Ptychography’s additional classes

Basic Parameter Structure

class proxtoolbox.Problems.ptychography.Pty(phi, obj, probe)[source]

Bases: object

In the Ptychography problem phi, the object and the probe do not have the same dimensions. This class combines them into a single data structure. Some standard operators are overloaded for convenience.

Initialization

Parameters:
phi : array_like

The p-th diffraction pattern

obj : array_like

The object

probe : array_like

The probe
__add__(p)[source]

Elementwise addition

__div__(t)[source]

Division

__getitem__(key)[source]

Getter for an item of phi

__init__(phi, obj, probe)[source]

Initialization

Parameters:
phi : array_like

The p-th diffraction pattern

obj : array_like

The object

probe : array_like

The probe
__rmul__(t)[source]

Multiplication

__sub__(p)[source]

Elementwise subtraction

__weakref__

list of weak references to the object (if defined)

copy()[source]

Creates a copy of a Pty-instance

ProxOperators

class proxtoolbox.Problems.ptychography.P_rodenburg(config)[source]

Bases: proxtoolbox.ProxOperators.proxoperators.ProxOperator

Rodenburg’s Ptychography algorithm

Notes

Done by
  • Pär Mattsson, Universität Göttingen, 2013
  • Matthew Tam, CARMA Centre, University of Newcastle, 2014

Initialization

__init__(config)[source]

Initialization

work(u)[source]
Parameters:
u : Pty - Ptychography has special needs, so u must be a Pty.
Returns:
Pty - The output iterate

See also

Pty

class proxtoolbox.Problems.ptychography.P_rodenburg_probe_fixed(config)[source]

Bases: proxtoolbox.Problems.ptychography.P_rodenburg

Rodenburg’s Ptychography algorithm

Initialization

work(u)[source]
Parameters:
u : Pty - Input
class proxtoolbox.Problems.ptychography.P_thibault_f(config)[source]

Bases: proxtoolbox.ProxOperators.proxoperators.ProxOperator

Fourier magnitude update for the Thibault et al. algorithm

Initialization

__init__(config)[source]

Initialization

work(u)[source]
Parameters:
u : Pty - Input
class proxtoolbox.Problems.ptychography.P_thibault_o(config)[source]

Bases: proxtoolbox.ProxOperators.proxoperators.ProxOperator

The object update for the Thibault et al. algorithm, with probe fixed

Initialization

__init__(config)[source]

Initialization

work(u)[source]
Parameters:
u : Pty - Input
class proxtoolbox.Problems.ptychography.P_thibault_op(config)[source]

Bases: proxtoolbox.ProxOperators.proxoperators.ProxOperator

The simultaneous probe and object update for the Thibault et al. algorithm

Initialization

__init__(config)[source]

Initialization

work(u)[source]
Parameters:
u : Pty - Input
class proxtoolbox.Problems.ptychography.P_PHeBIE_probe(config)[source]

Bases: proxtoolbox.ProxOperators.proxoperators.ProxOperator

Probe update for the PHeBIE algorithm

Initialization

__init__(config)[source]

Initialization

work(u)[source]
Parameters:
u : Pty - Input
class proxtoolbox.Problems.ptychography.P_PHeBIE_probe_ptwise(config)[source]

Bases: proxtoolbox.Problems.ptychography.P_PHeBIE_probe

Probe update for the PHeBIE algorithm

Initialization

work(u)[source]
Parameters:
u : Pty - Input
class proxtoolbox.Problems.ptychography.P_PHeBIE_object(config)[source]

Bases: proxtoolbox.ProxOperators.proxoperators.ProxOperator

Object update for the PHeBIE algorithm

Initialization

__init__(config)[source]

Initialization

work(u)[source]
Parameters:
u : Pty - Input
class proxtoolbox.Problems.ptychography.P_PHeBIE_object_ptwise(config)[source]

Bases: proxtoolbox.Problems.ptychography.P_PHeBIE_object

Object update for the PHeBIE algorithm

Initialization

work(u)[source]
Parameters:
u : Pty - Input
class proxtoolbox.Problems.ptychography.P_PHeBIE_phi(config)[source]

Bases: proxtoolbox.ProxOperators.proxoperators.ProxOperator

Phi update for the PHeBIE algorithm

Initialization

__init__(config)[source]

Initialization

work(u)[source]
Parameters:
u : Pty - Input
class proxtoolbox.Problems.ptychography.P_PHeBIE_phi_regularized(config)[source]

Bases: proxtoolbox.ProxOperators.proxoperators.ProxOperator

Regularized phi update for the PHeBIE algorithm

Initialization

__init__(config)[source]

Initialization

work(u)[source]
Parameters:
u : Pty - Input

Statistics

class proxtoolbox.Problems.ptychography.PtychographyStats(config)[source]

Bases: object

The Ptychography algorithms have special statistics. The necessary routines are collected in this class.

Initialization

__init__(config)[source]

Initialization

__weakref__

list of weak references to the object (if defined)

_dftregistration(buf1ft, buf2ft, usfac=1)[source]

Efficient subpixel image registration by crosscorrelation.

Parameters:
buf1ft : array_like - Fourier transform of the reference image
buf2ft : array_like - Fourier transform of the image to register
usfac : int, optional - Upsampling factor
Returns:
error : number - Translation invariant normalized RMS error between f and g
diffphase : number - Global phase difference between the two images
row_shift : number - Pixel shifts between images (rows)
col_shift : number - Pixel shifts between images (columns)

References

[1]Manuel Guizar-Sicairos, Samuel T. Thurman, James R. Fienup, “Efficient subpixel image registration algorithms”, Opt.Lett. 33, 156-158 (2008)
_dftups(inp, nor=None, noc=None, usfac=1, roff=0, coff=0)[source]

Upsampled DFT by matrix multiplications

Parameters:
inp : array_like - Input data
nor, noc : int, optional - Number of pixels in the output unsampled DFT (Default = inp.shape)
usfac : number, optional - Upsampling factor (Default = 1)
roff, coff : int, optional - Row and column offsets (Default = 0)
Returns:
ndarray - DFT
change(x1, x2)[source]

Computes the change for x1 and x2

Parameters:
x1 : Pty - Input 1
x2 : Pty - Input 2
Returns:
float - The change
customerror(x1, x2, x3=None)[source]

Computes the custom error for x1, x2 and x3

Parameters:
x1 : Pty - Input 1
x2 : Pty - Input 2
x3 : Pty, optional - Input 3
Returns:
(rms_error_obj, rms_error_probe, change_phi, r_factor, objective) : tuple - The error
objective(x1)[source]

Get the value of the objective function (used in PALM)

Parameters:
x1 : Pty - Input
Returns:
float - Objective function value of x1

Blocking

class proxtoolbox.Problems.ptychography.Blocking_meta(config)[source]

Bases: proxtoolbox.Algorithms.algorithms.Algorithm

A meta-algorithm used if a blocking scheme is applied. Although this is probably not the most efficient implementation if blocking schemes, it can be applied independently of the algorithm.

Initialization

__getstate__()[source]

a multiprocessing workaround

__init__(config)[source]

Initialization

__setstate__(state)[source]

a multiprocessing workaround

_do_block(u, b, tol)[source]

Creates and solves the subproblem for the current block

Parameters:
u : Pty - Input data
b : int - Current block’s index
tol : number - Tolerance
Returns:
sequence of array_like - A tuple of the form (phi,obj,probe)
_do_view(u, view, tol)[source]

Creates and solves the subproblem for the current view

Parameters:
u : Pty - Input data
view : array_like - Current view
tol : number - Tolerance
Returns:
sequence of array_like - A tuple of the form (phi,obj,probe)
run(u, tol, maxiter)[source]

Runs Blocking_meta

ptychography._unwrap_do_block(**kwarg)
ptychography._unwrap_do_view(**kwarg)

Algorithms

class proxtoolbox.Problems.ptychography.PALM(config)[source]

Bases: proxtoolbox.Algorithms.algorithms.Algorithm

Proximal alternating (linearized) minimization algorithm

Parameters:
config : dict

Dictionary containing the problem configuration. It must contain the following mappings:

projector_sequence: sequence of Proxoperator

Sequence of prox operators to be used. (The classes, no instances)

Nx: int

Row-dim of the product space elements

Ny: int

Column-dim of the product space elements

Nz: int

Depth-dim of the product space elements

dim: int

Size of the product space

normM: number

?

ignore_error: boolean

Whether to ignore errors
__init__(config)[source]
Parameters:
config : dict

Dictionary containing the problem configuration. It must contain the following mappings:

projector_sequence: sequence of Proxoperator

Sequence of prox operators to be used. (The classes, no instances)

Nx: int

Row-dim of the product space elements

Ny: int

Column-dim of the product space elements

Nz: int

Depth-dim of the product space elements

dim: int

Size of the product space

normM: number

?

ignore_error: boolean

Whether to ignore errors
run(u, tol, maxiter)[source]

Runs the algorithm one some input data

Parameters:
u : array_like - Input data
tol : number - Tolerance
maxiter : int - Maximum number of iterations
Returns:
u : array_like - Result
u_final : array_like - Result
iters : int - Number of iterations the algorithm performed
change : array_like - The percentage change in the norm
gap : array_like - Squared gap distance normalized by the magnitude constraint
class proxtoolbox.Problems.ptychography.RAAR_PALM(config)[source]

Bases: proxtoolbox.Algorithms.algorithms.Algorithm

RAAR implementation that produces error statistics comparable to PALM

Parameters:
config : dict

Dictionary containing the problem configuration. It must contain the following mappings:

projector_sequence: sequence of ProxOperator

Sequence of prox operators to be used. (The classes, no instances). For RAAR_PALM it must contain 2 ProxOperators.

Nx: int

Row-dim of the product space elements

Ny: int

Column-dim of the product space elements

Nz: int

Depth-dim of the product space elements

dim: int

Size of the product space

beta0: number

Starting relaxation parmater

beta_max: number

Maximum relaxation parameter

beta_switch: int

Iteration at which beta moves from beta0 -> beta_max

normM: number

?

ignore_error: boolean

Whether to ignore errors
__init__(config)[source]
Parameters:
config : dict

Dictionary containing the problem configuration. It must contain the following mappings:

projector_sequence: sequence of ProxOperator

Sequence of prox operators to be used. (The classes, no instances). For RAAR_PALM it must contain 2 ProxOperators.

Nx: int

Row-dim of the product space elements

Ny: int

Column-dim of the product space elements

Nz: int

Depth-dim of the product space elements

dim: int

Size of the product space

beta0: number

Starting relaxation parmater

beta_max: number

Maximum relaxation parameter

beta_switch: int

Iteration at which beta moves from beta0 -> beta_max

normM: number

?

ignore_error: boolean

Whether to ignore errors
run(u, tol, maxiter)[source]

Runs the algorithm one some input data

Parameters:
u : array_like - Input data
tol : number - Tolerance
maxiter : int - Maximum number of iterations
Returns:
tmp_u1 : array_like - Result
tmp_u1.copy() : array_like - Copy of result
iters : int - Number of iterations the algorithm performed
change : array_like - The percentage change in the norm
custom_errors : array_like - Squared gap distance normalized by the magnitude constraint