Before starting you will need Python 2.7.x or Python 3. You need to have numpy, scipy, and pip installed and we recommend installing Anaconda/Miniconda for your desired Python version.
There are a couple of optional components of pySOT that needs to be installed manually:
scikit-learn: Necessary in order to use the Gaussian process regression. The minimum version is 0.18.1. Can be installed using
pip install "scikit-learn >= 0.18.1"
py-earth: Implementation of MARS. Can be installed using:
pip install six http://github.com/scikit-learn-contrib/py-earth/tarball/master
git clone git://github.com/scikit-learn-contrib/py-earth.git cd py-earth pip install six python setup.py install
mpi4py: This module is necessary in order to use pySOT with MPI. Can be installed through pip:
pip install mpi4py
or through conda (Anaconda/Miniconda) where it can be channeled with your favorite MPI implementation such as mpich:
conda install --channel mpi4py mpich mpi4py
subprocess32: A backport of the subprocess module for Python 3.2 that works for Python 2.7. This is the recommended way of launching workers through subprocesses for Python 2.7 and this module is easily installed using:
pip install subprocess32
matlab_wrapper: A module that can be used to create MATLAB sessions for older MATLAB versions where there is no default MATLAB engine. Easily instead using:
pip install matlab_wrapper
PySide: If you want to use the GUI you need to install PySide. This can be done with pip:
pip install PySide
There are currently two ways to install pySOT:
(Recommended) The easiest way to install pySOT is through pip in which case the following command should suffice:
pip install pySOT
The other option is cloning the repository and installing.
git clone https://github.com/dme65/pySOT
python setup.py install
Several examples problems are available at ./pySOT/test or in the pySOT.test module