The Computational Convex Analysis numerical library contains efficient
algorithms to manipulate convex and nonconvex univariate functions, and to
compute fundamental convex analysis transforms. Functions can be entered
explicitly or through a black box, and are approximated with quadratic splines.
The primary transforms implemented are the Legendre-Fenchel transform, the
Moreau envelope, the (generalized) proximal average, the convex hull, the
subdifferential, and Rockafellar and Fitzpatrick functions.
The library contains extensive demos, and numerous tests (476 unit tests with a
statement coverage of 85%).