On Linearity of Nonclassical Differentiation
Serguei Samborski
Abstract
We introduce a real vector space composed of set-valued maps on an open set X and note it by S. It is a complete metric space and a complete lattice. The set of continuous functions on X is dense in S as in a metric space and as in a lattice. Thus the constructed space plays the same role for the space of continuous functions with uniform convergence as the field of reals plays for the field of rationals. The classical gradient may be extended in the space S as a close operator. If a function f belongs to its domain then f is locally lipschitzian and the values of our gradient coincide with the values of Clarke's gradient. However, unlike Clarke's gradient, our gradient is a linear operator.
Create a lesson
Related papers
Risk-Sensitive Stochastic Optimal Control on Positive Cones
Mohamad Al Ahdab, John-Josef Leth
Deterministic Oracle Complexity for Goldstein Stationarity in Nonsmooth Nonconvex Optimization with Constraints
Yixi Ding, Ya-xiang Yuan
Technical Report: Partially Asynchronous Distributed Optimization on Riemannian Manifolds
Adam Pooley, William Warke, Matthew T. Hale
Multi-Step Performance Estimation for Gradient Descent under the Polyak Lojasiewicz Condition
Wanli Hong, Shuyang Ling
Robust Markov Decision Processes: An Introduction Using Convex Optimization and Game Theory
Julien Grand-Clément
Dual-Mode Extremum Seeking for Diffusion Systems with Boundary-Cost Feedback
Martin Guay, Denis Dochain