Q-MM is a python toolbox for Quadratic Majorization-Minimization for fast optimization of differentiable criterion.

You can also download a zip file. To cite the package

   title = {Q-MM: The Quadratic Majorize-Minimize Python toolbox},
   author = {Orieux, Fran\c{c}ois},
   url = {https://github.com/forieux/qmm},



I contribute to the deconvolution algorithms of the scikit-image library.


  • A Conjugate Gradient descent optimisation function design for quadratic criterion and linear problem defined on large unknown (compatible with complex unknown). Any feedback, patch and issue report are welcome.
  • An unsupervised deconvolution based on the Wiener-Hunt filter. The estimation of the regularisation parameter is done thanks to a Gibbs sampler.

Some extra

  • Several python utility and small tools here
  • Several Matlab functions (related to Fourier transform, optimisation, MCMC and various utility) in this archive

Jupyter notebooks

  • A notebook of PO algorithm illustrated on a Super-Resolution problem

In a general way, the source code are under free license (open source license if you prefer), typically MIT or BSD or GPL. This mean that you can copy, share and adapt the work with minor restriction.