Applications of algorithmic differentiation to phase retrieval algorithms.

@article{Jurling2014ApplicationsOA,
  title={Applications of algorithmic differentiation to phase retrieval algorithms.},
  author={Alden S. Jurling and James R. Fienup},
  journal={Journal of the Optical Society of America. A, Optics, image science, and vision},
  year={2014},
  volume={31 7},
  pages={
          1348-59
        }
}
  • Alden S. Jurling, J. Fienup
  • Published 1 July 2014
  • Computer Science
  • Journal of the Optical Society of America. A, Optics, image science, and vision
In this paper, we generalize the techniques of reverse-mode algorithmic differentiation to include elementary operations on multidimensional arrays of complex numbers. We explore the application of the algorithmic differentiation to phase retrieval error metrics and show that reverse-mode algorithmic differentiation provides a framework for straightforward calculation of gradients of complicated error metrics without resorting to finite differences or laborious symbolic differentiation. 
View on PubMed
optics.rochester.edu
75 Citations
Highly Influential Citations
1
Background Citations
6
Methods Citations
20

75 Citations

Citation Type

Citation Type

Phase retrieval with unknown sampling factors via the two-dimensional chirp z-transform.
  • Alden S. Jurling, J. Fienup
  • Computer Science
    Journal of the Optical Society of America. A, Optics, image science, and vision
  • 2014
TLDR
The analytic gradient of a phase retrieval error metric is derived with respect to the sampling factor or the f-number that produced the measured point-spread function, thereby improving the accuracy of the phase estimate.
Efficient ptychographic phase retrieval via a matrix-free Levenberg-Marquardt algorithm.
TLDR
This algorithm can be used to solve both the unconstrained ptychographic object retrieval problem and the constrained "blind" ptyChographic object and probe retrieval problems, under the popular Gaussian noise model as well as the Poisson noise model.
An Optimization Approach To EO Phase Retrieval
TLDR
An optimization approach to optical phase retrieval by defining a non-linear error function and compute its gradient, which enables a quasi-Newton optimizatio n to find a stationary point, which corresponds to the final complex image estimate.
Phase retrieval and design with automatic differentiation: tutorial
The principal limitation in many areas of astronomy, especially for directly imaging exoplanets, arises from instability in the point spread function (PSF) delivered by the telescope and instrument.…
Utilizing Machine Learning for Smart Starting Guesses for Phase Retrieval
TLDR
This work uses machine learning to produce better starting guesses for wavefront-sensing phase retrieval problems with large amounts of wavefront error that often do not converge without a good starting point.
Extending capture range for piston retrieval in segmented systems.
TLDR
It is shown that this method for finding errors using broadband light and a specialized grid search as part of a more global search has a high rate of success for a case where nonlinear optimization gets stuck in local minima.
Overcoming Large Piston Capture Range Problems in Segmented Systems Using Broadband Light
TLDR
A new algorithm is presented that uses broadband light to fit large piston error with a high convergence rate to retrieve piston errors greater than one wave in segmented systems.
Using energy consistency to improve phase retrievals with discrete Fourier transform cropping
TLDR
It is found that the cropping methods had better wavefront fits compared to the downsampled method, and the Parseval error metric had better retrieval success rates over the other two methods, although with greater computational requirements.
Efficient Prescription Retrieval from PSF Data
Reverse-accumulation automated differentiation is applied to retrieve optical prescription parameters from point spread function data, yielding a 4.6× speed improvement over existing methods.
Nonlinear Optimization Algorithm for Partially Coherent Phase Retrieval and Source Recovery
TLDR
The gradient and Hessian of the cost function are derived and it is shown that the second-order optimization approach outperforms previously proposed phase retrieval algorithms, for datasets taken with both coherent and partially coherent illumination.
...
1
2
3
4
5
...

References

SHOWING 1-10 OF 20 REFERENCES
Algorithmic differentiation in Python with AlgoPy
Evaluating derivatives - principles and techniques of algorithmic differentiation, Second Edition
TLDR
This second edition has been updated and expanded to cover recent developments in applications and theory, including an elegant NP completeness argument by Uwe Naumann and a brief introduction to scarcity, a generalization of sparsity.
Algorithm 755: ADOL-C: a package for the automatic differentiation of algorithms written in C/C++
The C++ package ADOL-C described here facilitates the evaluation of first and higher derivatives of vector functions that are defined by computer programs written in C or C++. The resulting…
Phase retrieval algorithms: a comparison.
TLDR
Iterative algorithms for phase retrieval from intensity data are compared to gradient search methods and it is shown that both the error-reduction algorithm for the problem of a single intensity measurement and the Gerchberg-Saxton algorithm forThe problem of two intensity measurements converge.
Phase-retrieval algorithms for a complicated optical system.
TLDR
Improved algorithms permit a more accurate characterization of complicated optical systems from their point spread functions, independent of the number of coefficients or phase points.
The Tapenade automatic differentiation tool: Principles, model, and specification
TLDR
The principles of Tapenade are described, a subset of the general principles of AD, and the extensions of the tool that are planned in a foreseeable future are presented, deriving from the ongoing research on AD.
Graph-Based Software Design for Managing Complexity and Enabling Concurrency in Multiphysics PDE Software
TLDR
This work presents a design strategy that shifts focus away from high-level algorithmic concerns to low-level data dependencies, and makes problems with complex dependencies entirely tractable, and removes virtually all logic from the algorithm itself.
Phase retrieval for undersampled broadband images
TLDR
Generalizations of phase-retrieval algorithms, which accommodate broadband light, allow for undersampled images, permit fitting of multiple images simultaneously, and have a flexible description of the aberrations, are described in this study.
Quasi-Newton Methods
The problem to be solved is formulated precisely and the introduction of quasi-Newton methods is motivated by considering the classical Newton and secant methods and their properties. Three highly…
Joint estimation of object and aberrations by using phase diversity
TLDR
The joint estimation of an object and the aberrations of an incoherent imaging system from multiple images incorporating phase diversity is investigated and an expectation-maximization algorithm that enforces a nonnegativity constraint in a natural fashion is constructed for use in the Poisson noise case.
...
1
2
...