regpy 1.1.0

A python library to solve Inverse and Ill-Posed problems by regularization

RegPy: Python tools for regularization methods

RegPy is a Python library developed at the Institute for Numerical and Applied Mathematics at the University of Göttingen. It provides tools to implement custom forward models -- both linear and non-linear -- as well as a variety of regularization methods and stopping rules.

This project is currently approaching beta quality state, but remains under active development. As a result, you may run into bugs or partially undocumented features. If you run into any issues we welcome any information on our GitHub issue tracker.

Detailed information and documentation of the current version can be found at https://num.math.uni-goettingen.de/regpy/.

Usage examples

We offer an explanation on how to use RegPy here and our website features several detailed usage examples. These examples are provided as Jupyter notebooks that serve as a tutorial-style introduction to RegPy.

For a more comprehensive overview of RegPy's capabilities, we provide numerous examples in the examples GitHub repository. These examples are also part of the docker image provided on DockerHub (see in the installation instructions for details). Most examples include both a commented Python script and a Jupyter notebook with more detailed explanations.

Installation

We provide different installation methods, such installation using pip, listed and explained in INSTALLATION.md.

Dependencies

• numpy >= 1.14
• scipy >= 1.1

Optional dependencies

• ngsolve, for some forward operators that require solving PDEs. We provide an optional installation tag ngsolve when installing with pip.
• matplotlib (for some of the examples)
• sphinx (for generating the documentation) further requirements in doc/sphinx/requirements.txt