rsdfoq 1.0

A derivative-free solver for large-scale minimization

A Python package for general minimization, where derivatives are not available, using random subspaces. For a description of this algorithm, see this paper.

For lower-dimensional problems, consider using the more actively maintained Py-BOBYQA.

Citation

If you use RSDFO-Q in an academic work, please cite the following paper:

C. Cartis and L. Roberts, Randomized Subspace Derivative-Free Optimization with Quadratic Models and Second-Order Convergence. Optimization Methods and Software, to appear.

A preprint version of this paper can be found on arXiv.

Installation

You can install RSDFO-Q by cloning this repository and installing with pip:

$ git clone https://github.com/lindonroberts/rsdfoq.git
$ cd rsdfoq
$ ls                     <-- check for pyproject.toml
$ pip install -e .

RSDFO-Q requires NumPy, SciPy and pandas, but these will be installed automatically if they are not already available.

Usage

Examples for how to use RSDFO-Q may be found in the examples directory.