Skip to main content

Linear optimization with N-D labeled arrays in Python

Project description

linopy: Optimization with array-like variables and constraints

PyPI License Tests doc codecov

        Linear
        Integer
        Non-linear
        Optimization in
        PYthon

linopy is an open-source python package that facilitates optimization with real world data. It builds a bridge between data analysis packages like xarray & pandas and problem solvers like cbc, gurobi (see the full list below). Linopy supports Linear, Integer, Mixed-Integer and Quadratic Programming while aiming to make linear programming in Python easy, highly-flexible and performant.

Benchmarks

linopy is designed to be fast and efficient. The following benchmark compares the performance of linopy with the alternative popular optimization packages.

Performance Benchmark

Main features

linopy is heavily based on xarray which allows for many flexible data-handling features:

  • Define (arrays of) continuous or binary variables with coordinates, e.g. time, consumers, etc.
  • Apply arithmetic operations on the variables like adding, substracting, multiplying with all the broadcasting potentials of xarray
  • Apply arithmetic operations on the linear expressions (combination of variables)
  • Group terms of a linear expression by coordinates
  • Get insight into the clear and transparent data model
  • Modify and delete assigned variables and constraints on the fly
  • Use lazy operations for large linear programs with dask
  • Choose from different commercial and non-commercial solvers
  • Fast import and export a linear model using xarray's netcdf IO

Installation

So far linopy is available on the PyPI repository

pip install linopy

or on conda-forge

conda install -c conda-forge linopy

In a Nutshell

Linopy aims to make optimization programs transparent and flexible. To illustrate its usage, let's consider a scenario where we aim to minimize the cost of buying apples and bananas over a week, subject to daily and weekly vitamin intake constraints.

>>> import pandas as pd
>>> import linopy

>>> m = linopy.Model()

>>> days = pd.Index(['Mon', 'Tue', 'Wed', 'Thu', 'Fri'], name='day')
>>> apples = m.add_variables(lower=0, name='apples', coords=[days])
>>> bananas = m.add_variables(lower=0, name='bananas', coords=[days])
>>> apples
Variable (day: 5)
-----------------
[Mon]: apples[Mon] ∈ [0, inf]
[Tue]: apples[Tue] ∈ [0, inf]
[Wed]: apples[Wed] ∈ [0, inf]
[Thu]: apples[Thu] ∈ [0, inf]
[Fri]: apples[Fri] ∈ [0, inf]

Add daily vitamin constraints

>>> m.add_constraints(3 * apples + 2 * bananas >= 8, name='daily_vitamins')
Constraint `daily_vitamins` (day: 5):
-------------------------------------
[Mon]: +3 apples[Mon] + 2 bananas[Mon] ≥ 8
[Tue]: +3 apples[Tue] + 2 bananas[Tue] ≥ 8
[Wed]: +3 apples[Wed] + 2 bananas[Wed] ≥ 8
[Thu]: +3 apples[Thu] + 2 bananas[Thu] ≥ 8
[Fri]: +3 apples[Fri] + 2 bananas[Fri] ≥ 8

Add weekly vitamin constraint

>>> m.add_constraints((3 * apples + 2 * bananas).sum() >= 50, name='weekly_vitamins')
Constraint `weekly_vitamins`
----------------------------
+3 apples[Mon] + 2 bananas[Mon] + 3 apples[Tue] ... +2 bananas[Thu] + 3 apples[Fri] + 2 bananas[Fri] ≥ 50

Define the prices of apples and bananas and the objective function

>>> apple_price = [1, 1.5, 1, 2, 1]
>>> banana_price = [1, 1, 0.5, 1, 0.5]
>>> m.objective = apple_price * apples + banana_price * bananas

Finally, we can solve the problem and get the optimal solution:

>>> m.solve()
>>> m.objective.value
17.166

... and display the solution as a pandas DataFrame

>>> m.solution.to_pandas()
        apples  bananas
day
Mon    2.667      0
Tue    0          4
Wed    0          9
Thu    0          4
Fri    0          4

Supported solvers

linopy supports the following solvers

Note that these do have to be installed by the user separately.

Development Setup

To set up a local development environment for linopy and to run the same tests that are run in the CI, you can run:

python -m venv venv
source venv/bin/activate
pip install uv
uv pip install -e .[dev,solvers]
pytest

The -e flag of the install command installs the linopy package in editable mode, which means that the virtualenv (and thus the tests) will run the code from your local checkout.

Citing Linopy

If you use Linopy in your research, please cite the following paper:

A BibTeX entry for LaTeX users is

@article{Hofmann2023,
    doi = {10.21105/joss.04823},
    url = {https://doi.org/10.21105/joss.04823},
    year = {2023}, publisher = {The Open Journal},
    volume = {8},
    number = {84},
    pages = {4823},
    author = {Fabian Hofmann},
    title = {Linopy: Linear optimization with n-dimensional labeled variables},
    journal = {Journal of Open Source Software}
}

License

Copyright 2021 Fabian Hofmann

This package is published under MIT license. See LICENSE.txt for details.

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

linopy-0.5.5.tar.gz (1.2 MB view details)

Uploaded Source

Built Distribution

linopy-0.5.5-py3-none-any.whl (95.8 kB view details)

Uploaded Python 3

File details

Details for the file linopy-0.5.5.tar.gz.

File metadata

  • Download URL: linopy-0.5.5.tar.gz
  • Upload date:
  • Size: 1.2 MB
  • Tags: Source
  • Uploaded using Trusted Publishing? Yes
  • Uploaded via: twine/6.1.0 CPython/3.12.9

File hashes

Hashes for linopy-0.5.5.tar.gz
Algorithm Hash digest
SHA256 2ce6897bd6618d863efb63f2a2710385d40a432b60b3e32bff0e7c699dbf3804
MD5 3c042c3a42f3dc4de7aa66722c9ec0d6
BLAKE2b-256 a387988a991b27596404ee6bd2a7d34e1eef3a4a2e9e719bede6c0cb472ca013

See more details on using hashes here.

Provenance

The following attestation bundles were made for linopy-0.5.5.tar.gz:

Publisher: release.yml on PyPSA/linopy

Attestations: Values shown here reflect the state when the release was signed and may no longer be current.

File details

Details for the file linopy-0.5.5-py3-none-any.whl.

File metadata

  • Download URL: linopy-0.5.5-py3-none-any.whl
  • Upload date:
  • Size: 95.8 kB
  • Tags: Python 3
  • Uploaded using Trusted Publishing? Yes
  • Uploaded via: twine/6.1.0 CPython/3.12.9

File hashes

Hashes for linopy-0.5.5-py3-none-any.whl
Algorithm Hash digest
SHA256 811be075771b5735811540fd564d498387d556a28436dec6a5e152d005a234fe
MD5 572a40cb36fdbe9c18d53b371e0a4b32
BLAKE2b-256 28cc07f7b9d9f4af42d02c7ae0fa1997096b0d2c3e87ab4cb0a92d955e5d3cc0

See more details on using hashes here.

Provenance

The following attestation bundles were made for linopy-0.5.5-py3-none-any.whl:

Publisher: release.yml on PyPSA/linopy

Attestations: Values shown here reflect the state when the release was signed and may no longer be current.

Supported by

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