Skip to main content

high-level tools for linearizing and solving systems of equations

Project description


Run Tests Code Coverage

linsolve is a module providing high-level tools for linearizing and solving systems of equations.


The solvers in linsolve include LinearSolver, LogProductSolver, and LinProductSolver. LinearSolver solves linear equations of the form 'a*x + b*y + c*z'. LogProductSolver uses logrithms to linearize equations of the form 'x*y*z'. LinProductSolver uses symbolic Taylor expansion to linearize equations of the form 'x*y + y*z'.

See linsolve_example.ipynb for a tutorial on how to use these functionalities.

Below we give a brief example on the general usage of LinearSolver.

Assume we have a linear system of equations, with a data vector y containing measurements and a model vector b containing parameters we would like to solve for. Let's simplify to the problem of fitting a line to three data points, which amounts to solving for a slope and an offset. In this case, our linear system of equations can be written as

where b_1 is the slope and b_2 is the offset, and the A matrix contains the mapping from model vector b to data vector y. In our case, the a_x1 values are the x-values of the data points, and the a_x2 values are equal to unity. Let's assume the data vector measurements are y_1 = 2, y_2 = 4 and y_3 = 7, and their corresponding dependent variable values are a_11 = 0, a_21 = 2 and a_31 = 4.

We will use LinearSolver to solve this system of equations in the following manner. First we setup a data dictionary, which contains as keys strings of the RHS of our linear model equation, and as values the y-data measurements:

data = {'b_2': 2.0, '2.0*b_1 + b_2': 4.0, '4*b_1 + b_2': 7.0}

Alternatively, we can write the data dictionary more generally by also writing dictionary of constants we don't want to solve for (i.e. the values of the A matrix):

data = {'a_11*b_1 + a_12*b_2': 2.0, 'a_21*b_1 + a_22*b_2': 4.0, 'a_31*b_1 + a_32*b_2': 7.0}
consts = {'a_11': 0.0, 'a_21': 2.0, 'a_31': 4.0, 'a_12': 1.0, 'a_22': 1.0, 'a_32': 1.0}

We then feed this into linsolve.LinearSolver (optionally passing the consts dictionary as keyword arguments.)

ls = linsolve.LinearSolver(data) # or linsolve.LinearSolver(data, **consts) if we use constants
solution = ls.solve()

The output, solution, is a dictionary with solution of our model vector:

{'b_1': 1.2499999999999998, 'b_2': 1.8333333333333324}

Weighting of measurements can be implemented through an optional wgts dictionary that parallels the construction of data. To see a more in-depth example, please consult the linsolve_example.ipynb tutorial.

Package Details

Known Issues and Planned Improvements

For details see the issue log.

Community Guidelines

Contributions to this package to add new file formats or address any of the issues in the issue log are very welcome. Please submit improvements as pull requests against the repo after verifying that the existing tests pass and any new code is well covered by unit tests.

Bug reports or feature requests are also very welcome, please add them to the issue log after verifying that the issue does not already exist. Comments on existing issues are also welcome.


Preferred method of installation is pip install . (or pip install git+ This will install all dependencies. See below for manual management of dependencies.


If you use conda and would like to ensure that dependencies are installed with conda rather than pip, you should execute::

$ conda install "numpy>=1.14" scipy


If you are developing linsolve, it is recommended to create a fresh environment by::

$ git clone
$ cd linsolve
$ conda create -n linsolve python=3
$ conda activate linsolve
$ conda env update -n linsolve -f environment.yml
$ pip install -e .

This will install extra dependencies required for testing/development as well as the standard ones.

To run tests, just run nosetests in the top-level directory.

Project details

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distribution

linsolve-1.1.3.tar.gz (106.6 kB view hashes)

Uploaded Source

Built Distribution

linsolve-1.1.3-py3-none-any.whl (17.6 kB view hashes)

Uploaded Python 3

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page