The gradient method in problems of infinite-dimensional optimization

Convergence of gradient procedures is established for non-convex optimization problems when the data on non-singularity of the minimum point is lacking. Applications to optimal control variational calculus and mechanics problems are indicated.