CHARACTERIZING AND TESTING SUBDIFFERENTIAL REGULARITY FOR PIECEWISE SMOOTH OBJECTIVE FUNCTIONS

Functions defined by evaluation programs involving smooth elementals and absolute values as well as the maxand min-operator are piecewise smooth. Using piecewise linearization we derived in [7] for this class of nonsmooth functions φ first and second order conditions for local optimality (MIN). They are necessary and sufficient, respectively. These generalizations of the classical KKT and SSC theory assumed that the given representation of φ satisfies the LinearIndependence-Kink-Qualification (LIKQ). In this paper we relax LIKQ to the Mangasarin-FromovitzKink-Qualification (MFKQ) and develop a constructive condition for a local convexity concept, i.e., the convexity of the local piecewise linearization on a neighborhood. As a consequence we show that this first order convexity (FOC) is always required by subdifferential regularity (REG) as defined in [20], and is even equivalent to it under MFKQ. Whereas it was observed in [7] that testing for MIN is polynomial under LIKQ, we show here that even under this strong kink qualification, testing for FOC and thus REG is co-NP complete. We conjecture that this is also true for testing MFKQ itself.

[1]  Andreas Griewank,et al.  First- and second-order optimality conditions for piecewise smooth objective functions , 2016, Optim. Methods Softw..

[2]  Andreas Griewank,et al.  On Lipschitz optimization based on gray-box piecewise linearization , 2016, Math. Program..

[3]  Paul I. Barton,et al.  A vector forward mode of automatic differentiation for generalized derivative evaluation , 2015, Optim. Methods Softw..

[4]  Heinz H. Bauschke,et al.  On the convexity of piecewise-defined functions , 2014, 1408.3771.

[5]  Andreas Griewank,et al.  On stable piecewise linearization and generalized algorithmic differentiation , 2013, Optim. Methods Softw..

[6]  Paul I. Barton,et al.  Evaluating an element of the Clarke generalized Jacobian of a composite piecewise differentiable function , 2013, TOMS.

[7]  Yves Lucet,et al.  Computing the conjugate of convex piecewise linear-quadratic bivariate functions , 2013, Math. Program..

[8]  A. Griewank,et al.  A method for pointwise evaluation of polyconvex envelopes , 2013 .

[9]  Mert Gürbüzbalaban,et al.  On Nesterov's Nonsmooth Chebyshev-Rosenbrock Functions , 2012 .

[10]  S. Scholtes Piecewise Differentiable Functions , 2012 .

[11]  B. Mordukhovich Variational analysis and generalized differentiation , 2006 .

[12]  Yurii Nesterov,et al.  Lexicographic differentiation of nonsmooth functions , 2005, Math. Program..

[13]  Robert Mifflin,et al.  Primal-Dual Gradient Structured Functions: Second-Order Results; Links to Epi-Derivatives and Partly Smooth Functions , 2003, SIAM J. Optim..

[14]  Andreas Griewank,et al.  Evaluating derivatives - principles and techniques of algorithmic differentiation, Second Edition , 2000, Frontiers in applied mathematics.

[15]  R. Tyrrell Rockafellar,et al.  Variational Analysis , 1998, Grundlehren der mathematischen Wissenschaften.

[16]  A. S. Cullick,et al.  New strategy for phase equilibrium and critical point calculations by thermodynamic energy analysis. Part I. Stability analysis and flash , 1991 .

[17]  A. S. Cullick,et al.  New strategy for phase equilibrium and critical point calculations by thermodynamic energy analysis. Part II. Critical point calculations , 1991 .

[18]  O. Mangasarian,et al.  The Fritz John Necessary Optimality Conditions in the Presence of Equality and Inequality Constraints , 1967 .