Welcome to pykinsol’s documentation!

Indices and tables


Build status Build status on CircleCI Build status on Travis-CI PyPI version License coverage

pykinsol provides a Python binding to the Nonlinear system of equation solver algorithms from kinsol in the SUNDIALS suite. pykinsol allows a user to numerically solve (systems of) non-linear equations. Note that not all functionality of KINSOL are currently exposed in the python wrapper. Additional functionality is however easy to add if needed.

You may also want to know that you can use pykinsol from pyneqsys which can e.g. derive the Jacobian analytically for you.


Autogenerated API documentation is found here.


Simplest way to install pykinsol is to use the conda package manager:

$ conda install -c bjodah pykinsol pytest
$ python -m pytest --pyargs pykinsol

tests should pass.

Binary distribution is available here: https://anaconda.org/bjodah/pykinsol

Source distribution is available here: https://pypi.python.org/pypi/pykinsol

To compile from source you may proceed e.g. as follows:

$ LLAPACK=openblas python3.4 setup.py install --user

where we specified specifically what LAPACK implementation we used for building SUNDIALS.


See examples/, and rendered jupyter notebooks here: http://hera.physchem.kth.se/~pykinsol/branches/master/examples

Here is a simple example:

>>> def f(x, fout):
...     fout[0] = x[0] + (x[0] - x[1])**3/2 - 1
...     fout[1] = x[1] + (x[1] - x[0])**3/2
>>> def j(x, Jout, fx):
...     Jout[0, 0] =  1 + 3/2 * (x[0] - x[1])**2
...     Jout[0, 1] =  -3/2 * (x[0] - x[1])**2
...     Jout[1, 0] =  -3/2 * (x[1] - x[0])**2
...     Jout[1, 1] =  1 + 3/2 * (x[1] - x[0])**2
>>> from pykinsol import solve
>>> result = solve(f, j, [0, 0])
>>> assert result['success']
>>> print(result['x'])  
[ 0.84116389  0.15883611]


The source code is Open Source and is released under the simplified 2-clause BSD license. See LICENSE for further details.

Contributors are welcome to suggest improvements at https://github.com/bjodah/pykinsol


Björn I. Dahlgren, contact:

  • gmail address: bjodah