Data Format¶

PtyLab.py loads experimental data from HDF5 files. All fields are stored at the root level of the file.

CPM required fields¶

Field	Type	Unit	Description
`ptychogram`	`float32 (N, Nd, Nd)`	counts	Stack of N diffraction intensity patterns
`wavelength`	`float64`	m	Illumination wavelength
`encoder`	`float64 (N, 2)`	m	Scan positions (x, y) in the sample plane
`dxd`	`float64`	m	Detector pixel size
`zo`	`float64`	m	Sample-to-detector distance

CPM optional fields¶

Field	Type	Unit	Description
`entrancePupilDiameter`	`float64`	m	Diameter of the probe aperture
`spectralDensity`	`float64 (nlambda,)`	—	Relative spectral weights for polychromatic reconstruction
`theta`	`float64`	rad	Tilt angle for reflection ptychography
`emptyBeam`	`float32 (Nd, Nd)`	counts	Reference probe image (for modulus-enforced probe)

Loading data¶

from PtyLab import ExperimentalData

data = ExperimentalData("path/to/data.hdf5", operationMode="CPM")
data.loadData("path/to/data.hdf5")

Or pass the filename directly to easyInitialize:

import PtyLab
experimentalData, reconstruction, params, monitor, engine = PtyLab.easyInitialize(
    "path/to/data.hdf5",
    engine=...,
    operationMode="CPM",
)

Built-in example datasets¶

PtyLab.py ships with simulation datasets for testing and tutorials. Use these special filename strings:

Name	Description
`"example:simulation_cpm"`	Synthetic CPM ptychogram with known object and probe
`"example:simulation_fpm"`	Synthetic FPM ptychogram
`"test:nodata"`	Empty object for unit testing (no diffraction data)

data = ExperimentalData(operationMode="CPM")
data.loadData("example:simulation_cpm")

Creating an HDF5 file¶

Use h5py to create a compatible data file from your own measurements:

import h5py
import numpy as np

ptychogram = ...   # shape (N, Nd, Nd), dtype float32
encoder = ...      # shape (N, 2), dtype float64, units: meters

with h5py.File("mydata.hdf5", "w") as f:
    f.create_dataset("ptychogram", data=ptychogram.astype(np.float32))
    f.create_dataset("wavelength", data=np.float64(532e-9))   # 532 nm
    f.create_dataset("encoder", data=encoder.astype(np.float64))
    f.create_dataset("dxd", data=np.float64(6.5e-6))           # 6.5 µm pixels
    f.create_dataset("zo", data=np.float64(50e-3))             # 50 mm distance

Note

All distances must be in meters. The ptychogram should contain raw intensity values (counts), not amplitude. Negative values and NaNs are not supported.