Welcome to the UnLocBox (Matlab convex optimization toolbox) sourceforge page.
Get the latest version from the download page.
- New version of the UNLocBoX. Now you can stop the algorithms using the keyboard "ctrl+d" in the command window! 17th September 2014
- Alpha release of the pyunlocbox: a python implementation of the UNLocBoX. 11th June. 2014
- New version released. 29th April. 2014
- Add 4 dimentional TV proximal operator
- Add weights option to tv proximal operators
- Simplification of structured sparsity through the use of matrices
- Correction of small bugs in ADMM
- New version released. 24th Feb. 2014
- Refactor the code of all demonstration files
- Add "utils" and "signals" fonctions
- Corrections of many bugs
- New version released. 1st oct. 2013
- WARNING: use of cell of structure to group functionals.
- Start official support for GPU computation. Documentation coming soon
- Customize you algorithm... Documentation coming soon.
- To evaluate a function representing by a structure called f, we use f.eval instead of f.norm... 28th jan. 2013
- Version 1.1 availlable (GPU experimental), documentation for GPU not availlable 28th jan. 2013
- Adding gpu computation (TV) --> availlable in the release 1.1 (next week) 14th jan. 2013
- New proximal operators (prox_l2 and prox_l2grad) 28th nov. 2013
- The version 1 of the UnLocBoX is is available. 1st nov. 2013
- The prox used in grouped sparsity(prox_l12 and prox_l1inf) have been debugged.
- If you are using the old version 0.3 of the UNLocBox and upgrade to the newest version, please read this document. It takes 3 minutes, but will save hours.
This site contains a brief description of the convex optimization, as well as the MATLAB toolbox implementing the main algorithms.
Summary of the UnlocBox algorithms are in the following papers
"Proximal Splitting Methods in Signal Processing"
Patrick L. Combettes, Jean-Christophe Pesquet
Fixed-Point algorithms for inverse problems in science and engineering.