iminuit

Jupyter-friendly Python interface for C++ MINUIT2

https://github.com/scikit-hep/iminuit

Keywords

Keywords from Contributors

Repository

Jupyter-friendly Python interface for C++ MINUIT2

README.rst

.. |iminuit| image:: doc/_static/iminuit_logo.svg
   :alt: iminuit

|iminuit|
=========

.. version-marker-do-not-remove

.. image:: https://scikit-hep.org/assets/images/Scikit--HEP-Project-blue.svg
   :target: https://scikit-hep.org
.. image:: https://img.shields.io/pypi/v/iminuit.svg
   :target: https://pypi.org/project/iminuit
.. image:: https://img.shields.io/conda/vn/conda-forge/iminuit.svg
   :target: https://github.com/conda-forge/iminuit-feedstock
.. image:: https://coveralls.io/repos/github/scikit-hep/iminuit/badge.svg?branch=develop
   :target: https://coveralls.io/github/scikit-hep/iminuit?branch=develop
.. image:: https://github.com/scikit-hep/iminuit/actions/workflows/docs.yml/badge.svg?branch=main
   :target: https://scikit-hep.org/iminuit
.. image:: https://zenodo.org/badge/DOI/10.5281/zenodo.3949207.svg
   :target: https://doi.org/10.5281/zenodo.3949207
.. image:: https://img.shields.io/badge/ascl-2108.024-blue.svg?colorB=262255
   :target: https://ascl.net/2108.024
   :alt: ascl:2108.024
.. image:: https://img.shields.io/gitter/room/Scikit-HEP/iminuit
   :target: https://gitter.im/Scikit-HEP/iminuit
.. image:: https://mybinder.org/badge_logo.svg
   :target: https://mybinder.org/v2/gh/scikit-hep/iminuit/develop?filepath=doc%2Ftutorial

``iminuit`` is a Jupyter-friendly Python interface for the ``Minuit2`` C++ library maintained by CERN's `ROOT team `_.

Minuit was designed to optimize statistical cost functions, for maximum-likelihood and least-squares fits. It provides the best-fit parameters and error estimates from likelihood profile analysis.

The iminuit package brings additional features:

- Builtin cost functions for statistical fits to N-dimensional data

  - Unbinned and binned maximum-likelihood + extended versions
  - `Template fits with error propagation `_
  - Least-squares (optionally robust to outliers)
  - Gaussian penalty terms for parameters
  - Cost functions can be combined by adding them: ``total_cost = cost_1 + cost_2``
  - Visualization of the fit in Jupyter notebooks
- Support for SciPy minimizers as alternatives to Minuit's MIGRAD algorithm (optional)
- Support for Numba accelerated functions (optional)

Minimal dependencies
--------------------

``iminuit`` is promised to remain a lean package which only depends on ``numpy``, but additional features are enabled if the following optional packages are installed.

- ``numba``: Cost functions are partially JIT-compiled if ``numba`` is installed.
- ``matplotlib``: Visualization of fitted model for builtin cost functions
- ``ipywidgets``: Interactive fitting, see example below (also requires ``matplotlib``)
- ``scipy``: Compute Minos intervals for arbitrary confidence levels
- ``unicodeitplus``: Render names of model parameters in simple LaTeX as Unicode

Documentation
-------------

Checkout our large and comprehensive list of `tutorials`_ that take you all the way from beginner to power user. For help and how-to questions, please use the `discussions`_ on GitHub or `gitter`_.

**Lecture by Glen Cowan**

`In the exercises to his lecture for the KMISchool 2022 `_, Glen Cowan shows how to solve statistical problems in Python with iminuit. You can find the lectures and exercises on the Github page, which covers both frequentist and Bayesian methods.

`Glen Cowan `_ is a known for his papers and international lectures on statistics in particle physics, as a member of the Particle Data Group, and as author of the popular book `Statistical Data Analysis `_.

In a nutshell
-------------

``iminuit`` can be used with a user-provided cost functions in form of a negative log-likelihood function or least-squares function. Standard functions are included in ``iminuit.cost``, so you don't have to write them yourself. The following example shows how to perform an unbinned maximum likelihood fit.

.. code:: python

    import numpy as np
    from iminuit import Minuit
    from iminuit.cost import UnbinnedNLL
    from scipy.stats import norm

    x = norm.rvs(size=1000, random_state=1)

    def pdf(x, mu, sigma):
        return norm.pdf(x, mu, sigma)

    # Negative unbinned log-likelihood, you can write your own
    cost = UnbinnedNLL(x, pdf)

    m = Minuit(cost, mu=0, sigma=1)
    m.limits["sigma"] = (0, np.inf)
    m.migrad()  # find minimum
    m.hesse()   # compute uncertainties

.. image:: doc/_static/demo_output.png
    :alt: Output of the demo in a Jupyter notebook

Interactive fitting
-------------------

``iminuit`` optionally supports an interactive fitting mode in Jupyter notebooks.

.. image:: doc/_static/interactive_demo.gif
   :alt: Animated demo of an interactive fit in a Jupyter notebook

High performance when combined with numba
-----------------------------------------

When ``iminuit`` is used with cost functions that are JIT-compiled with `numba`_ (JIT-compiled pdfs are provided by `numba_stats`_ ), the speed is comparable to `RooFit`_ with the fastest backend. `numba`_ with auto-parallelization is considerably faster than the parallel computation in `RooFit`_.

.. image:: doc/_static/roofit_vs_iminuit+numba.svg

More information about this benchmark is given `in the Benchmark section of the documentation `_.

Citation
--------

If you use iminuit in a scientific work, please cite us. A generic BibTeX entry is::

    @article{iminuit,
      author={Hans Dembinski and Piti Ongmongkolkul et al.},
      title={scikit-hep/iminuit},
      DOI={10.5281/zenodo.3949207},
      publisher={Zenodo},
      year={2020},
      month={Dec},
      url={https://doi.org/10.5281/zenodo.3949207}
    }

The DOI and URL in this entry point always to the latest release of iminuit. You can also cite the actual release that you used, please follow the `Zenodo link `_, which offers entries for common bibliography formats for all iminuit releases.

The recommended scientific reference for the MINUIT algorithms is::

    @article{James:1975dr,
        author = "James, F. and Roos, M.",
        title = "{Minuit: A System for Function Minimization and Analysis of the Parameter Errors and Correlations}",
        reportNumber = "CERN-DD-75-20",
        doi = "10.1016/0010-4655(75)90039-9",
        journal = "Comput. Phys. Commun.",
        volume = "10",
        pages = "343--367",
        year = "1975"
    }

Partner projects
----------------

* `numba_stats`_ provides faster implementations of probability density functions than scipy, and a few specific ones used in particle physics that are not in scipy.
* `boost-histogram`_ from Scikit-HEP provides fast generalized histograms that you can use with the builtin cost functions.
* `jacobi`_ provides a robust, fast, and accurate calculation of the Jacobi matrix of any transformation function and building a function for generic error propagation.

Versions
--------

The 2.x series has introduced breaking interfaces changes with respect to the 1.x series. There are no plans to introduce further breaking changes.

All interface changes from 1.x to 2.x are documented in the `changelog`_ with recommendations how to upgrade. To keep old scripts running, pin your major iminuit version to <2: the command ``pip install 'iminuit<2'`` installs the 1.x series.

.. _changelog: https://scikit-hep.org/iminuit/changelog.html
.. _tutorials: https://scikit-hep.org/iminuit/tutorials.html
.. _discussions: https://github.com/scikit-hep/iminuit/discussions
.. _gitter: https://gitter.im/Scikit-HEP/iminuit
.. _jacobi: https://github.com/hdembinski/jacobi
.. _numba_stats: https://github.com/HDembinski/numba-stats
.. _boost-histogram: https://github.com/scikit-hep/boost-histogram
.. _numba: https://numba.pydata.org
.. _RooFit: https://root.cern/manual/roofit/

Owner

Citation (CITATION.cff)

# This CITATION.cff file was generated with cffinit.
# Visit https://bit.ly/cffinit to generate yours today!

cff-version: 1.2.0
title: scikit-hep/iminuit
message: >-
  If you use this software, please cite it using the
  metadata from this file and the MINUIT paper at
  https://doi.org/10.1016/0010-4655(75)90039-9.

  This software includes source code under other
  licenses, see LICENSE in the repository for details.
type: software
authors:
  - given-names: Hans
    family-names: Dembinski
    email: hans.dembinski@gmail.com
    affiliation: TU Dortmund
    orcid: 'https://orcid.org/0000-0003-3337-3850'
  - given-names: Piti
    family-names: Ongmongkolkul
  - given-names: Christoph
    family-names: Deil
  - given-names: Henry
    family-names: Schreiner
  - given-names: Matthew
    family-names: Feickert
  - given-names: Chris
    family-names: Burr
  - given-names: Jason
    family-names: Watson
  - given-names: Fabian
    family-names: Rost
  - given-names: Alex
    family-names: Pearce
  - given-names: Lukas
    family-names: Geiger
  - given-names: Ahmed
    family-names: Abdelmotteleb
  - given-names: Aman
    family-names: Desai
  - given-names: Bernhard M.
    family-names: Wiedemann
  - given-names: Christoph
    family-names: Gohlke
  - given-names: Jeremy
    family-names: Sanders
  - given-names: Jonas
    family-names: Drotleff
  - given-names: Jonas
    family-names: Eschle
  - given-names: Ludwig
    family-names: Neste
  - given-names: Marco Edward
    family-names: Gorelli
  - given-names: Max
    family-names: Baak
  - given-names: Michael
    family-names: Eliachevitch
  - given-names: Omar
    family-names: Zapata
identifiers:
  - type: doi
    value: 10.5281/zenodo.7695764
repository-code: 'https://github.com/scikit-hep/iminuit'
url: 'https://scikit-hep.org/iminuit/'
abstract: >-
  iminuit is a Jupyter-friendly Python interface for the
  Minuit2 C++ library maintained by CERN's ROOT team.
keywords:
  - Python
  - C++
  - fitting
  - optimization
  - statistics
  - data analysis
  - Scikit-HEP
license: MIT

GitHub Events

Committers

Last synced: 7 months ago

Issues and Pull Requests

Last synced: 4 months ago

Packages