\(\varepsilon\)-subgradient algorithms for locally Lipschitz functions on Riemannian manifolds (Q273664)

The paper deals with the optimization of non-smooth functions on Riemannian manifolds, and to this end it is necessary to generalize non-smooth analysis to Riemannian manifolds. To this end, the authors introduce the concept of Riemannian \(\varepsilon\)-subdifferential of a function considered as an extension of the Clarke generalized directional derivative, and it is shown that the problem is completely defined by the knowledge of this generalized derivative. The optimization process works via a so-called method of descent direction. Numerical examples illustrate the theory.