ssnp 0.0.2rc1

Split-step non-paraxial beam propagation simulation package

SSNP

Split-step non-paraxial beam propagation method

Features

• Forward model calculation based on CUDA
• Read/write data of common file type
• Gradient calculation
• Image reconstruction with regularization

Reference

Sharma, A., & Agrawal, A. (2004). Split-step non-paraxial beam propagation method. Physics and Simulation of Optoelectronic Devices XII, 5349, 132. https://doi.org/10.1117/12.528172

Lim, J., Ayoub, A. B., Antoine, E. E., & Psaltis, D. (2019). High-fidelity optical diffraction tomography of multiple scattering samples. Light: Science & Applications, 8(1), 82. https://doi.org/10.1038/s41377-019-0195-1

Changelog

0.0.2 (developing)

• Add skcuda as a substitution of reikna
• Enable to compute several illumination on the same object in a batch
• Support background refractive index (n0 > 1) in model and gradient computation
• Record partial intermediate results to effectively save memory when tracking operations
• Add stream parameter for async operation and related small tools
• Add cache to avoid unnecessary recompiling
• Fix errors:
  • data.write: fix unexpectedly changing input data when scaling before write

0.0.1 (Sep 7, 2020)

• Add gradient calculation support (tracking operations and doing autograd)
• Add config to set package-wise constant
• Add Multiplier class to generate auto-cached numpy/gpu array
• Add MATLAB .mat file read & write support (rely on scipy lib)