DFBGN is a Python package for nonlinear least-squares minimization, where derivatives are not available. It is particularly useful when evaluations of the objective are expensive and/or noisy, and the number of variables to be optimized is large.
This is an implementation of the algorithm from our paper: C. Cartis, L. Roberts, Scalable Subspace Methods for Derivative-Free Nonlinear Least-Squares Optimization, (2021) [preprint].
DFBGN is based on DFO-LS, but is better-suited when there are many variables to be optimized (so the linear algebra in DFO-LS is too slow). Unlike DFO-LS, DFBGN does not currently support bound constraints on the variables.
If you are interested in solving general optimization problems (without a least-squares structure), you may wish to try Py-BOBYQA.
If you use DFBGN in a paper, please cite:
Cartis, C., and Roberts, L., Scalable Subspace Methods for Derivative-Free Nonlinear Least-Squares Optimization, arXiv preprint arXiv:2102.12016, (2021).
DFBGN 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, see Installation using pip):
- NumPy 1.11 or higher (http://www.numpy.org/)
- SciPy 0.18 or higher (http://www.scipy.org/)
- Pandas 0.17 or higher (http://pandas.pydata.org/)
For easy installation, use pip as root:
$ [sudo] pip install dfbgn
or alternatively easy_install:
$ [sudo] easy_install dfbgn
If you do not have root privileges or you want to install DFBGN for your private use, you can use:
$ pip install --user dfbgn
which will install DFBGN in your home directory.
Note that if an older install of DFBGN is present on your system you can use:
$ [sudo] pip install --upgrade dfbgn
to upgrade DFBGN to the latest version.
Alternatively, you can download the source code from Github and unpack as follows:
$ git clone https://github.com/numericalalgorithmsgroup/dfbgn $ cd dfbgn
DFBGN is written in pure Python and requires no compilation. It can be installed using:
$ [sudo] pip install .
If you do not have root privileges or you want to install DFBGN for your private use, you can use:
$ pip install --user .
instead.
To upgrade DFBGN to the latest version, navigate to the top-level directory (i.e. the one containing setup.py) and rerun the installation using pip, as above:
$ git pull $ [sudo] pip install . # with admin privileges
If you installed DFBGN manually, you can test your installation by running:
$ python setup.py test
Alternatively, the HTML documentation provides some simple examples of how to run DFBGN.
Examples of how to run DFBGN are given in the documentation, and the examples directory in Github.
If DFBGN was installed using pip you can uninstall as follows:
$ [sudo] pip uninstall dfbgn
If DFBGN was installed manually you have to remove the installed files by hand (located in your python site-packages directory).
Please report any bugs using GitHub's issue tracker.
This algorithm is released under the GNU GPL license. Please contact NAG for alternative licensing.