Interactive IPython Friendly Mimizer based on SEAL Minuit. (It's included in the package no need to install it separately)

It is designed from ground up to be fast, interactive and cython friendly. iminuit extract function signature very permissively starting from checking func_code down to last resort of parsing docstring(or you could tell iminuit to stop looking and take your answer). The interface is inspired heavily by PyMinuit and the status printout is inspired by ROOT Minuit. iminuit is mostly compatible with PyMinuit(with few exceptions). Existing PyMinuit code can be ported to iminuit by just changing the import statement.

In a nutshell:

from iminuit import Minuit
def f(x,y,z):
    return (x-2)**2 + (y-3)**2 + (z-4)**2
m = Minuit(f)
m.migrad()
print m.values #{'x':2,'y':3,'z':4}
print m.errors

python setup.py install

or from pip:

pip install iminuit

for windows Christoph Gohlke made a nice windows library to save you all from windows compilation nightmare

http://www.lfd.uci.edu/~gohlke/pythonlibs/#iminuit

All the tutorials are in tutorial directory. You can view it online too.

• Quick start
• Hard Core Cython tutorial. If you need to do a huge likelihood fit that need speed or learn how to parallelize your stuff, this is for you.

http://iminuit.readthedocs.org/

Using it as a black box is a bad idea. Here are some fun read the order is given by the order I think you should read.

• Wikipedia for Quasi Newton Method and DFP formula. The magic behind migrad.
• Variable Metric Method for Minimization William Davidon 1991
• A New Approach to Variable Metric Algorithm. (R.Fletcher 1970)
• Original Paper: MINUIT - A SYSTEM FOR FUNCTION MINIMIZATION AND ANALYSIS OF THE PARAMETER ERRORS AND CORRELATIONS by Fred James and Matts Roos.

Github allows you to contribute to this project very easily just fork the repository, make changes and submit a pull request.

Here's the list of concrete open issues and feature requests: https://github.com/iminuit/iminuit

More generally any contribution to the docs, tests and package itself is welcome!

• Documentation. Tell us what's missing, what's incorrect or misleading.
• Tests. If you have an example that shows a bug or problem, please file an issue!
• Performance. I you are a C/cython/python hacker go ahead and make it faster.