06 – CPM Interferometric Intensity Constraint¶

This tutorial demonstrates an extended forward model for reflection ptychography that accounts for interferometric contributions from a reference plane. It shows how the modified intensity constraint improves reconstruction of multi-layer samples.

What you'll learn:

• How to apply an interferometric intensity constraint in the forward model
• How to reconstruct a two-layer reflection dataset (USAF target at 700 nm)
• How the extended model compares to the standard intensity constraint

!!! note "Dataset" The dataset (USAF_speckle_bin4.hdf5) is downloaded automatically into the example_data/ folder at the project root the first time you run the notebook. If the file already exists it is not re-downloaded.

!!! tip "Prerequisites" Complete 05 – CPM Reflection before this tutorial.

In [31]:

import matplotlib
import matplotlib.pyplot as plt
import numpy as np
from pathlib import Path
import h5py

import matplotlib import matplotlib.pyplot as plt import numpy as np from pathlib import Path import h5py

In [32]:

# import the PtyLab module
import PtyLab
from PtyLab import ExperimentalData
from PtyLab import Reconstruction
from PtyLab import Monitor
from PtyLab import Params
from PtyLab import Engines
from PtyLab.utils.utils import posit

# import the PtyLab module import PtyLab from PtyLab import ExperimentalData from PtyLab import Reconstruction from PtyLab import Monitor from PtyLab import Params from PtyLab import Engines from PtyLab.utils.utils import posit

The dataset (USAF_speckle_bin4.hdf5) is downloaded automatically into the example_data/ folder at the project root the first time you run the next cell. If the file already exists it is not re-downloaded.

In [33]:

from PtyLab.io import getExampleDataFolder
import urllib.request

fileName = "USAF_speckle_bin4.hdf5"
filePath = getExampleDataFolder() / fileName

if not filePath.exists():
    print(f"Downloading {fileName}...")
    urllib.request.urlretrieve(
        "https://ndownloader.figshare.com/files/28713885",
        filePath,
    )
    print("Download complete.")
else:
    print(f"{fileName} found, skipping download.")

from PtyLab.io import getExampleDataFolder import urllib.request fileName = "USAF_speckle_bin4.hdf5" filePath = getExampleDataFolder() / fileName if not filePath.exists(): print(f"Downloading {fileName}...") urllib.request.urlretrieve( "https://ndownloader.figshare.com/files/28713885", filePath, ) print("Download complete.") else: print(f"{fileName} found, skipping download.")

USAF_speckle_bin4.hdf5 found, skipping download.

In [34]:

experimentalData = ExperimentalData(filePath, operationMode="CPM")
experimentalData.showPtychogram()

experimentalData = ExperimentalData(filePath, operationMode="CPM") experimentalData.showPtychogram()

Found encoder with shape (102, 2)
Min max ptychogram: 0.9099568724632263, 16222.2021484375
Min max ptychogram: 0.2810235917568207, 4.210136890411377
0.2810236 4.210137

VBox(children=(IntSlider(value=0, description='Frame', max=101), Output()))

In [35]:

experimentalData.zo = 25.02e-3
experimentalData.entrancePupilDiameter = (
    0.4e-3  # exampleData.Np / 3 * exampleData.dxp  # initial estimate of beam size
)
backgroundOffset = 20
experimentalData.ptychogram = posit(experimentalData.ptychogram - backgroundOffset)
# experimentalData.showPtychogram()

experimentalData.zo = 25.02e-3 experimentalData.entrancePupilDiameter = ( 0.4e-3 # exampleData.Np / 3 * exampleData.dxp # initial estimate of beam size ) backgroundOffset = 20 experimentalData.ptychogram = posit(experimentalData.ptychogram - backgroundOffset) # experimentalData.showPtychogram()

In [36]:

# Set monitor properties
monitor = Monitor()
monitor.figureUpdateFrequency = 5
monitor.objectPlot = "complex"  # complex abs angle
monitor.verboseLevel = "low"  # high: plot two figures, low: plot only one figure
monitor.probeZoom = 1  # control probe plot FoV
monitor.objectZoom = 2  # control object plot FoV
monitor.objectPlotContrast = 0.8
monitor.probePlotContrast = 0.5

# Set monitor properties monitor = Monitor() monitor.figureUpdateFrequency = 5 monitor.objectPlot = "complex" # complex abs angle monitor.verboseLevel = "low" # high: plot two figures, low: plot only one figure monitor.probeZoom = 1 # control probe plot FoV monitor.objectZoom = 2 # control object plot FoV monitor.objectPlotContrast = 0.8 monitor.probePlotContrast = 0.5

In [37]:

# Set the reconstruction parameters
params = Params()
## switches
params.gpuSwitch = True  # this is also autodetected by default, but you can set it to False to force CPU usage
params.positionOrder = "random"  # 'sequential' or 'random'
params.propagator = (
    "Fraunhofer"  # Fraunhofer Fresnel ASP scaledASP polychromeASP scaledPolychromeASP
)
params.probePowerCorrectionSwitch = True
params.comStabilizationSwitch = True
params.fftshiftSwitch = False
params.backgroundModeSwitch = True

# Set the reconstruction parameters params = Params() ## switches params.gpuSwitch = True # this is also autodetected by default, but you can set it to False to force CPU usage params.positionOrder = "random" # 'sequential' or 'random' params.propagator = ( "Fraunhofer" # Fraunhofer Fresnel ASP scaledASP polychromeASP scaledPolychromeASP ) params.probePowerCorrectionSwitch = True params.comStabilizationSwitch = True params.fftshiftSwitch = False params.backgroundModeSwitch = True

INFO:GPU:cupy and CUDA available, switching to GPU

In [38]:

reconstruction = Reconstruction(experimentalData, params)
reconstruction.No = 2**11
reconstruction.initialProbe = "circ"
reconstruction.initialObject = "ones"
# initialize probe and object and related params
reconstruction.initializeObjectProbe()

# customize initial probe quadratic phase
reconstruction.probe = reconstruction.probe * np.exp(
    1.0j
    * 2
    * np.pi
    / reconstruction.wavelength
    * (reconstruction.Xp**2 + reconstruction.Yp**2)
    / (3 * 6e-3)
)

reconstruction = Reconstruction(experimentalData, params) reconstruction.No = 2**11 reconstruction.initialProbe = "circ" reconstruction.initialObject = "ones" # initialize probe and object and related params reconstruction.initializeObjectProbe() # customize initial probe quadratic phase reconstruction.probe = reconstruction.probe * np.exp( 1.0j * 2 * np.pi / reconstruction.wavelength * (reconstruction.Xp**2 + reconstruction.Yp**2) / (3 * 6e-3) )

INFO:Reconstruction:Copying attribute wavelength
INFO:Reconstruction:Copying attribute dxd
INFO:Reconstruction:Copying attribute theta
INFO:Reconstruction:Copying attribute spectralDensity
INFO:Reconstruction:Copying attribute entrancePupilDiameter
INFO:Reconstruction:Initial object set to ones
INFO:Reconstruction:Initial probe set to circ

We first use the standard intensityConstraint, which means that the USAF sample is considered as a thin single-layer object

In [39]:

params.intensityConstraint = "standard"

mPIE = Engines.mPIE(reconstruction, experimentalData, params, monitor)
mPIE.numIterations = 100
mPIE.betaProbe = 0.05
mPIE.betaObject = 0.25
mPIE.reconstruct()

params.intensityConstraint = "standard" mPIE = Engines.mPIE(reconstruction, experimentalData, params, monitor) mPIE.numIterations = 100 mPIE.betaProbe = 0.05 mPIE.betaObject = 0.25 mPIE.reconstruct()

INFO:mPIE:Doing probe com stabilization

mPIE:  97%|█████████▋| 97/100 [00:10<00:00,  9.09it/s]

INFO:mPIE:Doing probe com stabilization

INFO:mPIE:Doing probe com stabilization

mPIE:  99%|█████████▉| 99/100 [00:10<00:00, 10.18it/s]

INFO:mPIE:Doing probe com stabilization

mPIE: 100%|██████████| 100/100 [00:10<00:00,  9.55it/s]

INFO:mPIE:switch to cpu

Now we switch to interferometric constraint, which takes into account the extra reflection from the plane side of the USAF glass substrate, and treats it as an external 'reference'. The results show that the noise in object reconstruction is significantly cleaned up.

In [40]:

params.intensityConstraint = "interferometric"

mPIE = Engines.mPIE(reconstruction, experimentalData, params, monitor)
mPIE.numIterations = 200
mPIE.betaProbe = 0.25
mPIE.betaObject = 0.25
mPIE.reconstruct()

params.intensityConstraint = "interferometric" mPIE = Engines.mPIE(reconstruction, experimentalData, params, monitor) mPIE.numIterations = 200 mPIE.betaProbe = 0.25 mPIE.betaObject = 0.25 mPIE.reconstruct()

INFO:mPIE:Doing probe com stabilization

mPIE:  98%|█████████▊| 197/200 [00:20<00:00,  9.01it/s]

INFO:mPIE:Doing probe com stabilization

INFO:mPIE:Doing probe com stabilization

mPIE: 100%|█████████▉| 199/200 [00:20<00:00, 10.22it/s]

INFO:mPIE:Doing probe com stabilization

mPIE: 100%|██████████| 200/200 [00:20<00:00,  9.82it/s]

INFO:mPIE:switch to cpu

In [41]:

## now save the data
# reconstruction.saveResults('reconstruction.hdf5')

## now save the data # reconstruction.saveResults('reconstruction.hdf5')