Convergence of nonsmooth descent methods via Kurdyka-Lojasiewicz inequality on Riemannian manifolds

Seyedehsomayeh Hosseini

We examine convergence of subgradient-oriented descent methods
in nonsmooth optimization on Riemannian manifolds. We prove convergence
in the sense of subsequences for nonsmooth functions whose standard model
is strict. Adding the Kurdyka-Lojasiewicz condition, we demonstrate convergence to a singular critical point.