Mathematical SoftwareUniversity of GrazMathematics and Scientific Computing, Uni Graz
Back

Gratopy

Kristian Bredies, Richard Huber, Benjamin Hackl

  • Topic: tomography
  • Topic: inverse problems
  • Topic: numerical analysis
  • Topic: GPU computing
  • Target audience: research
  • Maintenance: Actively maintained
Package

Gratopy — Graz accelerated tomographic projections for Python provides Radon transforms, fanbeam transforms, and their associated backprojections. Its pixel-driven projection operators are implemented with OpenCL for efficient GPU execution and integration with other PyOpenCL code.

The toolbox is intended for developing iterative tomographic reconstruction methods, including optimization-based approaches. It includes Landweber, conjugate-gradient, and total-variation reconstruction schemes, together with examples and test data.

An experimental Radon operator interface supports adjoints and compositions such as . Fanbeam execution remains available through the established projection interface. OpenCL drivers and a suitable PyOpenCL installation are required.

Publications

Documentation

Read the documentation

Funding

  • Austrian Science Fund (FWF) — P-29192 — Regularization Graphs for Variational Imaging
  • German Research Foundation (DFG) and Austrian Science Fund (FWF) — IGDK 1754 / W-1244 — Optimization and Numerical Analysis for Partial Differential Equations with Nonsmooth Structures
License
GPL-3.0-or-later
Maintenance
Actively maintained