Computational Convex Analysis (CCA) numerical library CCA numerical library for Computational Convex Analysis Copyright (C) 2008-2017, Yves Lucet
This toolbox implements functions for convex operations.
Here are the main files provided by the package.
README.txt The present file
plq_version.m Operations on Piecewise Linear-Quadratic functions, which are represented as matrices. Algorithms are included to perform PLQ addition, scalar multiplication, evaluation, minimization and maximization, projection, and inf-convolution and deconvolution, as well as compute Fenchel conjugates, Moreau envelopes, convex hulls, proximal averages, and Rockafellar functions, and restricted Fitzpatrick functions. A PLQ approximation of an arbitrary convex function can be built. There are functions to plot PLQ functions and their averages. See help plq_function for the complete list of functions.
For more details on the Scilab functions defined in the main toolbox files, refer to the help documentation.
runtests('tests') Matlab script that runs all unit tests and displays the results
test.m Unit test files that validate the algorithms and provide numerous examples of their use. Test files included are: plq_test.m: Unit tests and examples for the PLQ functions provided in plq_test.m. plt_test.m: Tests the Parametric Moreau Envelope algorithm pl_me_plt, which uses the Parametric Legendre Transform. rock_test.m: Unit tests for the computation of Rockafellar functions. Each of the unit test files can be called independently by giving the command exec "/tests/name_of_test.m"; or exec (CCADIR+'/tests/name_of_test.m'); Then calling: runtest('name_of_test.m') will return a true if the tests succeeded, false otherwise.
See copyright.txt for the list of contributors to this library.