Analysis of a Quadratic Programming Decomposition Algorithm

نویسندگان

  • Guy Bencteux
  • Eric Cancès
  • William W. Hager
  • Claude Le Bris
چکیده

We analyze a decomposition algorithm for minimizing a quadratic objective function, separable in x1 and x2, subject to the constraint that x1 and x2 are orthogonal vectors on the unit sphere. Our algorithm consists of a local step where we minimize the objective function in either variable separately, while enforcing the constraints, followed by a global step where we minimize over a subspace generated by solutions to the local subproblems. We establish a local convergence result when the global minimizers nondegenerate. Our analysis employs necessary and su cient conditions and continuity properties for a global optimum of a quadratic objective function subject to a sphere constraint and a linear constraint. The analysis is connected with a new domain decomposition algorithm for electronic structure calculations. Key-words: quadratic programming, orthogonality constraints, domain decomposition method, electronic structure calculations ∗ August 29, 2007. This material is based upon work supported by the National Science Foundation under Grants 0619080 and 0620286. † [email protected] .edu, http://www.math.u .edu/∼hager, PO Box 118105, Department of Mathematics, University of Florida, Gainesville, FL 32611-8105. Phone (352) 392-0281. Fax (352) 392-8357. ‡ EDF R&D, 1 Avenue du Général de Gaulle, 92141 Clamart Cedex, France and INRIA, MICMAC team project, BP 105, 78153 Rocquencourt, France. § CERMICS, École Nationale des Ponts et Chaussées, 6 & 8, Avenue Blaise Pascal, Cité Descartes, 77455 Marne-La-Vallée Cedex 2, France and INRIA, MICMAC team project, BP 105, 78153 Rocquencourt, France. ¶ CERMICS, École Nationale des Ponts et Chaussées, 6 & 8, Avenue Blaise Pascal, Cité Descartes, 77455 Marne-La-Vallée Cedex 2, France and INRIA, MICMAC team project, BP 105, 78153 Rocquencourt, France. in ria -0 01 69 08 0, v er si on 5 2 Ju n 20 09 Analyse d'un algorithme de décomposition pour un problème de programmation quadratique Résumé : On présente l'analyse numérique d'un algorithme de décomposition pour la minimisation d'une fonction coût quadratique, séparable en x1 et x2, sous la contrainte que x1 et x2 sont orthogonaux sur la sphère unité. Notre algorithme consiste en une étape locale où la fonction coût est minimisée séparément en chacune de ses deux variables, en respectant les contraintes. Cette première étape est suivie d'une étape globale où on minimise la fonction coût sur un sous-espace généré par les solutions de l'étape locale. Un théorème de convergence locale est établi quand les minimiseurs globaux ne sont pas dégénérés. Notre analyse utilise les conditions nécessaires et su santes et les propriétés de continuité du minimum global d'une fonction coût quadratique minimisée sous une double contrainte sphérique et linéaire. Cette analyse est reliée à un nouvel algorithme de décomposition de domaine pour les calculs de structure électronique. Mots-clés : programmation quadratique, contraintes d'orthogonalité, méthode de décomposition de domaine, calculs de structure électronique in ria -0 01 69 08 0, v er si on 5 2 Ju n 20 09 Analysis of a Quadratic Programming Decomposition Algorithm 3

برای دانلود متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید

ثبت نام

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

منابع مشابه

An Integer Programming Model and a Tabu Search Algorithm to Generate α-labeling of Special Classes of Quadratic Graphs

First, an integer programming model is proposed to find an α-labeling for quadratic graphs. Then, a Tabu search algorithm is developed to solve large scale problems. The proposed approach can generate α-labeling for special classes of quadratic graphs, not previously reported in the literature. Then, the main theorem of the paper is presented. We show how a problem in graph theory c...

متن کامل

A TRUST-REGION SEQUENTIAL QUADRATIC PROGRAMMING WITH NEW SIMPLE FILTER AS AN EFFICIENT AND ROBUST FIRST-ORDER RELIABILITY METHOD

The real-world applications addressing the nonlinear functions of multiple variables could be implicitly assessed through structural reliability analysis. This study establishes an efficient algorithm for resolving highly nonlinear structural reliability problems. To this end, first a numerical nonlinear optimization algorithm with a new simple filter is defined to locate and estimate the most ...

متن کامل

An iterative method for tri-level quadratic fractional programming problems using fuzzy goal programming approach

Tri-level optimization problems are optimization problems with three nested hierarchical structures, where in most cases conflicting objectives are set at each level of hierarchy. Such problems are common in management, engineering designs and in decision making situations in general, and are known to be strongly NP-hard. Existing solution methods lack universality in solving these types of pro...

متن کامل

A New Approach for Solving Interval Quadratic Programming Problem

This paper discusses an Interval Quadratic Programming (IQP) problem, where the constraints coefficients and the right-hand sides are represented by interval data. First, the focus is on a common method for solving Interval Linear Programming problem. Then the idea is extended to the IQP problem. Based on this method each IQP problem is reduced to two classical Quadratic Programming (QP) proble...

متن کامل

Benders Decomposition Algorithm for Competitive Supply Chain Network Design under Risk of Disruption and Uncertainty

In this paper, bi-level programming is proposed for designing a competitive supply chain network. A two-stage stochastic programming approach has been developed for a multi-product supply chain comprising a capacitated supplier, several distribution centers, retailers and some resellers in the market. The proposed model considers demand’s uncertainty and disruption in distribution centers and t...

متن کامل

Global convergence of an inexact interior-point method for convex quadratic‎ ‎symmetric cone programming‎

‎In this paper‎, ‎we propose a feasible interior-point method for‎ ‎convex quadratic programming over symmetric cones‎. ‎The proposed algorithm relaxes the‎ ‎accuracy requirements in the solution of the Newton equation system‎, ‎by using an inexact Newton direction‎. ‎Furthermore‎, ‎we obtain an‎ ‎acceptable level of error in the inexact algorithm on convex‎ ‎quadratic symmetric cone programmin...

متن کامل

ذخیره در منابع من


  با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید

برای دانلود متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید

ثبت نام

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

عنوان ژورنال:
  • SIAM J. Numerical Analysis

دوره 47  شماره 

صفحات  -

تاریخ انتشار 2010