xptycho

xptycho: ptychographic reconstruction with projected multi-agent consensus equilibrium (PMACE) using PyTorch.

Features include:

• Reconstruction of the complex object image, magnitude and phase, from far-field diffraction frames.
• Estimation of the probe together with the object, with one or more probe modes.
• Refinement of the probe positions.
• Preprocessing of raw detector frames: dark subtraction, outlier removal, centering, and cropping.
• One HDF5 file format for scans, samples, and reconstructions.
• Demos on simulated data and on measured data, from raw file to image.
• Seamless operation on 1 or more GPUs, Mac MPS, or CPU.

Full documentation at https://xptycho.readthedocs.io/

Design pages at https://cabouman.github.io/xptycho/

Install:

pip install xptycho

Reconstruct in a few lines:

import xptycho as xpt
scan = xpt.Scan.load('scan.h5')
model = xpt.PtychoModel.from_scan(scan)
recon = model.recon(scan)
xpt.view_sample(recon)

xptycho implements the PMACE method of Qiuchen Zhai, Gregery T. Buzzard, Kevin Mertes, Brendt Wohlberg, and Charles A. Bouman. For the papers to cite, the source of the demo data, and the funding support, see Credits.