====================================================== DFO-GN: Derivative-Free Nonlinear Least-Squares Solver

.. image:: https://travis-ci.org/numericalalgorithmsgroup/dfogn.svg?branch=master :target: https://travis-ci.org/numericalalgorithmsgroup/dfogn :alt: Build Status

.. image:: https://img.shields.io/badge/License-GPL%20v3-blue.svg :target: https://www.gnu.org/licenses/gpl-3.0 :alt: GNU GPL v3 License

.. image:: https://img.shields.io/pypi/v/DFOGN.svg :target: https://pypi.python.org/pypi/DFOGN :alt: Latest PyPI version

.. image:: https://zenodo.org/badge/DOI/10.5281/zenodo.2629875.svg :target: https://doi.org/10.5281/zenodo.2629875 :alt: DOI:10.5281/zenodo.2629875

DFO-GN is a package for solving nonlinear least-squares minimisation, without requiring derivatives of the objective.

This is an implementation of the algorithm from our paper: A Derivative-Free Gauss-Newton Method <https://doi.org/10.1007/s12532-019-00161-7>, C. Cartis and L. Roberts, Mathematical Programming Computation (2019). For reproducibility of all figures in this paper, please feel free to contact the authors. A preprint of the paper is available here <https://arxiv.org/abs/1710.11005>.

Note: we have released a newer package, called DFO-LS, which is an upgrade of DFO-GN to improve its flexibility and robustness to noisy problems. See here <https://github.com/numericalalgorithmsgroup/dfols>_ for details.

Citation To cite DFO-GN, please use :: @Article{DFOGN, Title = {A derivative-free {G}auss-{N}ewton method}, Author = {Cartis, Coralia and Roberts, Lindon}, Journal = {Mathematical Programming Computation}, Year = {2019}, Doi = {10.1007/s12532-019-00161-7}, Url = {https://doi.org/10.1007/s12532-019-00161-7} }

Documentation

See manual.pdf or here <https://numericalalgorithmsgroup.github.io/dfogn/>_.

Requirements

DFO-GN requires the following software to be installed:

• Python 2.7 or Python 3 <http://www.python.org/>_

Additionally, the following python packages should be installed (these will be installed automatically if using pip <http://www.pip-installer.org/>, see Installation using pip):

• NumPy 1.11 or higher <http://www.numpy.org/>_
• SciPy 0.18 or higher <http://www.scipy.org/>_

Installation using pip

For easy installation, use pip <http://www.pip-installer.org/>_ as root:

.. code-block:: bash

$ [sudo] pip install --pre dfogn

If you do not have root privileges or you want to install DFO-GN for your private use, you can use:

.. code-block:: bash

$ pip install --pre --user dfogn

which will install DFO-GN in your home directory.

Note that if an older install of DFO-GN is present on your system you can use:

.. code-block:: bash

$ [sudo] pip install --pre --upgrade dfogn

to upgrade DFO-GN to the latest version.

Manual installation

The source code for DFO-GN is available on Github <https://https://github.com/numericalalgorithmsgroup/dfogn>_:

.. code-block:: bash

$ git clone https://github.com/numericalalgorithmsgroup/dfogn
$ cd dfogn

DFO-GN is written in pure Python and requires no compilation. It can be installed using:

.. code-block:: bash

$ [sudo] pip install --pre .

If you do not have root privileges or you want to install DFO-GN for your private use, you can use:

.. code-block:: bash

$ pip install --pre --user .

instead.

Testing

If you installed DFO-GN manually, you can test your installation by running:

.. code-block:: bash

$ python setup.py test

Alternatively, the documentation <https://numericalalgorithmsgroup.github.io/dfogn/>_ provides some simple examples of how to run DFO-GN, which are also available in the examples directory.