Multiobjective Nonlinear Symmetric Duality Involving Generalized Pseudoconvexity

Total Page:16

File Type:pdf, Size:1020Kb

Multiobjective Nonlinear Symmetric Duality Involving Generalized Pseudoconvexity Applied Mathematics, 2011, 2, 1236-1242 doi:10.4236/am.2011.210172 Published Online October 2011 (http://www.SciRP.org/journal/am) Multiobjective Nonlinear Symmetric Duality Involving Generalized Pseudoconvexity Mohamed Abd El-Hady Kassem Mathematical Department, Faculty of Science, Tanta University, Tanta, Egypt E-mail: [email protected] Received April 8, 2011; revised June 9, 2011; accepted June 16, 2011 Abstract The purpose of this paper is to introduce second order (K, F)-pseudoconvex and second order strongly (K, F)- pseudoconvex functions which are a generalization of cone-pseudoconvex and strongly cone-pseudoconvex functions. A pair of second order symmetric dual multiobjective nonlinear programs is formulated by using the considered functions. Furthermore, the weak, strong and converse duality theorems for this pair are es- tablished. Finally, a self duality theorem is given. Keywords: Multiobjective Programming, Second-Order Symmetric Dual Models, Duality Theorems, Pseudoconvex Functions, Cones 1. Introduction rems were proved for these programs. Yang et al. [6] formulated a pair of Wolf type non-differentiable second Duality is an important concept in the study of nonlinear order symmetric primal and dual problems in mathemati- programming. Symmetric duality in nonlinear program- cal programming. The weak and strong duality theorems ming in which the dual of the dual is the primal was first were established under second order F-convexity assump- introduced by Dorn [1]. Subsequently Dantzig et al. [2] tions. Symmetric minimax mixed integer primal and dual established symmetric duality results for convex/concave problems were also investigated. Khurana [10] intro- functions with nonnegative orthant as the cone. The sym- duced cone-pseudoinvex and strongly cone-pseudoinvex metric duality result was generalized by Bazaraa and functions, and formulated a pair of Mond-Weir type Goode [3] to arbitrary cones. Kim et al. [4] formulated a symmetric dual multiobjective programs over arbitrary pair of multiobjective symmetric dual programs for pseu- cones. The duality theorems and the self-dual theorem doinvex functions and arbitrary cones. The weak, strong, were established under these functions. Yang et al. [8] converse and self duality theorems were established for proved the weak, strong and converse duality theorems that pair of dual models. under F-convexity conditions for a pair of second order The study of second order duality is significant due to symmetric dual programs. Yang et al. [7] established the computational advantage over first order duality as it various duality results for nonlinear programming with provides tighter bounds for the value of the objective cone constraints and its four dual models introduced by function when approximations are used (see Hou and Chandra and Abha [11]. Yang [5], Yang et al. [6,7], Yang et al. [8]). In this paper, we present new definitions dealing with Hou and Yang [5] introduced a pair of second order second order (K, F)-pseudoconvex and second order symmetric dual non-differentiable programs and second strongly (K, F)-pseudoconvex functions which are a gen- order F-pseudoconvex and proved the weak and strong eralized of cone-pseudoconvex and strongly cone-pseu- duality theorems for these second order symmetric dual doconvex functions. We suggest a pair of multiobjective programs under the F-pseudoconvex assumption. Suneja nonlinear second order symmetric dual programs. More- et al. [9] formulated a pair of multiobjective symmetric over, we establish the duality theorems using the above dual programs over arbitrary cones for cone-convex func- generalization of cone-pseudoconvex functions. Finally, tions. The weak, strong, converse and self-duality theo- a self-duality theorem is given by assuming the skew- Copyright © 2011 SciRes. AM M. ABD EL-H. KASSEM 1237 symmetric of the functions. FfufupKxu, u uu int 1 T 2. Notations and Definitions f xfup uu fup int K ; 2 The following conventions for vectors in Rn will be used: and the function f is said to be second-order strongly (K, F)-pseudoconvex at uX if x,,pXRn x yxyiii ,1,2,, n, x≦≦yxyi,1,2,, n, ii Fxu, u fu uu fup int K ≦, x yxyiii 1, 2, , n but x y . A general multiobjective nonlinear programming 1 T f xfup uu fupK . problem can be expressed in the form: 2 f is second-order (K, F)-pseudoconcave if f is second- (P): min f xfxfxfx ,,, 12 m order (K, F)-pseudoconvex and f is second-order strongly (K, F)-pseudoconcave if f is second-order strongly subject to x X x Rn g x≦0, j 1,2, , k j (K, F)-pseudoconvex. Remark 1. If p = 0 and F fu xu, fu where f :and:RRnm gRR nk. xu, where :XX Rnn,,XR the second-order Definition 1. A point x X is said to be an efficient strongly (K, F)-pseudoconvex functions and second-order (or a Pareto optimal) solution of problem (P) if there (K, F)-pseudoconvex functions reduce to strongly K- exists no other x X such that f xfx , pseudoinvex functions and K-pseudoinvex functions de- f xf≦ x, im 1, 2,, but f xfx . ii fined by Khurana [10]. Recall the following three definitions aiming to give Remark 2. Every second-order strongly (K, F)-pseu- the desired definition (i.e., Definition 5). doconvex function is second-order (K, F)-pseudoconvex Definition 2. [5,7,8] A functional but converse is not necessarily true as can be seen from nn F : XXR RXR is sublinear in its third the following example. component if, for all x,,uXX Example 1. Let n 1) Fxua,; a≦ Fxua ,; Fx,;ua aa , R; 12 1 2 12 x and Kxyxy,4≦≦ ,0 x, 2) F xu,; a F xua ,; a Rn , R, ≧0. 2 For notational convenience, we write fx x2 xe,,1x p Fxu, aFxua ,; . and m 3 Let K be a closed convex pointed cone in R with Fxu, AAxu . int K and f : RRn m be a differentiable func- tion. It can be seen that f x is second-order (K, F)-pseu- Definition 3. [4,10,12] The polar cone K * of K is doconvex at u 0 but f x is not second-order strongly defined as (K, F)-pseudoconvex at u 0 because for x 1 * mT F fu fup int K K zRxz≧0. xK xu, u uu and Definition 4. [5] The function f is said to be second- 1 T n f xfup fupK . order F-pseudoconvex at uX if x, pXR , 2 uu ≧ The following example show that a function which is Ffufupxu, u uu 0 second-order strongly (K, F)-pseudoconvex but not sec- 1 T fx fu puu fup. ond-order F-pseudoconvex where K is a closed convex 2 cone. f is second-order F-pseudoconcave if f is second-order Example 2. Let F-pseudoconvex. Kxyyxyxx ,,, ≧≧ ≧0, Now, we are in position to give our definitions of sec- ond-order (K, F)-pseudoconvex functions and second- fx x22 3, xx , p 1 order strongly (K, F)-pseudoconvex functions. and Definition 5. The function f is said to be second-order n 3 (K, F)-pseudoconvex at uX if x,,pXR Fxu, AAxu . Copyright © 2011 SciRes. AM 1238 M. ABD EL-H. KASSEM Then f x is second-order strongly (K, F)-pseudoco- GY: Y Rl R YR l satisfying: nvex at u 0. However, f x is not F-pseudoconvex F aauT CaC, (5) at u 0, because for x 3, xu, 11 Gb byT CbC. (6) Ffufupxu, u uu 0 vy,22 but Furthermore, assume that either 1) f (.,v ) is second-order (K, F)-pseudoconvex at u 1 T and f (.,v ) is second-order (K, G)-pseudoconcave at y; f1 xfup uu fup. 2 or We formulate the following multiobjective nonlinear 2) f (.,v ) is second-order strongly (K, F)-pseudoco- symmetric dual problems: nvex at u and f (.,v) is second-order strongly (K, G)- pseudoconcave at y. 1 TT (SP): minf xy , pyy f xy , p Then 2 1 TT TT* fuv,,ruu fuvr subject toyyy f x,, y f x y p C2 , 2 (1) 1 TT f xy,,pyy f xy pint K 2 TTT yfxyfxypyy,,y ≧0, (2) Proof: Suppose the contrary, i.e., * K , xC1 . 1 TT fuv,,ruu fuvr 2 1 TT (7) (SD): maxf uv , ruu f uv , r 1 2 TT f xy,,intpyy f xy p K 2 subject toTT f u,, v f u v r C* , (3) uuu 1 Since (,x yp , , ) is a feasible solution for the prob- lem (SP) and (,,uv,)r is a feasible solution for the ufuvfuvrTT(,T ,≦0, (4) uu uproblem (SD), we have: * By the dual constraint (3), the vector K , vC, TT * 2 afovfu , ,vr belongs to C , uuu 1 where f : RRnl Rm is a thrice differentiable func- and so by (5) we get from (4) tion of x and y. C and C are closed convex cones- with 1 2 FfuvfuvrTT,, nonempty interiors in Rn and Rl respectively. For exam- xu, u uu n ple, the nonnegative orthant xRx ≧0 is a convex ≧≧ufuvfuvrTT,, T 0. * * uuu cone). C1 and C2 are positive polar cones of C1 and m C2, respectively. K is a closed convex pointed cone in R This gives such that int K and K * is its positive polar cone. TT T T Fxu, ufuv,, uu fuvr int K. (*) y f xy, and yy f xy, are the gradient T and the Hessian matrix of f xy, with respect to y, In a similar fashion, respectively. Similarly, T f xy, and T f xy, are GfxyfxypTT,, ≧0 u uu vy, y yy the gradient and the Hessian matrix of T f xy, with respect to u, respectively.
Recommended publications
  • Boosting Algorithms for Maximizing the Soft Margin
    Boosting Algorithms for Maximizing the Soft Margin Manfred K. Warmuth∗ Karen Glocer Gunnar Ratsch¨ Dept. of Engineering Dept. of Engineering Friedrich Miescher Laboratory University of California University of California Max Planck Society Santa Cruz, CA, U.S.A. Santa Cruz, CA, U.S.A. T¨ubingen, Germany Abstract We present a novel boosting algorithm, called SoftBoost, designed for sets of bi- nary labeled examples that are not necessarily separable by convex combinations of base hypotheses. Our algorithm achieves robustness by capping the distribu- tions on the examples. Our update of the distribution is motivated by minimizing a relative entropy subject to the capping constraints and constraints on the edges of the obtained base hypotheses. The capping constraints imply a soft margin in the dual optimization problem. Our algorithm produces a convex combination of hypotheses whose soft margin is within δ of its maximum. We employ relative en- ln N tropy projection methods to prove an O( δ2 ) iteration bound for our algorithm, where N is number of examples. We compare our algorithm with other approaches including LPBoost, Brown- Boost, and SmoothBoost. We show that there exist cases where the numberof iter- ations required by LPBoost grows linearly in N instead of the logarithmic growth for SoftBoost. In simulation studies we show that our algorithm converges about as fast as LPBoost, faster than BrownBoost, and much faster than SmoothBoost. In a benchmark comparison we illustrate the competitiveness of our approach. 1 Introduction Boosting methods have been used with great success in many applications like OCR, text classifi- cation, natural language processing, drug discovery, and computational biology [13].
    [Show full text]
  • On the Dual Formulation of Boosting Algorithms
    On the Dual Formulation of Boosting Algorithms Chunhua Shen, and Hanxi Li Abstract—We study boosting algorithms from a new perspective. We show that the Lagrange dual problems of ℓ1 norm regularized AdaBoost, LogitBoost and soft-margin LPBoost with generalized hinge loss are all entropy maximization problems. By looking at the dual problems of these boosting algorithms, we show that the success of boosting algorithms can be understood in terms of maintaining a better margin distribution by maximizing margins and at the same time controlling the margin variance. We also theoretically prove that, approximately, ℓ1 norm regularized AdaBoost maximizes the average margin, instead of the minimum margin. The duality formulation also enables us to develop column generation based optimization algorithms, which are totally corrective. We show that they exhibit almost identical classification results to that of standard stage-wise additive boosting algorithms but with much faster convergence rates. Therefore fewer weak classifiers are needed to build the ensemble using our proposed optimization technique. Index Terms—AdaBoost, LogitBoost, LPBoost, Lagrange duality, linear programming, entropy maximization. ✦ 1 INTRODUCTION that the hard-margin LPBoost does not perform well in OOSTING has attracted a lot of research interests since most cases although it usually produces larger minimum B the first practical boosting algorithm, AdaBoost, was margins. More often LPBoost has worse generalization introduced by Freund and Schapire [1]. The machine learn- performance. In other words, a higher minimum margin ing community has spent much effort on understanding would not necessarily imply a lower test error. Breiman [11] how the algorithm works [2], [3], [4].
    [Show full text]
  • Applications of LP Duality
    Lecture 6 Duality of LPs and Applications∗ Last lecture we introduced duality of linear programs. We saw how to form duals, and proved both the weak and strong duality theorems. In this lecture we will see a few more theoretical results and then begin discussion of applications of duality. 6.1 More Duality Results 6.1.1 A Quick Review Last time we saw that if the primal (P) is max c>x s:t: Ax ≤ b then the dual (D) is min b>y s:t: A>y = c y ≥ 0: This is just one form of the primal and dual and we saw that the transformation from one to the other is completely mechanical. The duality theorem tells us that if (P) and (D) are a primal-dual pair then we have one of the three possibilities 1. Both (P) and (D) are infeasible. 2. One is infeasible and the other is unbounded. 3. Both are feasible and if x∗ and y∗ are optimal solutions to (P) and (D) respectively, then c>x∗ = b>y∗. *Lecturer: Anupam Gupta. Scribe: Deepak Bal. 1 LECTURE 6. DUALITY OF LPS AND APPLICATIONS 2 6.1.2 A Comment about Complexity Note that the duality theorem (and equivalently, the Farkas Lemma) puts several problems related to LP feasibility and solvability in NP \ co-NP. E.g., Consider the question of whether the equational form LP Ax = b; x ≥ 0 is feasible. If the program is feasible, we may efficiently verify this by checking that a \certificate” point satisfies the equations. By taking this point to be a vertex and appealing to Hwk1 (Problem 4), we see that we may represent this certificate point in size polynomial in the size of the input.
    [Show full text]
  • Duality Theory and Sensitivity Analysis
    4 Duality Theory and Sensitivity Analysis The notion of duality is one of the most important concepts in linear programming. Basically, associated with each linear programming problem (we may call it the primal problem), defined by the constraint matrix A, the right-hand-side vector b, and the cost vector c, there is a corresponding linear programming problem (called the dual problem) which is constructed by the same set of data A, b, and c. A pair of primal and dual problems are closely related. The interesting relationship between the primal and dual reveals important insights into solving linear programming problems. To begin this chapter, we introduce a dual problem for the standard-form linear programming problem. Then we study the fundamental relationship between the primal and dual problems. Both the "strong" and "weak" duality theorems will be presented. An economic interpretation of the dual variables and dual problem further exploits the concepts in duality theory. These concepts are then used to derive two important simplex algorithms, namely the dual simplex algorithm and the primal dual algorithm, for solving linear programming problems. We conclude this chapter with the sensitivity analysis, which is the study of the effects of changes in the parameters (A, b, and c) of a linear programming problem on its optimal solution. In particular, we study different methods of changing the cost vector, changing the right-hand-side vector, adding and removing a variable, and adding and removing a constraint in linear programming. 55 56 Duality Theory and Sensitivity Analysis Chap. 4 4.1 DUAL LINEAR PROGRAM Consider a linear programming problem in its standard form: Minimize cT x (4.1a) subject to Ax = b (4.1b) x:::O (4.1c) where c and x are n-dimensional column vectors, A an m x n matrix, and b an m­ dimensional column vector.
    [Show full text]
  • On the Ekeland Variational Principle with Applications and Detours
    Lectures on The Ekeland Variational Principle with Applications and Detours By D. G. De Figueiredo Tata Institute of Fundamental Research, Bombay 1989 Author D. G. De Figueiredo Departmento de Mathematica Universidade de Brasilia 70.910 – Brasilia-DF BRAZIL c Tata Institute of Fundamental Research, 1989 ISBN 3-540- 51179-2-Springer-Verlag, Berlin, Heidelberg. New York. Tokyo ISBN 0-387- 51179-2-Springer-Verlag, New York. Heidelberg. Berlin. Tokyo No part of this book may be reproduced in any form by print, microfilm or any other means with- out written permission from the Tata Institute of Fundamental Research, Colaba, Bombay 400 005 Printed by INSDOC Regional Centre, Indian Institute of Science Campus, Bangalore 560012 and published by H. Goetze, Springer-Verlag, Heidelberg, West Germany PRINTED IN INDIA Preface Since its appearance in 1972 the variational principle of Ekeland has found many applications in different fields in Analysis. The best refer- ences for those are by Ekeland himself: his survey article [23] and his book with J.-P. Aubin [2]. Not all material presented here appears in those places. Some are scattered around and there lies my motivation in writing these notes. Since they are intended to students I included a lot of related material. Those are the detours. A chapter on Nemyt- skii mappings may sound strange. However I believe it is useful, since their properties so often used are seldom proved. We always say to the students: go and look in Krasnoselskii or Vainberg! I think some of the proofs presented here are more straightforward. There are two chapters on applications to PDE.
    [Show full text]
  • Lagrangian Duality
    Lagrangian Duality Richard Lusby Department of Management Engineering Technical University of Denmark Today's Topics (jg Lagrange Multipliers Lagrangian Relaxation Lagrangian Duality R Lusby (42111) Lagrangian Duality 2/30 Example: Economic Order Quantity Model (jg Parameters I Demand rate: d I Order cost: K I Unit cost: c I Holding cost: h Decision Variable: The optimal order quantity Q Objective Function: dK hQ minimize + dc + Q 2 Optimal Order Quantity: r 2dK Q∗ = h R Lusby (42111) Lagrangian Duality 3/30 EOQ Model Consider the following extension(jg Suppose we have several items with a space constraint q The problem P dj Kj hQj minimize + dj cj + j Qj 2 P subject to: Qj q j ≤ We have the following possibilities 1 The constraint is non-binding/slack, i.e r X 2dj Kj < q h j 2 The constraint is binding/active R Lusby (42111) Lagrangian Duality 4/30 Lagrange Multiplier (jg The problem P dj Kj hQj minimize T (Q ; Q ;:::; Qn) = + dj cj + 1 2 j Qj 2 P subject to: j Qj = q Lagrange multiplier λ X minimize T (Q ; Q ;:::; Qn) + λ( Qj q) 1 2 − j Differentiate with respect to Qj : @L dj Kj h = 2 + + λ = 0 j @Qj − Qj 2 8 Solving for Qj r 2dj Kj Qj = j h + 2λ 8 R Lusby (42111) Lagrangian Duality 5/30 Lagrange multiplier Continued(jg Substituting this into the constraint we have r X 2dj Kj = q h + 2λ j Solving for λ and Qj gives: P 2 p2dj Kj j h q − λ = λ∗ = 2 r 2dj Kj Q∗ = j j h + 2λ∗ 8 R Lusby (42111) Lagrangian Duality 6/30 Interpretation of λ (jg In linear programming a dual variable is a shadow price: ∗ ∗ @Z yi = @bi Similarly, in the EOQ
    [Show full text]
  • Arxiv:2011.09194V1 [Math.OC]
    Noname manuscript No. (will be inserted by the editor) Lagrangian duality for nonconvex optimization problems with abstract convex functions Ewa M. Bednarczuk · Monika Syga Received: date / Accepted: date Abstract We investigate Lagrangian duality for nonconvex optimization prob- lems. To this aim we use the Φ-convexity theory and minimax theorem for Φ-convex functions. We provide conditions for zero duality gap and strong duality. Among the classes of functions, to which our duality results can be applied, are prox-bounded functions, DC functions, weakly convex functions and paraconvex functions. Keywords Abstract convexity · Minimax theorem · Lagrangian duality · Nonconvex optimization · Zero duality gap · Weak duality · Strong duality · Prox-regular functions · Paraconvex and weakly convex functions 1 Introduction Lagrangian and conjugate dualities have far reaching consequences for solution methods and theory in convex optimization in finite and infinite dimensional spaces. For recent state-of the-art of the topic of convex conjugate duality we refer the reader to the monograph by Radu Bot¸[5]. There exist numerous attempts to construct pairs of dual problems in non- convex optimization e.g., for DC functions [19], [34], for composite functions [8], DC and composite functions [30], [31] and for prox-bounded functions [15]. In the present paper we investigate Lagrange duality for general optimiza- tion problems within the framework of abstract convexity, namely, within the theory of Φ-convexity. The class Φ-convex functions encompasses convex l.s.c. Ewa M. Bednarczuk Systems Research Institute, Polish Academy of Sciences, Newelska 6, 01–447 Warsaw Warsaw University of Technology, Faculty of Mathematics and Information Science, ul.
    [Show full text]
  • The Idea Behind Duality Lec12p1, ORF363/COS323
    Lec12p1, ORF363/COS323 This lecture Instructor: Amir Ali Ahmadi Fall 2014 Linear programming duality + robust linear programming Intuition behind the derivation of the dual TAs: Y. Chen, • G. Hall, • Weak and strong duality theorems J. Ye ○ Max-flow=Min-cut • Primal/dual possibilities • Interpreting the dual • An application: robust linear programming The idea behind duality • For any linear program (LP), there is a closely related LP called the dual. The feasible and optimal solutions of the dual provide very useful information about the original (aka primal) LP. • In particular, if the primal LP is a maximization problem, the dual can be used to find upper bounds on its optimal value. (Similarly, if the primal is a minimization problem, the dual gives lower bounds.) Note that this is useful in certifying optimality of a candidate solution to the primal. One does not get information of this type from feasible solutions to the primal alone. Let's understand this concept through an example first. Consider this LP (from [DPV08]): max. s.t. • Somebody comes to you and claims with objective value 1900 is optimal. • How can we check his claim? Well, one thing we can do is take combinations of the constraints to produce new "valid inequalities" that upper bound the objective function when evaluated on the feasible set. For example, • This means that it is impossible for the objective function to take value larger than 2000 when evaluated at a feasible point (why?). Can we bring down this upper bound further? Lec12 Page 1 Lec12p2, ORF363/COS323 • What if we try This indeed shows that is optimal.
    [Show full text]
  • Norm to Weak Upper Semicontinuity of Pre Duality
    J. Analysis Volume 12 (2005), 1–10 Norm-To-Weak Upper Semi-continuity of the Pre-duality Map S. Dutta and T. S. S. R. K. Rao Abstract. For a Banach space X we study the norm-weak upper semi-continuity of the pre-duality map. We give an example to show that these points in gen- eral do not continue to be points of norm-weak upper semi-continuity for the pre-duality map on the third dual. We give several geometric conditions on X that ensures the existence of these points independent of the predual and also that they continue to be points of norm-weak upper semi-continuity in all the duals of odd order. We also consider these questions in the space of operators. Keywords. Pre-duality map, Unitaries, Semi L-summand, Spaces of opera- tors. 2000 Mathematics Subject Classification. Primary 54C60, 46B20, 47L05. 1. Introduction Motivated by the recent work of Godefroy and Indumathi [4] on the points of norm-weak upper semi-continuity of the pre-duality map of a Banach space X, in this paper we under take a detailed study of these points. Throughout this paper we only work with unit vectors of X∗ that attain their norm (denoted by ∗ NA(X)). Given a unit vector x0 ∈ NA(X) we denote its pre-state space and ∗ x∗ ∗∗ ∗ state space by S ∗ = {x ∈ X : x (x) = 1} and S 0 = {τ ∈ X : τ (x ) = 1} x0 1 0 1 0 where for a Banach space X we denote its closed unit ball by X1.
    [Show full text]
  • Lecture 11: October 8 11.1 Primal and Dual Problems
    10-725/36-725: Convex Optimization Fall 2015 Lecture 11: October 8 Lecturer: Ryan Tibshirani Scribes: Tian Tong Note: LaTeX template courtesy of UC Berkeley EECS dept. Disclaimer: These notes have not been subjected to the usual scrutiny reserved for formal publications. They may be distributed outside this class only with the permission of the Instructor. 11.1 Primal and dual problems 11.1.1 Lagrangian Consider a general optimization problem (called as primal problem) min f(x) (11.1) x subject to hi(x) ≤ 0; i = 1; ··· ; m `j(x) = 0; j = 1; ··· ; r: We define its Lagrangian as m r X X L(x; u; v) = f(x) + uihi(x) + vj`j(x): i=1 j=1 Lagrange multipliers u 2 Rm; v 2 Rr. Lemma 11.1 At each feasible x, f(x) = supu≥0;v L(x; u; v), and the supremum is taken iff u ≥ 0 satisfying uihi(x) = 0; i = 1; ··· ; m. Pm Proof: At each feasible x, we have hi(x) ≤ 0 and `(x) = 0, thus L(x; u; v) = f(x) + i=1 uihi(x) + Pr j=1 vj`j(x) ≤ f(x). The last inequality becomes equality iff uihi(x) = 0; i = 1; ··· ; m. Proposition 11.2 The optimal value of the primal problem, named as f ?, satisfies: f ? = inf sup L(x; u; v): x u≥0;v ? Proof: First considering feasible x (marked as x 2 C), we have f = infx2C f(x) = infx2C supu≥0;v L(x; u; v). Second considering non-feasible x, since supu≥0;v L(x; u; v) = 1 for any x2 = C, infx=2C supu≥0;v L(x; u; v) = ? 1.
    [Show full text]
  • Lagrangian Duality and Convex Optimization
    Lagrangian Duality and Convex Optimization David S. Rosenberg Bloomberg ML EDU October 11, 2017 David S. Rosenberg (Bloomberg ML EDU) October 11, 2017 1 / 30 Introduction David S. Rosenberg (Bloomberg ML EDU) October 11, 2017 2 / 30 Why Convex Optimization? Historically: Linear programs (linear objectives & constraints) were the focus Nonlinear programs: some easy, some hard By early 2000s: Main distinction is between convex and non-convex problems Convex problems are the ones we know how to solve efficiently Mostly batch methods until... around 2010? (earlier if you were into neural nets) By 2010 +- few years, most people understood the optimization / estimation / approximation error tradeoffs accepted that stochatic methods were often faster to get good results (especially on big data sets) These days, nobody’s scared of non-convex problems - SGD seems to work well enough on problems of interest (i.e. neural networks). David S. Rosenberg (Bloomberg ML EDU) October 11, 2017 3 / 30 Your Reference for Convex Optimization Boyd and Vandenberghe (2004) Very clearly written, but has a ton of detail for a first pass. See the Extreme Abridgement of Boyd and Vandenberghe. David S. Rosenberg (Bloomberg ML EDU) October 11, 2017 4 / 30 Notation from Boyd and Vandenberghe f : Rp ! Rq to mean that f maps from some subset of Rp namely dom f ⊂ Rp, where dom f is the domain of f David S. Rosenberg (Bloomberg ML EDU) October 11, 2017 5 / 30 Convex Sets and Functions David S. Rosenberg (Bloomberg ML EDU) October 11, 2017 6 / 30 Convex Sets Definition A set C is convex if for any x1,x2 2 C and any θ with 0 6 θ 6 1 we have θx1 + (1 - θ)x2 2 C.
    [Show full text]
  • On Duality Without Convexity F :S+R; Q: T+Rk
    JOURNAL OF MATHEMATICAL ANALYSIS AND APPLICATIONS 18, 269-275 (1967) On Duality without Convexity J. RISSANEN IBM Sun Jose Research Laboratory, San Jose, California Submitted by Richard Bellman 1. INTRODUCTION Ordinarily, duality in mathematical programming is defined only for convex objective functions with similar inequality constraints, in which case a number of results have been derived relating the solutions of the so-called primal and the dual problems [l-4]. In Ref. [5] (Theorem 2), however, the converse duality theorem is extended for pseudo-convex objective functions with quasi-convex inequality constraints. 1 In this paper, both parts of the duality theorem are proved under somewhat weaker conditions than con- vexity. The weakening of the convexity assumption in the conditions for the direct duality theorem (i.e., when the solution of the primal problem solves the dual) is not particularly significant. More interesting than the results in this part is the proof which, based on somewhat unusual arguments, is very simple. The new conditions are readily seen to be satisfied for convex func- tions. The relaxation of the convexity assumptions for the converse duality theorem appears more significant. The given conditions are considerably weaker than those in Refs. [l-4], but neither weaker nor stronger than those in Ref. [5] (Theorem 2). 2. THE PRIMAL AND THE DUAL PROBLEMS Letf(x) and q(x) denote the mappings: f :S+R; q: T+Rk; S, T open subsets of Rn such that SnTAU#f. 1 An earlier but unsuccessful attempt to extend the duality theorem for what are called “functionally convex” functions was made in Ref.
    [Show full text]