DOCSLIB.ORG

Variational Properties of Polynomial Root Functions and Spectral Max Functions

Total Page：16

File Type：pdf, Size：1020Kb

Download full-text PDF Read full-text
Variational Properties of Polynomial Root Functions and Spectral Max Functions Julia Eaton Workshop on Numerical Linear Algebra and Optimization [email protected] University of British Columbia University of Washington Tacoma Vancouver, BC Canada Variational properties of polynomial root functions and spectral max functions Abstract 1 Spectral Functions 4 Damped oscillator abscissa (see panel 3) 9 Applications 14 n×n Eigenvalue optimization problems arise in the control of continuous We say :C !R=R [ f+1g is a spectral function if it • Smooth everywhere but -2 and 2. Recall damped oscillator problem in panel 3. The characteristic and discrete time dynamical systems. The spectral abscissa (largest • depends only on the eigenvalues of its argument • Non-Lipschitz at -2 and 2 polynomial of A(x) is: p(λ, x) = λ(λ+x)+1 = λ2+xλ+1: real part of an eigenvalue) and spectral radius (largest eigenvalue • is invariant under permutations of those eigenvalues • Non-convex and abscissa graph is shown in panel 3. in modulus) are examples of functions of eigenvalues, or spectral • @h^ (2) = [−1=2; 1) = @h(2) A spectral max function ': n×n ! is a spectral fcn. of the form Consider the parameterization H : !P2, where functions, connected to these problems. A related class of functions C R • subdifferentially regular at 2 R '(A) = maxff (λ) j det(λI −A) = 0g H(x) = (λ−λ )2+h (x)(λ−λ )+h (x); λ 2 , h ; h : ! . are polynomial root functions. In 2001, Burke and Overton showed 0 1 0 2 0 C 1 2 R R where f : C ! R. The spectral abscissa and radius are spectral max 2 2 that the abscissa mapping on polynomials is subdifferentially Variational properties of spectral functions 10 Require p(λ, x) = λ +xλ+1 = (λ−λ0) +h1(x)(λ−λ0)+h2(x) so functions. (Note: f C ! R is treated as a function from R2 to R.) 2 T regular on the monic polynomials of degree n. In 2012 we extended Familiar spectral functions: abscissa α and radius ρ. h1(x) = 2λ0+x, h2(x) = λ0+xλ0+1, rH(x) = [0; 1; λ0] n×n these results to a class of max polynomial root functions which • α and ρ continuous on C , not Lipschitz continuous Set h = a ◦ H, so h(x) = a(p(λ, x)): By chain rule (panel 10), Optimizing Spectral Functions 5 includes both the polynomial abscissa and the polynomial radius. • α and ρ are not differentiable in general @h^ (x) = rH(x)T @^a(p(λ, x)): We are currently working to extend these results to the matrix setting. We saw an example of a simple abscissa minimization problem Compute regular subgradients? If subdifferentially regular... e e e where the matrix depended on a single parameter. More generally Global min of h is 2, so p(λ, 2) = λ2+2λ+1 = (λ+1)2 and (Chain Rule, Rockafellar & Wets ‘98) ^ T 2 we can consider min ( A(x)) @h(2) = rH(2) @a((λ+1) ): Let g : m ! be regular at H(x), H smooth, h=g ◦ H. CQ: the only 2 x2U R R e 2 Motivation k From 2001, @a((λ+1) ) = fz jz = −(λ+1)=2 + µ ; Re (µ ) ≤ 0g: where U ⊂C is some parameter set. How do we verify a candidate vector z 2@1g(H(x)) with rH(x)∗z =0 is z =0. If (CQ) is satisfied, then 2 2 Consider the linear continuous-time dynamical system e e D E 0 soln is optimal? We need to know about the structure of . @h(x)=rH(x)∗@g(H(x));@1h(x)=rH(x)∗@1g(H(x)): ^ T 2 (DE) y = Ay: e e e e e e Set λ0 =−1: @h(2) = [0; 1; −1] ;@a((λ+1) ) n×n where A2C . What is the long-term behavior of a solution? • Gradients can give necessary conditions, provided they exist. = fRe(−1=2−µ )j Re(µ ) ≤ 0g = [−1=2; 1) α regular at X iff X’s active evals are nonderogatory (Burke, Overton 01) 2 2 • Parameterization means a composite fcn—use a chain rule? Therefore 02int (@h^ (2)). Minimizer 2 is sharp. Decomposition 11 n Some variational analysis 6 •P : linear space of polynomials of degree 0; 1; : : : ; n Applying the polynomial radius to the family 15 Let ye be an equilibrium solution to dy=dt = f(t; y). Let h : E ! . •M n ⊂Pn: polynomials of degree n 2 R fp(λ, x)gx2R = fλ +xλ+1gx2R, • ye is Lyapunov stable if for any " > 0, 9 δ" > 0 such that The regular subdifferential of h at x is the set • Define Φ: n×n !Pn by Φ(A) = det(λI −A): e C we have ( p jy(t) − yej ≤ " whenever jy(0) − yej ≤ δ" ^ Define the max root function f : Pn ! by (jxj + x2 − 4)=2 jxj ≥ 2 @h(xe) = fz j h(x) − h(xe) ≥ hz; x−xei + o(kx−xek); 8xg R • y asymptotically stable 9γ > 0 r(p(λ, x)) = e is if it is LS and such that We say x is sharp if 9 δ; ">0 such that h(x)−h(x)≥δ kx−xk f(p) = maxff (λ) j p(λ) = 0g. 1 jxj < 2 jy(t) − y j ! 0 jy(t) − y j < γ e e e e whenever e for all x for which kx−xk<". Then ' = f ◦ Φ. Let h = r ◦ H, so h(x) = r(p(λ, x)). A global minimum is at 2. e D E A stable solution ye to (DE) is asymptotically stable if the @h^ (2) = [0; 1; −1]T ;@r((λ+1)2 ) Fact: (Burke, Overton, 2000): xe is a sharp local minimizer of h iff Properties of polynomial root functions 12 eigenvalues of A lie in the left-half plane. ^ 02int(@h(xe)). If f is continuous on its domain, = fRe(1=2−µ2) jRe(µ2) ≤ 1=2g • Define the spectral abscissa α : n×n ! by • f Mn \ dom (f) = [0; 1) C R Example reg. subgradients of the continuous on p , not generally Lipschitz α(A) = maxfRe(λ) j det(λI −A)=0g n n ^ damped oscillator abscissa: a(λ − ") = " 0 2= int(@h(2)), so minimizer is not sharp. • y α(A)<0 Then e is asympotically stable if . Set h(·)=α(A(·)): • f not differentiable in general n n • Also, the rate of decay of a soln. depends on α(A). z 2f−1=2g=@h^ (−3=5) • f is unbounded in a neighborhood of q for all q(λ)2P n M n−1 ^ ^ let q(λ)2P , then q(λ)(ελ−1) ! q(λ), Matrix Setting 16 k+1 k n×n z 2@h(−2), z 2@h(2) Next consider the discrete-time analog: y =Ay ;A2C . but root 1=" is unbounded In 2001 Burke & Overton derived many properties of a broad class The subderivative (generalizes the directional derivative) • Asymptotic stability tied to max modulus of eigenvalues of A. Restrict our attention to Mn. of spectral functions as well as the subdifferential regularity of the n×n dh(x)(v)= lim inf ( h(x+tv) − h(x))=t: • The spectral radius ρ : C ! R is given by e e e e spectral abscissa at a matrix with nonderogatory active eigenvalues. t&0;v!ve Polynomial max root history ρ(A)=maxfjλj j det(λI −A)=0g. In 2013 Grundel & Overton derived the formula for the A key subderivative & regular subdifferential relationship: • A stable soln is asymptotically stable if ρ(A)<1. Let f : C ! R be proper, convex, lsc. subdifferential of the simplest example of a derogatory, defective @^h(x)=fz j hz; vi ≤ dh(x)(v) 8 vg e e f(q)=maxff (q) j q(λ) = 0g matrix: a 3 × 3 matrix w/ one eigenvalue and two Jordan blocks. • If f =Re(·), the polynomial abscissa a. Well studied for general We have been working to extend the abscissa results to a broader Example 3 The general subdifferential (limits of regular subgradients) 7 p=Q(λ−λ )nj 2Mn (Burke & Overton ‘01) 00 0 j 1 class of spectral functions. We have shown the subdifferential (Damped oscillator) Consider the system v + xv + v = 0. 9fxνg⊆dom (h) and fzνg⊆E such that • For general f , f is studied for polynomials (λ−λ )n, regularity 0 @h(xe) = z ^ 0 regularity in the case of nonderogatory active eigenvalues for a very v 0 1 v zν 2@h(xν) 8 ν; xν !x;e h(xν)!h(xe); and zν !z not shown. (Burke, Lewis, Overton ‘05) Linearizing gives 0 = 0 , narrow class of functions which includes the abscissa but not the v −1 −x v The horizon subdifferential (captures non-Lipschitz-ness) • Done: fill the gap, and clarify process, go beyond max root fcn. p radius. Extending the methods in the 2001 paper has proven 2 1 9fxνg⊆dom (h); ftνg⊆[0; 1), fzνg⊆E; s.t. 8 ν • To do: push through to the matrix setting. The matrix (call it A(x)) has eigenvalues (−x ± x − 4)=2 @ h(x) = z difficult. As demonstrated in G & O’s 2013 paper, even the 3 × 3 p e zν 2@^h(xν) xν !x; h(xν)!h(x); tν #0 and tνzν !z ( 2 e e case is extremely challenging and does not easily lend itself to (−x + x − 4)=2 jxj≥2 Polynomial radius subdifferential 13 α (A(x)) = ^ ^ 1 1 generalization. We are currently exploring other avenues to derive Computing, h is regular at x if @h(x)=6 ;, @h(x)=@h(x) and @h(x) =@ h(x) r(p) = maxfjλj j p(λ) = 0g −x=2 jxj<2 e e e e e e variational properties of spectral functions.
Load more

Recommended publications
  • Subderivative-Subdifferential Duality Formula
    Subderivative-subdifferential duality formula Marc Lassonde Universit´edes Antilles, BP 150, 97159 Pointe `aPitre, France; and LIMOS, Universit´eBlaise Pascal, 63000 Clermont-Ferrand, France E-mail: [email protected] Abstract. We provide a formula linking the radial subderivative to other subderivatives and subdifferentials for arbitrary extended real-valued lower semicontinuous functions. Keywords: lower semicontinuity, radial subderivative, Dini subderivative, subdifferential. 2010 Mathematics Subject Classification: 49J52, 49K27, 26D10, 26B25. 1 Introduction Tyrrell Rockafellar and Roger Wets [13, p. 298] discussing the duality between subderivatives and subdifferentials write In the presence of regularity, the subgradients and subderivatives of a function f are completely dual to each other. [. ] For functions f that aren’t subdifferentially regular, subderivatives and subgradients can have distinct and independent roles, and some of the duality must be relinquished. Jean-Paul Penot [12, p. 263], in the introduction to the chapter dealing with elementary and viscosity subdifferentials, writes In the present framework, in contrast to the convex objects, the passages from directional derivatives (and tangent cones) to subdifferentials (and normal cones, respectively) are one-way routes, because the first notions are nonconvex, while a dual object exhibits convexity properties. In the chapter concerning Clarke subdifferentials [12, p. 357], he notes In fact, in this theory, a complete primal-dual picture is available: besides a normal cone concept, one has a notion of tangent cone to a set, and besides a subdifferential for a function one has a notion of directional derivative. Moreover, inherent convexity properties ensure a full duality between these notions. [. ]. These facts represent great arXiv:1611.04045v2 [math.OC] 5 Mar 2017 theoretical and practical advantages.
    [Show full text]
  • Techniques of Variational Analysis
    J. M. Borwein and Q. J. Zhu Techniques of Variational Analysis An Introduction October 8, 2004 Springer Berlin Heidelberg NewYork Hong Kong London Milan Paris Tokyo To Tova, Naomi, Rachel and Judith. To Charles and Lilly. And in fond and respectful memory of Simon Fitzpatrick (1953-2004). Preface Variational arguments are classical techniques whose use can be traced back to the early development of the calculus of variations and further. Rooted in the physical principle of least action they have wide applications in diverse ¯elds. The discovery of modern variational principles and nonsmooth analysis further expand the range of applications of these techniques. The motivation to write this book came from a desire to share our pleasure in applying such variational techniques and promoting these powerful tools. Potential readers of this book will be researchers and graduate students who might bene¯t from using variational methods. The only broad prerequisite we anticipate is a working knowledge of un- dergraduate analysis and of the basic principles of functional analysis (e.g., those encountered in a typical introductory functional analysis course). We hope to attract researchers from diverse areas { who may fruitfully use varia- tional techniques { by providing them with a relatively systematical account of the principles of variational analysis. We also hope to give further insight to graduate students whose research already concentrates on variational analysis. Keeping these two di®erent reader groups in mind we arrange the material into relatively independent blocks. We discuss various forms of variational princi- ples early in Chapter 2. We then discuss applications of variational techniques in di®erent areas in Chapters 3{7.
    [Show full text]
  • Proximal Analysis and Boundaries of Closed Sets in Banach Space Part Ii: Applications
    Can. J. Math., Vol. XXXIX, No. 2, 1987, pp. 428-472 PROXIMAL ANALYSIS AND BOUNDARIES OF CLOSED SETS IN BANACH SPACE PART II: APPLICATIONS J. M. BORWEIN AND H. M. STROJWAS Introduction. This paper is a direct continuation of the article "Proximal analysis and boundaries of closed sets in Banach space, Part I: Theory", by the same authors. It is devoted to a detailed analysis of applications of the theory presented in the first part and of its limitations. 5. Applications in geometry of normed spaces. Theorem 2.1 has important consequences for geometry of Banach spaces. We start the presentation with a discussion of density and existence of improper points (Definition 1.3) for closed sets in Banach spaces. Our considerations will be based on the "lim inf ' inclusions proven in the first part of our paper. Ti IEOREM 5.1. If C is a closed subset of a Banach space E, then the K-proper points of C are dense in the boundary of C. Proof If x is in the boundary of C, for each r > 0 we may find y £ C with \\y — 3c|| < r. Theorem 2.1 now shows that Kc(xr) ^ E for some xr e C with II* - *,ll ^ 2r. COROLLARY 5.1. ( [2] ) If C is a closed convex subset of a Banach space E, then the support points of C are dense in the boundary of C. Proof. Since for any convex set C and x G C we have Tc(x) = Kc(x) = Pc(x) = P(C - x) = WTc(x) = WKc(x) = WPc(x\ the /^-proper points of C at x, where Rc(x) is any one of the above cones, are exactly the support points.
    [Show full text]
  • Convexity: an Introduction
    Convexity: an introduction Convexity: an introduction Geir Dahl CMA, Dept. of Mathematics and Dept. of Informatics University of Oslo 1 / 74 Convexity: an introduction 1. Introduction 1. Introduction what is convexity where does it arise main concepts and results Literature: Rockafellar: Convex analysis, 1970. Webster: Convexity, 1994. Gr¨unbaum: Convex polytopes, 1967. Ziegler: Lectures on polytopes, 1994. Hiriart-Urruty and Lemar´echal: Convex analysis and minimization algorithms, 1993. Boyd and Vandenberghe: Convex optimization, 2004. 2 / 74 Convexity: an introduction 1. Introduction roughly: a convex set in IR2 (or IRn) is a set \with no holes". more accurately, a convex set C has the following property: whenever we choose two points in the set, say x; y 2 C, then all points in the line segment between x and y also lie in C. a sphere (ball), an ellipsoid, a point, a line, a line segment, a rectangle, a triangle, halfplane, the plane itself the union of two disjoint (closed) triangles is nonconvex. 3 / 74 Convexity: an introduction 1. Introduction Why are convex sets important? Optimization: mathematical foundation for optimization feasible set, optimal set, .... objective function, constraints, value function closely related to the numerical solvability of an optimization problem Statistics: statistics: both in theory and applications estimation: \estimate" the value of one or more unknown parameters in a stochastic model. To measure quality of a solution one uses a \loss function" and, quite often, this loss function is convex. statistical decision theory: the concept of risk sets is central; they are convex sets, so-called polytopes. 4 / 74 Convexity: an introduction 1.
    [Show full text]
  • Arxiv:1703.03069V1 [Math.OC]
    Upper semismooth functions and the subdifferential determination property March 10, 2017 Marc Lassonde Universit´edes Antilles, BP 150, 97159 Pointe `aPitre, France; and LIMOS, Universit´eBlaise Pascal, 63000 Clermont-Ferrand, France E-mail: [email protected] Dedicated to the memory of Jon Borwein. Abstract. In this paper, an upper semismooth function is defined to be a lower semicon- tinuous function whose radial subderivative satisfies a mild directional upper semicontinuity property. Examples of upper semismooth functions are the proper lower semicontinuous convex functions, the lower-C1 functions, the regular directionally Lipschitz functions, the Mifflin semismooth functions, the Thibault-Zagrodny directionally stable functions. It is shown that the radial subderivative of such functions can be recovered from any subdiffer- ential of the function. It is also shown that these functions are subdifferentially determined, in the sense that if two functions have the same subdifferential and one of the functions is upper semismooth, then the two functions are equal up to an additive constant. Keywords: upper semismooth, Dini subderivative, radial subderivative, subdifferential, sub- differential determination property, approximately convex function, regular function. 2010 Mathematics Subject Classification: 49J52, 49K27, 26D10, 26B25. 1 Introduction Jon Borwein discussing generalisations in the area of nonsmooth optimisation [1, p. 4] writes: In his thesis Francis Clarke extended Moreau-Rockafellar max formula to all locally Lip- schitz functions. Clarke replaced the right Dini directional derivative Dhf(x) by c f(y + th) f(y) Dhf(x) = lim sup − . 0<t→0,y→x t arXiv:1703.03069v1 [math.OC] 8 Mar 2017 c Somewhat miraculously the mapping p sending h Dhf(x) is always continuous and C → ∗ c sublinear in h, and so if we define ∂ f(x)= ∂p(0) = y X : y,h Dhf(x), h X , Moreau-Rockafellar max formula leads directly to: { ∈ h i≤ ∀ ∈ } Theorem (Clarke).
    [Show full text]
  • For Lower Semicontinuous Functions
    TRANSACTIONSof the AMERICAN MATHEMATICALSOCIETY Volume 347, Number 10, October 1995 MEAN VALUE PROPERTY AND SUBDIFFERENTIAL CRITERIA FOR LOWER SEMICONTINUOUS FUNCTIONS DIDIER AUSSEL, JEAN-NOËL CORVELLEC, AND MARC LASSONDE Abstract. We define an abstract notion of subdifferential operator and an as- sociated notion of smoothness of a norm covering all the standard situations. In particular, a norm is smooth for the Gâteaux (Fréchet, Hadamard, Lipschitz- smooth) subdifferential if it is Gâteaux (Fréchet, Hadamard, Lipschitz) smooth in the classical sense, while on the other hand any norm is smooth for the Clarke-Rockafellar subdifferential. We then show that lower semicontinüous functions on a Banach space satisfy an Approximate Mean Value Inequality with respect to any subdifferential for which the norm is smooth, thus pro- viding a new insight on the connection between the smoothness of norms and the subdifferentiability properties of functions. The proof relies on an adapta- tion of the "smooth" variational principle of Bonvein-Preiss. Along the same vein, we derive subdifferential criteria for coercivity, Lipschitz behavior, cone- monotonicity, quasiconvexity, and convexity of lower semicontinüous functions which clarify, unify and extend many existing results for specific subdifferen- tials. 1. Introduction Let X be a Banach space. Since the pioneering observation of Asplund [ 1] that there is a close connection between Gâteaux-differentiability of the norm of X and Gâteaux-differentiability of convex continuous functions on dense
    [Show full text]
  • Twice Epi-Differentiability of Extended-Real-Valued Functions with Applications in Composite Optimization
    TWICE EPI-DIFFERENTIABILITY OF EXTENDED-REAL-VALUED FUNCTIONS WITH APPLICATIONS IN COMPOSITE OPTIMIZATION ASHKAN MOHAMMADI1 and M. EBRAHIM SARABI2 Abstract. The paper is devoted to the study of the twice epi-differentiablity of extended-real-valued functions, with an emphasis on functions satisfying a certain composite representation. This will be conducted under parabolic regularity, a second-order regularity condition that was recently utilized in [14] for second-order variational analysis of constraint systems. Besides justifying the twice epi-differentiablity of composite functions, we obtain precise formulas for their second subderivatives under the metric subregularity constraint qualification. The latter allows us to derive second-order optimality conditions for a large class of composite optimization problems. Key words. Variational analysis, twice epi-differentiability, parabolic regularity, composite optimization, second-order optimality conditions Mathematics Subject Classification (2000) 49J53, 49J52, 90C31 1 Introduction This paper aims to provide a systematic study of the twice epi-differentiability of extend-real- valued functions in finite dimensional spaces. In particular, we pay special attention to the composite optimization problem minimize ϕ(x)+ g(F (x)) over all x ∈ X, (1.1) where ϕ : X → IR and F : X → Y are twice differentiable and g : Y → IR := (−∞, +∞] is a lower semicontinuous (l.s.c.) convex function and where X and Y are two finite dimen- sional spaces, and verify the twice epi-differentiability of the objective function in (1.1) under verifiable assumptions. The composite optimization problem (1.1) encompasses major classes of constrained and composite optimization problems including classical nonlinear programming problems, second-order cone and semidefinite programming problems, eigenvalue optimizations problems [24], and fully amenable composite optimization problems [19], see Example 4.7 for arXiv:1911.05236v2 [math.OC] 13 Apr 2020 more detail.
    [Show full text]
  • A Course in Machine Learning
    A Course in Machine Learning Hal Daumé III Draft: Copyright © 2012 Hal Daumé III http://ciml.info Do Not This book is for the use of anyone anywhere at no cost and with almost no re- strictions whatsoever. You may copy it or re-use it under the terms of the CIML License online at ciml.info/LICENSE. You may not redistribute it yourself, but are encouraged to provide a link to the CIML web page for others to download for free. You may not charge a fee for printed versions, though you can print it for your own use. Distribute version 0.8 , August 2012 6|LinearModels The essence of mathematics is not to make simple things Learning Objectives: complicated, but to make complicated things simple. -- Stan- • Define and plot four surrogate loss functions: squared loss, logistic loss, ley Gudder exponential loss and hinge loss. • Compare and contrast the optimiza- tion of 0/1 loss and surrogate loss functions. In Chapter ??, you learned about the perceptron algorithm • Solve the optimization problem for linear classification. This was both a model (linear classifier) and for squared loss with a quadratic algorithm (the perceptron update rule) in one. In this section, we regularizer in closed form. will separate these two, and consider general ways for optimizing • Implement and debug gradient descent and subgradient descent. linear models. This will lead us into some aspects of optimization (aka mathematical programming), but not very far. At the end of this chapter, there are pointers to more literature on optimization for those who are interested. The basic idea of the perceptron is to run a particular algorithm until a linear separator is found.
    [Show full text]
  • CONVEX ANALYSIS and NONLINEAR OPTIMIZATION Theory and Examples
    CONVEX ANALYSIS AND NONLINEAR OPTIMIZATION Theory and Examples JONATHAN M. BORWEIN Centre for Experimental and Constructive Mathematics Department of Mathematics and Statistics Simon Fraser University, Burnaby, B.C., Canada V5A 1S6 [email protected] http://www.cecm.sfu.ca/ jborwein ∼ and ADRIAN S. LEWIS Department of Combinatorics and Optimization University of Waterloo, Waterloo, Ont., Canada N2L 3G1 [email protected] http://orion.uwaterloo.ca/ aslewis ∼ To our families 2 Contents 0.1 Preface . 5 1 Background 7 1.1 Euclidean spaces . 7 1.2 Symmetric matrices . 16 2 Inequality constraints 22 2.1 Optimality conditions . 22 2.2 Theorems of the alternative . 30 2.3 Max-functions and first order conditions . 36 3 Fenchel duality 42 3.1 Subgradients and convex functions . 42 3.2 The value function . 54 3.3 The Fenchel conjugate . 61 4 Convex analysis 78 4.1 Continuity of convex functions . 78 4.2 Fenchel biconjugation . 90 4.3 Lagrangian duality . 103 5 Special cases 113 5.1 Polyhedral convex sets and functions . 113 5.2 Functions of eigenvalues . 120 5.3 Duality for linear and semidefinite programming . 126 5.4 Convex process duality . 132 6 Nonsmooth optimization 143 6.1 Generalized derivatives . 143 3 6.2 Nonsmooth regularity and strict differentiability . 151 6.3 Tangent cones . 158 6.4 The limiting subdifferential . 167 7 The Karush-Kuhn-Tucker theorem 176 7.1 An introduction to metric regularity . 176 7.2 The Karush-Kuhn-Tucker theorem . 184 7.3 Metric regularity and the limiting subdifferential . 191 7.4 Second order conditions . 197 8 Fixed points 204 8.1 Brouwer's fixed point theorem .
    [Show full text]
  • A Generalized Variational Principle
    Canad. J. Math. Vol. 53 (6), 2001 pp. 1174–1193 A Generalized Variational Principle Philip D. Loewen and Xianfu Wang Abstract. We prove a strong variant of the Borwein-Preiss variational principle, and show that on As- plund spaces, Stegall’s variational principle follows from it via a generalized Smulyan test. Applications are discussed. 1 Introduction The Borwein-Preiss smooth variational principle [1] is an important tool in infinite dimensional nonsmooth analysis. Its statement, given a Banach space X,readsas follows. Theorem 1.1 Given f : X → (−∞, +∞] lower semicontinuous, x0 ∈ X, ε>0, λ>0,andp≥ 1, suppose f (x0) <ε+inff . X µ ≥ ∞ µ = Then there exist a sequence n 0,with n=1 n 1, and a point v in X, expressible as the (norm-) limit of some sequence (vn),suchthat ε ε (1.1) f (x)+ (x) ≥ f (v)+ (v) ∀x ∈ X, λp p λp p = ∞ µ − p − <λ ≤ ε where p(x): n=1 n x vn .Moreover, x0 v ,andf(v) +infX f. If, in addition, X has a β-smooth norm and p > 1,then∂β f (v) ∩ (ε/λ)pB∗ = ∅. In a subsequent development [4], Deville, Godefroy and Zizler improved Theo- rem 1.1 by considering a certain Banach space Y of bounded continuous functions g : X → R, and showing that the following subset of Y is residual: {g ∈ Y : f + g attains a strong minimum somewhere in X}. (See Definition 2.1.) However, their result gives no information about the location of the strong minimizer, and offers no way to identify explicitly a perturbation g with the desired property.
    [Show full text]
  • Arithmetic Subderivatives: P-Adic Discontinuity and Continuity
    1 2 Journal of Integer Sequences, Vol. 23 (2020), 3 Article 20.7.3 47 6 23 11 Arithmetic Subderivatives: p-adic Discontinuity and Continuity Pentti Haukkanen and Jorma K. Merikoski1 Faculty of Information Technology and Communication Sciences FI-33014 Tampere University Finland [email protected] [email protected] Timo Tossavainen Department of Arts, Communication and Education Lulea University of Technology SE-97187 Lulea Sweden [email protected] Abstract In a previous paper, we proved that the arithmetic subderivative DS is discontinuous at any rational point with respect to the ordinary absolute value. In the present paper, we study this question with respect to the p-adic absolute value. In particular, we show that DS is in this sense continuous at the origin if S is finite or p∈ / S. 1 Introduction Let 0 =6 x ∈ Q. There exists a unique sequence (νp(x))p∈P of integers (with only finitely many nonzero terms) such that 1Corresponding author. 1 x = (sgn x) pνp(x). (1) Y p∈P Here P stands for the set of primes, and sgn x = x/|x|. Define that sgn0 = 0 and νp(0) = ∞ for all p ∈ P. In addition to the ordinary axioms of ∞, we state that 0 · ∞ = 0. Then (1) holds also for x = 0. We recall the basic properties of the p-adic order νp. Proposition 1. For all x,y ∈ Q, (a) νp(x)= ∞ if and only if x = 0; (b) νp(xy)= νp(x)+ νp(y); (c) νp(x + y) ≥ min(νp(x),νp(y)); (d) νp(x + y) = min(νp(x),νp(y)) if νp(x) =6 νp(y).
    [Show full text]
  • A Smooth Variational Principle with Applications to Subdifferentiability and to Differentiability of Convex Functions
    transactions of the american mathematical society Volume 303, Number 2, October 1987 A SMOOTH VARIATIONAL PRINCIPLE WITH APPLICATIONS TO SUBDIFFERENTIABILITY AND TO DIFFERENTIABILITY OF CONVEX FUNCTIONS J. M. BORWEIN AND D. PREISS ABSTRACT. We show that, typically, lower semicontinuous functions on a Ba- nach space densely inherit lower subderivatives of the same degree of smooth- ness as the norm. In particular every continuous convex function on a space with a Gâteaux (weak Hadamard, Fréchet) smooth renorm is densely Gâteaux (weak Hadamard, Fréchet) differentiable. Our technique relies on a more pow- erful analogue of Ekeland's variational principle in which the function is per- turbed by a quadratic-like function. This "smooth" variational principle has very broad applicability in problems of nonsmooth analysis. 1. Introduction. Ekeland's variational principle [11-13] has proved, along with its variants, to be a potent and flexible tool in analysis and in optimization theory [2, 7, 12, 13, 17]. One notable limitation on its application is that even when the original function is differentiable the perturbed function is not. A reasonable smooth variant has long been sought. In §2 we provide such a "smooth" variational principle. The geometric idea behind this proof is as follows. Given a fixed penalty function one cannot, in gen- eral, use penalization techniques to densely obtain minima for lower semicontinuous functions on a given Banach space (but see Theorem 5.2). One can however adap- tively adjust the penalty as one moves around the epigraph of the function, and the final cummulative penalty can be well enough controlled so as to inherit the differentiability properties of the underlying norm on the space.
    [Show full text]