrt(n)

“forward” - the forward operation is divided by n

“backward” - the backard operation is divided by n

resample(factor)[source]#

Return a new ExitWaveOptions with the parameters rescaled.

Parameters

factor (float) –

Return type

ExitWaveOptions

step_length_start: float = 0.5#

We use an iterative/recursive method for finding the steplengths, and this is what we use as initialization for that method.

step_length_usemodes: str = 'all_modes'#

When computing steplength for exitwave updates, there are two ways we do this:

“dominant_mode” - use the dominant mode to compute the steplength and use that steplength for the other less dominant modes

“all_modes” - compute the steplength for each mode independently

step_length_weight: float = 0.5#

When computing steplength, we use a weighted average of the previous calculated step and the current calculated step, options are 0.0 <= step_length_weight <= 1.0, with being closer to 1.0 favoring the current calculated step

unmeasured_pixels_scaling: float = 1.0#

Depending on how we control scaling for the exitwaves, we might need to scale up or down the number of photons in the unmeasured regions for the exitwave updates in Fourier space. 1.0 for no scaling.

tike.ptycho.exitwave.crop_fourier_space(x, w)[source]#

Crop x assuming a 2D frequency space image with zero frequency in corner.

Parameters

  • x (numpy.ndarray) –

  • w (int) –

Return type

numpy.ndarray

tike.ptycho.exitwave.poisson_steplength_all_modes(xi, abs2_Psi, I_e, I_m, measured_pixels, step_length, weight_avg)[source]#

Compute the optimal steplength for each exitwave mode independently.

Parameters

  • xi ((FRAME, 1, 1, WIDE, HIGH) float32) – xi = 1 - I_m / I_e

  • abs2_Psi ((FRAME, 1, SHARED, WIDE, HIGH ) float32) – the squared absolute value of the calulated exitwaves

  • I_m ((FRAME, WIDE, HIGH) float32) – measured diffraction intensity

  • I_e ((FRAME, WIDE, HIGH) float32) – calculated diffraction intensity

  • measured_pixels ((WIDE, HIGH) float32) – the regions on the detector where we have defined measurements

  • step_length ((FRAME, 1, SHARED, 1, 1) float32) – the steplength initializations

  • weight_avg (float) – the weight we use when computing a weighted average with ( 0.0 <= weight_avg <= 1.0 )

tike.ptycho.exitwave.poisson_steplength_dominant_mode(xi, I_e, I_m, measured_pixels, step_length, weight_avg)[source]#

Compute the optimal steplength for each exitwave mode using only the dominant mode.

Parameters

  • xi ((FRAME, 1, 1, WIDE, HIGH) float32) – xi = 1 - I_m / I_e

  • I_m ((FRAME, WIDE, HIGH) float32) – measured diffraction intensity

  • I_e ((FRAME, WIDE, HIGH) float32) – calculated diffraction intensity

  • measured_pixels ((WIDE, HIGH) float32) – the regions on the detector where we have defined measurements

  • step_length ((FRAME, 1, SHARED, 1, 1) float32) – the steplength initializations

  • weight_avg (float) – the weight we use when computing a weighted average with ( 0.0 <= weight_avg <= 1.0 )

Contents