Automatic Directional Differentiationof Nonsmooth Composite
نویسنده
چکیده
We consider continuous functions that are deened by programs for their evaluation. The basic arithmetic operations and univariate special functions are real analytic in the interior of their domains. However , another frequent ingredient, the absolute value function, has a kink at the origin, which occurs similarly for max and min. A slightly more serious complication arises with the introduction of Euclidean vector norms. It is shown here that the resulting class of composite functions is still directionally real analytic and we develop formulas for propagating the corresponding directional Taylor-coeecients in the forward mode of automatic diierentiation. Finally, we discuss possibilities for using the reverse mode to compute generalized gradients.
منابع مشابه
Sufficiency and duality for a nonsmooth vector optimization problem with generalized $alpha$-$d_{I}$-type-I univexity over cones
In this paper, using Clarke’s generalized directional derivative and dI-invexity we introduce new concepts of nonsmooth K-α-dI-invex and generalized type I univex functions over cones for a nonsmooth vector optimization problem with cone constraints. We obtain some sufficient optimality conditions and Mond-Weir type duality results under the foresaid generalized invexity and type I cone-univexi...
متن کاملAutomatic Face Recognition via Local Directional Patterns
Automatic facial recognition has many potential applications in different areas of humancomputer interaction. However, they are not yet fully realized due to the lack of an effectivefacial feature descriptor. In this paper, we present a new appearance based feature descriptor,the local directional pattern (LDP), to represent facial geometry and analyze its performance inrecognition. An LDP feat...
متن کاملController Design via Nonsmooth Multidirectional Search
We propose an algorithm which combines multi-directional search (MDS) with nonsmooth optimization techniques to solve difficult problems in automatic control. Applications include static and fixed-order output feedback controller design, simultaneous stabilization, H2/H∞ synthesis and much else. We show how to combine direct search techniques with nonsmooth descent steps in order to obtain conv...
متن کاملA necessary second-order optimality condition in nonsmooth mathematical programming
Generalized second–order directional derivatives for nonsmooth real–valued functions are studied and their connections with second–order variational sets are investigated. A necessary second–order optimality condition for problems with inequality constraints is obtained.
متن کاملStudy of a Coeecient Control Problem Using Nonsmooth Analysis
This report is concerned with a control problem associated to a semilinear second order ordinary diierential equation with pointwise state constraints. The control acts as a coeecient of the state equation. The nonlinear part of the equation is governed by a Nemytskij operator deened by a nonsmooth function. We prove the existence of optimal controls and obtain a necessary optimality condition ...
متن کامل