Combinatorial Optimization Spring Term 2015 Rico Zenklusen. 2 a = ( 3 2 ) 1 E(a, A) = E(( 3 2 ), ( 4 0

Size: px
Start display at page:

Download "Combinatorial Optimization Spring Term 2015 Rico Zenklusen. 2 a = ( 3 2 ) 1 E(a, A) = E(( 3 2 ), ( 4 0"

Transcription

1 3 2 a = ( 3 2 ) 1 E(a, A) = E(( 3 2 ), ( )) Figure 9: An example of an axis parallel ellipsoid E(a, A) in two dimensions. Notice that the eigenvectors of A correspond to the axes of the ellipsoid, and the square roots of the eigenvalues correspond to the radii of the corresponding axes. Additionally to a separation oracle for P, the ellipsoid algorithm also needs an ellipsoid E 0 that contains P. We recall that an ellipsoid is the image of the unit ball under an affine bijection, which can be defined as follows. Definition 23 (Ellipsoid). An ellipsoid in R n is a set E(a, A) := {x R n (x a) T A 1 (x a) 1}, where a R n and A R n n is a positive definite matrix. The point a is called the center of the ellipsoid E(a, A). In particular, an ellipsoid is always full-dimensional. Notice that the above definition of ellipsoid is indeed an affine bijection of the unit ball. This can be seen as follows. A matrix A R n n is positive definite if and only if there is a full rank matrix Q R n n such that A = QQ T. Hence, A 1 = (Q T ) 1 (Q 1 ) = (Q 1 ) T Q 1, and therefore E(a, A) = {x R n Q 1 (x a) 2 1} = {y + a y R n, Q 1 y 2 1} (Substitution with y = x a.) = {Qz + a z R n, z 2 1}. (Substitution with z = Q 1 y.) Figure 9 shows an example of an ellipsoid for n = 2. The separation oracle for P and the ellipsoid E 0 P is all that is needed to run the ellipsoid algorithm and obtain a point x P. However, to make sure that the ellipsoid algorithm runs in polynomial time we need a further( condition. ) One sufficient condition for the ellipsoid algorithm to run in polynomial time is that log vol(e0 ) is polynomially bounded in the input. We will see later that this holds for most cases we are interested in. Description of ellipsoid method The ellipsoid method is described in Algorithm 1. Algorithm 1: Ellipsoid method input : Separation oracle for a polytope P R n with dim(p ) = n, and an ellipsoid E 0 = E(a 0, A 0 ) with P E 0. output: A point y P. 1 i = 0; 2 while a i P (checked with separation oracle) do 3 Get non-zero c R n such that P {x R n c T x c T a i }, using sep. oracle; 4 Find min. volume ellipsoid E i+1 = E(a i+1, A i+1 ) containing E i {x R n c T x c T a i }; 5 i = i + 1; 6 return a i ;

2 P E(a i+1, A i+1 ) a i+1 {x R n c T x = c T a i } a i E(a i, A i ) c Figure 10: Illustration of a single iteration of the ellipsoid method. The polytope P is inside each ellipsoid considered by the ellipsoid method. We will soon give some more details on how to compute E i+1. It turns out that there is relatively simple way to describe E i+1 in terms of E i and c. Unfortunately, as we will see soon, it turns out that this description of E i+1 involves taking a square root, which is an operation that we cannot perform up to arbitrary precision under the typical computational assumptions. Hence, in practice, one only computes an ellipsoid E i+1 that approximates E i+1 and has a polynomial-size description. To simplify the exposition we will not go into these numerical details and assume that we use an exact description of E i+1. Figure 10 illustrates one iteration of the ellipsoid method. Notice that we have P E i for each ellipsoid considered in the ellipsoid method. This can easily be verified by induction. The first ellipsoid E 0 contains P by assumption. Furthermore, for each iteration i, P {x R n c T x c T a i }, and hence P E i {x R n c T x c T a i }, since P E i. We thus obtain P E i+1 because E i {x R n c T x c T a i } E i+1 by definition of E i+1. Getting a bound on the number of iterations The key property of the constructed ellipsoid, besides the fact that they contain P, is that they shrink in terms of volume. More precisely, we have the following. Lemma 24. vol(e i+1 ) vol(e i ) < e 1 2(n+1). Before proving Lemma 24, we observe that it immediately implies an upper bound on the number of iterations that the ellipsoid algorithm performs. ( ) Lemma 25. The ellipsoid method will stop after at most 2(n + 1) ln vol(e0 ) iterations. Proof. Let L Z 0 be the last iteration of the ellipsoid algorithm. Since E L contains P, we must have vol(e L ) which, combined with Lemma 24, leads to vol(e L ) vol(e 0 )e L 2(n+1), and thus L 2(n + 1) ln ( ) vol(e0 ). The rest of this section is devoted to proving Lemma 24. Later, we will show how the bound on the number of iterations given by Lemma 25 can be used to prove that the ellipsoid method runs in polynomial time for any {0, 1}-polytope. For this we will first make the link between checking feasibility of a polytope and optimizing an LP more explicit.

3 Proof of Lemma 24 and explicit description for E i+1 A key simplification for proving Lemma 24 is to observe that it suffices to consider the special case defined by the following properties. Assumption 26. E i is the unit ball, i.e., E i = E(0, I), where 0 R n is the zero vector, and I R n n is the n n identity matrix. The halfspace H i = {x R n c T x c T a i } is equal to H = {x = (x 1,..., x n ) R n x 1 0}. More precisely, a description of a minimum volume ellipsoid under this assumption can be transformed to a description of a minimum volume ellipsoid containing E i H i. Proof that Assumption 26 is without loss of generality. We first sketch the proof plan. We will define an affine bijection ρ : R n R n such that ρ(e(0, I)) = E i and ρ( H) = H i. Furthermore, since ρ is affine, applying it to any measurable set will scale its volume by a factor that only depends on ρ. Using this, we can derive that if Ē is a minimum volume ellipsoid containing E(0, I) H, then ρ(ē) is a minimum volume ellipsoid containing E i H i. We start by defining a bijective affine transformation φ that maps E(0, I) to E i. Let E i = (a i, A i ) and let Q i R n n be such that A i = Q i Q T i. Define As we have already observed previously, φ(x) = Q i x + a i x R n. φ(e(0, I)) = E i. If furthermore we had φ 1 (H i ) = H, then we could chose ρ = φ. However, this is not true in general. Still, a small change to φ (composing φ with an orthonormal matrix) will fix this. To see how we have to change φ to obtain ρ, we start by considering φ 1 (H i ). φ 1 (H i ) = {φ 1 (x) x R n, c T x c T a i } = {y R n c T (Q i y + a i ) c T a i } (y = φ 1 (x) = Q 1 i (x a i )) = {y R n c T Q i y 0}. Notice that c T Q i 0 since c 0 and Q i is full rank. Hence, Q T i c 2 > 0. Moreover, the hyperplane H i does not change when scaling c. Hence, we can assume Q T i c 2 = 1. Ideally, we would have liked to have Q T i c = e 1, where e 1 = (1, 0,..., 0) {0, 1} n is the vector with a single one in the first coordinate and zeros everywhere else. This would have implied φ 1 (H i ) = H. To achieve this we slightly modify φ to get ρ. Since Q T i c 2 = 1, there is a orthonormal matrix R R n n such that R T Q T i c = e 1. (19) We define We now obtain ρ(x) = φ(rx) = Q i Rx + a i x R n. ρ 1 (H i ) = {ρ 1 (x) x R n, c T x c T a i } = {y R n c T (Q i Ry + a i ) c T a i } (y = ρ 1 (x)) = {y R n c T Q i Ry 0} = {y R n y 1 0} (by (19)) = {y R n y 1 0} = H. Thus, ρ is indeed a bijective affine transformation with ρ( H) = H i and ρ(e(0, I)) = E i. Let Ē be a minimum volume ellipsoid that contains E(0, I) H. We will show that ρ(ē) is a minimum volume ellipsoid containing E i H i, which will finish the proof. Notice that all our steps

4 were constructive. In particular one can find efficiently a matrix R as claimed (though, we did not give details for this here). Furthermore, a matrix Q i such that A i = Q i Q T i can be found through a Cholesky decomposition. Since Ē E(0, I) H and ρ is a bijection, we have ρ(ē) ρ(e(0, I)) ρ( H) = E i H i. It remains to show that ρ(ē) has minimum volume among all ellipsoids that contain E i H i. Let Ẽ be an arbitrary ellipsoid in R n such that Ẽ E i H i. We will finish the proof by showing vol(ẽ) vol(ρ(ē)). We recall a well-known property of affine functions x Ax+v, namely that they scale any volume by the same factor det A. For ρ this implies that for any measurable set U R n we have vol(ρ(u)) = vol(u) det(qr) = vol(u) det Q det R = vol(u) det Q. (20) Notice that ρ 1 (Ẽ) is an ellipsoid containing ρ 1 (E i ) ρ 1 (H i ) = E(0, I) H, and hence We thus obtain 1 vol(ρ (Ẽ)) vol(ē). (21) vol(ẽ) = vol(ρ 1 (Ẽ)) det Q (by 20) vol(ē) det Q (by 21) = vol(ρ(ē)) (by 20). Hence, we can accept Assumption 26 without loss of generality. We claim that under Assumption 26, the ellipsoid E i+1 is given by ( ) E i+1 = x Rn n ( x 1 1 ) 2 + n2 1 n n + 1 n 2 x 2 j 1. (22) We will not show formally that the above description is a minimum ellipsoid containing E(0, I) {x R n x 1 0}. However, we will show that E i+1 as described above contains E(0, I) {x R n x 1 0}, and that the ratio of its volume and the volume of E i satisfy the inequality of Lemma 24. Of course, even without accepting that E i+1 is indeed the smallest ellipsoid containing E(0, I) {x R n x 1 0}, this shows Lemma 24. Furthermore, also in the analysis that follows, we never need to prove that E i+1 is the smallest ellipsoid containing E(0, I) {x R n x 1 0}, since we can simply assume that we work with the description of E i+1 given by (22) in the ellipsoid algorithm, without assuming that it has minimum volume. We will later generalize the description of E i+1 given in (22) to the general case when E i is not necessarily the unit ball, and {x R x 1 0} is replaced by a general halfspace going through the center of E i. Lemma 27. Under Assumption 26, we have that E i+1 as described by (22) satisfies E i+1 E i {x R n x 1 0}.

5 Proof. Let x E i {x R n x 1 0}. We have ( n + 1 n ) 2 ( x 1 1 ) 2 + n2 1 n + 1 n 2 = n2 + 2n + 1 n 2 x 2 1 ( n + 1 n x 2 j = 2n + 2 n 2 x 2 1 2n + 2 n 2 x n 2 + n2 1 n 2 ) 2 2x 1 n n 2 + n2 1 n 2 j=1 x 2 j }{{} 1 x 2 i (x E i = E(0, I)) = 2n + 2 n 2 x 1 (x 1 1) +1 (0 x }{{} 1 1) 1, and thus x E i+1. 0 We are now ready to prove Lemma 24.

CS675: Convex and Combinatorial Optimization Spring 2018 The Ellipsoid Algorithm. Instructor: Shaddin Dughmi

CS675: Convex and Combinatorial Optimization Spring 2018 The Ellipsoid Algorithm. Instructor: Shaddin Dughmi CS675: Convex and Combinatorial Optimization Spring 2018 The Ellipsoid Algorithm Instructor: Shaddin Dughmi History and Basics Originally developed in the mid 70s by Iudin, Nemirovski, and Shor for use

More information

7. Lecture notes on the ellipsoid algorithm

7. Lecture notes on the ellipsoid algorithm Massachusetts Institute of Technology Michel X. Goemans 18.433: Combinatorial Optimization 7. Lecture notes on the ellipsoid algorithm The simplex algorithm was the first algorithm proposed for linear

More information

Lecture notes on the ellipsoid algorithm

Lecture notes on the ellipsoid algorithm Massachusetts Institute of Technology Handout 1 18.433: Combinatorial Optimization May 14th, 007 Michel X. Goemans Lecture notes on the ellipsoid algorithm The simplex algorithm was the first algorithm

More information

Topics in Theoretical Computer Science: An Algorithmist's Toolkit Fall 2007

Topics in Theoretical Computer Science: An Algorithmist's Toolkit Fall 2007 MIT OpenCourseWare http://ocw.mit.edu 18.409 Topics in Theoretical Computer Science: An Algorithmist's Toolkit Fall 2007 For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.

More information

The Ellipsoid Algorithm

The Ellipsoid Algorithm The Ellipsoid Algorithm John E. Mitchell Department of Mathematical Sciences RPI, Troy, NY 12180 USA 9 February 2018 Mitchell The Ellipsoid Algorithm 1 / 28 Introduction Outline 1 Introduction 2 Assumptions

More information

6.854J / J Advanced Algorithms Fall 2008

6.854J / J Advanced Algorithms Fall 2008 MIT OpenCourseWare http://ocw.mit.edu 6.854J / 18.415J Advanced Algorithms Fall 2008 For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms. 18.415/6.854 Advanced

More information

Integer Programming ISE 418. Lecture 12. Dr. Ted Ralphs

Integer Programming ISE 418. Lecture 12. Dr. Ted Ralphs Integer Programming ISE 418 Lecture 12 Dr. Ted Ralphs ISE 418 Lecture 12 1 Reading for This Lecture Nemhauser and Wolsey Sections II.2.1 Wolsey Chapter 9 ISE 418 Lecture 12 2 Generating Stronger Valid

More information

Week Quadratic forms. Principal axes theorem. Text reference: this material corresponds to parts of sections 5.5, 8.2,

Week Quadratic forms. Principal axes theorem. Text reference: this material corresponds to parts of sections 5.5, 8.2, Math 051 W008 Margo Kondratieva Week 10-11 Quadratic forms Principal axes theorem Text reference: this material corresponds to parts of sections 55, 8, 83 89 Section 41 Motivation and introduction Consider

More information

A strongly polynomial algorithm for linear systems having a binary solution

A strongly polynomial algorithm for linear systems having a binary solution A strongly polynomial algorithm for linear systems having a binary solution Sergei Chubanov Institute of Information Systems at the University of Siegen, Germany e-mail: sergei.chubanov@uni-siegen.de 7th

More information

LMI MODELLING 4. CONVEX LMI MODELLING. Didier HENRION. LAAS-CNRS Toulouse, FR Czech Tech Univ Prague, CZ. Universidad de Valladolid, SP March 2009

LMI MODELLING 4. CONVEX LMI MODELLING. Didier HENRION. LAAS-CNRS Toulouse, FR Czech Tech Univ Prague, CZ. Universidad de Valladolid, SP March 2009 LMI MODELLING 4. CONVEX LMI MODELLING Didier HENRION LAAS-CNRS Toulouse, FR Czech Tech Univ Prague, CZ Universidad de Valladolid, SP March 2009 Minors A minor of a matrix F is the determinant of a submatrix

More information

Algorithmic Convex Geometry

Algorithmic Convex Geometry Algorithmic Convex Geometry August 2011 2 Contents 1 Overview 5 1.1 Learning by random sampling.................. 5 2 The Brunn-Minkowski Inequality 7 2.1 The inequality.......................... 8 2.1.1

More information

Week 8. 1 LP is easy: the Ellipsoid Method

Week 8. 1 LP is easy: the Ellipsoid Method Week 8 1 LP is easy: the Ellipsoid Method In 1979 Khachyan proved that LP is solvable in polynomial time by a method of shrinking ellipsoids. The running time is polynomial in the number of variables n,

More information

Massachusetts Institute of Technology 6.854J/18.415J: Advanced Algorithms Friday, March 18, 2016 Ankur Moitra. Problem Set 6

Massachusetts Institute of Technology 6.854J/18.415J: Advanced Algorithms Friday, March 18, 2016 Ankur Moitra. Problem Set 6 Massachusetts Institute of Technology 6.854J/18.415J: Advanced Algorithms Friday, March 18, 2016 Ankur Moitra Problem Set 6 Due: Wednesday, April 6, 2016 7 pm Dropbox Outside Stata G5 Collaboration policy:

More information

Semidefinite and Second Order Cone Programming Seminar Fall 2001 Lecture 5

Semidefinite and Second Order Cone Programming Seminar Fall 2001 Lecture 5 Semidefinite and Second Order Cone Programming Seminar Fall 2001 Lecture 5 Instructor: Farid Alizadeh Scribe: Anton Riabov 10/08/2001 1 Overview We continue studying the maximum eigenvalue SDP, and generalize

More information

The Ellipsoid (Kachiyan) Method

The Ellipsoid (Kachiyan) Method Yinyu Ye, MS&E, Stanford MS&E310 Lecture Note: Ellipsoid Method 1 The Ellipsoid (Kachiyan) Method Yinyu Ye Department of Management Science and Engineering Stanford University Stanford, CA 94305, U.S.A.

More information

NATIONAL UNIVERSITY OF SINGAPORE Department of Mathematics MA4247 Complex Analysis II Lecture Notes Part II

NATIONAL UNIVERSITY OF SINGAPORE Department of Mathematics MA4247 Complex Analysis II Lecture Notes Part II NATIONAL UNIVERSITY OF SINGAPORE Department of Mathematics MA4247 Complex Analysis II Lecture Notes Part II Chapter 2 Further properties of analytic functions 21 Local/Global behavior of analytic functions;

More information

Two Lectures on the Ellipsoid Method for Solving Linear Programs

Two Lectures on the Ellipsoid Method for Solving Linear Programs Two Lectures on the Ellipsoid Method for Solving Linear Programs Lecture : A polynomial-time algorithm for LP Consider the general linear programming problem: δ = max cx S.T. Ax b, x R n, () where A =

More information

Lecture #21. c T x Ax b. maximize subject to

Lecture #21. c T x Ax b. maximize subject to COMPSCI 330: Design and Analysis of Algorithms 11/11/2014 Lecture #21 Lecturer: Debmalya Panigrahi Scribe: Samuel Haney 1 Overview In this lecture, we discuss linear programming. We first show that the

More information

Throughout these notes we assume V, W are finite dimensional inner product spaces over C.

Throughout these notes we assume V, W are finite dimensional inner product spaces over C. Math 342 - Linear Algebra II Notes Throughout these notes we assume V, W are finite dimensional inner product spaces over C 1 Upper Triangular Representation Proposition: Let T L(V ) There exists an orthonormal

More information

THE INVERSE FUNCTION THEOREM

THE INVERSE FUNCTION THEOREM THE INVERSE FUNCTION THEOREM W. PATRICK HOOPER The implicit function theorem is the following result: Theorem 1. Let f be a C 1 function from a neighborhood of a point a R n into R n. Suppose A = Df(a)

More information

Ellipsoidal Mixed-Integer Representability

Ellipsoidal Mixed-Integer Representability Ellipsoidal Mixed-Integer Representability Alberto Del Pia Jeff Poskin September 15, 2017 Abstract Representability results for mixed-integer linear systems play a fundamental role in optimization since

More information

LINEAR PROGRAMMING III

LINEAR PROGRAMMING III LINEAR PROGRAMMING III ellipsoid algorithm combinatorial optimization matrix games open problems Lecture slides by Kevin Wayne Last updated on 7/25/17 11:09 AM LINEAR PROGRAMMING III ellipsoid algorithm

More information

Ma/CS 6b Class 23: Eigenvalues in Regular Graphs

Ma/CS 6b Class 23: Eigenvalues in Regular Graphs Ma/CS 6b Class 3: Eigenvalues in Regular Graphs By Adam Sheffer Recall: The Spectrum of a Graph Consider a graph G = V, E and let A be the adjacency matrix of G. The eigenvalues of G are the eigenvalues

More information

(x, y) = d(x, y) = x y.

(x, y) = d(x, y) = x y. 1 Euclidean geometry 1.1 Euclidean space Our story begins with a geometry which will be familiar to all readers, namely the geometry of Euclidean space. In this first chapter we study the Euclidean distance

More information

Polynomiality of Linear Programming

Polynomiality of Linear Programming Chapter 10 Polynomiality of Linear Programming In the previous section, we presented the Simplex Method. This method turns out to be very efficient for solving linear programmes in practice. While it is

More information

Discrete Geometry. Problem 1. Austin Mohr. April 26, 2012

Discrete Geometry. Problem 1. Austin Mohr. April 26, 2012 Discrete Geometry Austin Mohr April 26, 2012 Problem 1 Theorem 1 (Linear Programming Duality). Suppose x, y, b, c R n and A R n n, Ax b, x 0, A T y c, and y 0. If x maximizes c T x and y minimizes b T

More information

Complexity of linear programming: outline

Complexity of linear programming: outline Complexity of linear programming: outline I Assessing computational e ciency of algorithms I Computational e ciency of the Simplex method I Ellipsoid algorithm for LP and its computational e ciency IOE

More information

In English, this means that if we travel on a straight line between any two points in C, then we never leave C.

In English, this means that if we travel on a straight line between any two points in C, then we never leave C. Convex sets In this section, we will be introduced to some of the mathematical fundamentals of convex sets. In order to motivate some of the definitions, we will look at the closest point problem from

More information

Grothendieck s Inequality

Grothendieck s Inequality Grothendieck s Inequality Leqi Zhu 1 Introduction Let A = (A ij ) R m n be an m n matrix. Then A defines a linear operator between normed spaces (R m, p ) and (R n, q ), for 1 p, q. The (p q)-norm of A

More information

EE/ACM Applications of Convex Optimization in Signal Processing and Communications Lecture 17

EE/ACM Applications of Convex Optimization in Signal Processing and Communications Lecture 17 EE/ACM 150 - Applications of Convex Optimization in Signal Processing and Communications Lecture 17 Andre Tkacenko Signal Processing Research Group Jet Propulsion Laboratory May 29, 2012 Andre Tkacenko

More information

VISCOSITY SOLUTIONS. We follow Han and Lin, Elliptic Partial Differential Equations, 5.

VISCOSITY SOLUTIONS. We follow Han and Lin, Elliptic Partial Differential Equations, 5. VISCOSITY SOLUTIONS PETER HINTZ We follow Han and Lin, Elliptic Partial Differential Equations, 5. 1. Motivation Throughout, we will assume that Ω R n is a bounded and connected domain and that a ij C(Ω)

More information

A LOCALIZATION PROPERTY AT THE BOUNDARY FOR MONGE-AMPERE EQUATION

A LOCALIZATION PROPERTY AT THE BOUNDARY FOR MONGE-AMPERE EQUATION A LOCALIZATION PROPERTY AT THE BOUNDARY FOR MONGE-AMPERE EQUATION O. SAVIN. Introduction In this paper we study the geometry of the sections for solutions to the Monge- Ampere equation det D 2 u = f, u

More information

Summer School: Semidefinite Optimization

Summer School: Semidefinite Optimization Summer School: Semidefinite Optimization Christine Bachoc Université Bordeaux I, IMB Research Training Group Experimental and Constructive Algebra Haus Karrenberg, Sept. 3 - Sept. 7, 2012 Duality Theory

More information

Lecture 15: October 15

Lecture 15: October 15 10-725: Optimization Fall 2012 Lecturer: Barnabas Poczos Lecture 15: October 15 Scribes: Christian Kroer, Fanyi Xiao Note: LaTeX template courtesy of UC Berkeley EECS dept. Disclaimer: These notes have

More information

CS 6820 Fall 2014 Lectures, October 3-20, 2014

CS 6820 Fall 2014 Lectures, October 3-20, 2014 Analysis of Algorithms Linear Programming Notes CS 6820 Fall 2014 Lectures, October 3-20, 2014 1 Linear programming The linear programming (LP) problem is the following optimization problem. We are given

More information

CSC Linear Programming and Combinatorial Optimization Lecture 8: Ellipsoid Algorithm

CSC Linear Programming and Combinatorial Optimization Lecture 8: Ellipsoid Algorithm CSC2411 - Linear Programming and Combinatorial Optimization Lecture 8: Ellipsoid Algorithm Notes taken by Shizhong Li March 15, 2005 Summary: In the spring of 1979, the Soviet mathematician L.G.Khachian

More information

15.081J/6.251J Introduction to Mathematical Programming. Lecture 18: The Ellipsoid method

15.081J/6.251J Introduction to Mathematical Programming. Lecture 18: The Ellipsoid method 15.081J/6.251J Introduction to Mathematical Programming Lecture 18: The Ellipsoid method 1 Outline Efficient algorithms and computational complexity Slide 1 The key geometric result behind the ellipsoid

More information

Principal Component Analysis

Principal Component Analysis Machine Learning Michaelmas 2017 James Worrell Principal Component Analysis 1 Introduction 1.1 Goals of PCA Principal components analysis (PCA) is a dimensionality reduction technique that can be used

More information

Functional Analysis Exercise Class

Functional Analysis Exercise Class Functional Analysis Exercise Class Week 9 November 13 November Deadline to hand in the homeworks: your exercise class on week 16 November 20 November Exercises (1) Show that if T B(X, Y ) and S B(Y, Z)

More information

15.083J/6.859J Integer Optimization. Lecture 10: Solving Relaxations

15.083J/6.859J Integer Optimization. Lecture 10: Solving Relaxations 15.083J/6.859J Integer Optimization Lecture 10: Solving Relaxations 1 Outline The key geometric result behind the ellipsoid method Slide 1 The ellipsoid method for the feasibility problem The ellipsoid

More information

arxiv: v3 [math.oc] 24 May 2016

arxiv: v3 [math.oc] 24 May 2016 Mathematical Programming manuscript No. (will be inserted by the editor) How to Convexify the Intersection of a Second Order Cone and a Nonconvex Quadratic Samuel Burer Fatma Kılınç-Karzan arxiv:1406.1031v3

More information

Design and Analysis of Algorithms Lecture Notes on Convex Optimization CS 6820, Fall Nov 2 Dec 2016

Design and Analysis of Algorithms Lecture Notes on Convex Optimization CS 6820, Fall Nov 2 Dec 2016 Design and Analysis of Algorithms Lecture Notes on Convex Optimization CS 6820, Fall 206 2 Nov 2 Dec 206 Let D be a convex subset of R n. A function f : D R is convex if it satisfies f(tx + ( t)y) tf(x)

More information

USE OF THE INTERIOR-POINT METHOD FOR CORRECTING AND SOLVING INCONSISTENT LINEAR INEQUALITY SYSTEMS

USE OF THE INTERIOR-POINT METHOD FOR CORRECTING AND SOLVING INCONSISTENT LINEAR INEQUALITY SYSTEMS An. Şt. Univ. Ovidius Constanţa Vol. 9(2), 2001, 121 128 USE OF THE INTERIOR-POINT METHOD FOR CORRECTING AND SOLVING INCONSISTENT LINEAR INEQUALITY SYSTEMS Elena Popescu Dedicated to Professor Mirela Ştefănescu

More information

3. Linear Programming and Polyhedral Combinatorics

3. Linear Programming and Polyhedral Combinatorics Massachusetts Institute of Technology 18.433: Combinatorial Optimization Michel X. Goemans February 28th, 2013 3. Linear Programming and Polyhedral Combinatorics Summary of what was seen in the introductory

More information

Input: System of inequalities or equalities over the reals R. Output: Value for variables that minimizes cost function

Input: System of inequalities or equalities over the reals R. Output: Value for variables that minimizes cost function Linear programming Input: System of inequalities or equalities over the reals R A linear cost function Output: Value for variables that minimizes cost function Example: Minimize 6x+4y Subject to 3x + 2y

More information

A Review of Linear Programming

A Review of Linear Programming A Review of Linear Programming Instructor: Farid Alizadeh IEOR 4600y Spring 2001 February 14, 2001 1 Overview In this note we review the basic properties of linear programming including the primal simplex

More information

COURSE ON LMI PART I.2 GEOMETRY OF LMI SETS. Didier HENRION henrion

COURSE ON LMI PART I.2 GEOMETRY OF LMI SETS. Didier HENRION   henrion COURSE ON LMI PART I.2 GEOMETRY OF LMI SETS Didier HENRION www.laas.fr/ henrion October 2006 Geometry of LMI sets Given symmetric matrices F i we want to characterize the shape in R n of the LMI set F

More information

5.6 Penalty method and augmented Lagrangian method

5.6 Penalty method and augmented Lagrangian method 5.6 Penalty method and augmented Lagrangian method Consider a generic NLP problem min f (x) s.t. c i (x) 0 i I c i (x) = 0 i E (1) x R n where f and the c i s are of class C 1 or C 2, and I and E are the

More information

A Topological Representation Theorem for Oriented Matroids

A Topological Representation Theorem for Oriented Matroids Smith ScholarWorks Computer Science: Faculty Publications Computer Science 4-2005 A Topological Representation Theorem for Oriented Matroids Jürgen Bokowski Technische Universitat Darmstadt Simon King

More information

8. Geometric problems

8. Geometric problems 8. Geometric problems Convex Optimization Boyd & Vandenberghe extremal volume ellipsoids centering classification placement and facility location 8 1 Minimum volume ellipsoid around a set Löwner-John ellipsoid

More information

Theory and Internet Protocols

Theory and Internet Protocols Game Lecture 2: Linear Programming and Zero Sum Nash Equilibrium Xiaotie Deng AIMS Lab Department of Computer Science Shanghai Jiaotong University September 26, 2016 1 2 3 4 Standard Form (P) Outline

More information

c 2000 Society for Industrial and Applied Mathematics

c 2000 Society for Industrial and Applied Mathematics SIAM J. OPIM. Vol. 10, No. 3, pp. 750 778 c 2000 Society for Industrial and Applied Mathematics CONES OF MARICES AND SUCCESSIVE CONVEX RELAXAIONS OF NONCONVEX SES MASAKAZU KOJIMA AND LEVEN UNÇEL Abstract.

More information

1 Overview. 2 Extreme Points. AM 221: Advanced Optimization Spring 2016

1 Overview. 2 Extreme Points. AM 221: Advanced Optimization Spring 2016 AM 22: Advanced Optimization Spring 206 Prof. Yaron Singer Lecture 7 February 7th Overview In the previous lectures we saw applications of duality to game theory and later to learning theory. In this lecture

More information

Topological properties of Z p and Q p and Euclidean models

Topological properties of Z p and Q p and Euclidean models Topological properties of Z p and Q p and Euclidean models Samuel Trautwein, Esther Röder, Giorgio Barozzi November 3, 20 Topology of Q p vs Topology of R Both R and Q p are normed fields and complete

More information

Week 4. (1) 0 f ij u ij.

Week 4. (1) 0 f ij u ij. Week 4 1 Network Flow Chapter 7 of the book is about optimisation problems on networks. Section 7.1 gives a quick introduction to the definitions of graph theory. In fact I hope these are already known

More information

U.C. Berkeley CS294: Spectral Methods and Expanders Handout 11 Luca Trevisan February 29, 2016

U.C. Berkeley CS294: Spectral Methods and Expanders Handout 11 Luca Trevisan February 29, 2016 U.C. Berkeley CS294: Spectral Methods and Expanders Handout Luca Trevisan February 29, 206 Lecture : ARV In which we introduce semi-definite programming and a semi-definite programming relaxation of sparsest

More information

47-831: Advanced Integer Programming Lecturer: Amitabh Basu Lecture 2 Date: 03/18/2010

47-831: Advanced Integer Programming Lecturer: Amitabh Basu Lecture 2 Date: 03/18/2010 47-831: Advanced Integer Programming Lecturer: Amitabh Basu Lecture Date: 03/18/010 We saw in the previous lecture that a lattice Λ can have many bases. In fact, if Λ is a lattice of a subspace L with

More information

10. Ellipsoid method

10. Ellipsoid method 10. Ellipsoid method EE236C (Spring 2008-09) ellipsoid method convergence proof inequality constraints 10 1 Ellipsoid method history developed by Shor, Nemirovski, Yudin in 1970s used in 1979 by Khachian

More information

Linear Programming. Chapter Introduction

Linear Programming. Chapter Introduction Chapter 3 Linear Programming Linear programs (LP) play an important role in the theory and practice of optimization problems. Many COPs can directly be formulated as LPs. Furthermore, LPs are invaluable

More information

On John type ellipsoids

On John type ellipsoids On John type ellipsoids B. Klartag Tel Aviv University Abstract Given an arbitrary convex symmetric body K R n, we construct a natural and non-trivial continuous map u K which associates ellipsoids to

More information

To hand in: (a) Prove that a group G is abelian (= commutative) if and only if (xy) 2 = x 2 y 2 for all x, y G.

To hand in: (a) Prove that a group G is abelian (= commutative) if and only if (xy) 2 = x 2 y 2 for all x, y G. Homework #6. Due Thursday, October 14th Reading: For this homework assignment: Sections 3.3 and 3.4 (up to page 167) Before the class next Thursday: Sections 3.5 and 3.4 (pp. 168-171). Also review the

More information

Symmetric Matrices and Eigendecomposition

Symmetric Matrices and Eigendecomposition Symmetric Matrices and Eigendecomposition Robert M. Freund January, 2014 c 2014 Massachusetts Institute of Technology. All rights reserved. 1 2 1 Symmetric Matrices and Convexity of Quadratic Functions

More information

Some Results Concerning Uniqueness of Triangle Sequences

Some Results Concerning Uniqueness of Triangle Sequences Some Results Concerning Uniqueness of Triangle Sequences T. Cheslack-Postava A. Diesl M. Lepinski A. Schuyler August 12 1999 Abstract In this paper we will begin by reviewing the triangle iteration. We

More information

In this section again we shall assume that the matrix A is m m, real and symmetric.

In this section again we shall assume that the matrix A is m m, real and symmetric. 84 3. The QR algorithm without shifts See Chapter 28 of the textbook In this section again we shall assume that the matrix A is m m, real and symmetric. 3.1. Simultaneous Iterations algorithm Suppose we

More information

Largest dual ellipsoids inscribed in dual cones

Largest dual ellipsoids inscribed in dual cones Largest dual ellipsoids inscribed in dual cones M. J. Todd June 23, 2005 Abstract Suppose x and s lie in the interiors of a cone K and its dual K respectively. We seek dual ellipsoidal norms such that

More information

THEOREM OF OSELEDETS. We recall some basic facts and terminology relative to linear cocycles and the multiplicative ergodic theorem of Oseledets [1].

THEOREM OF OSELEDETS. We recall some basic facts and terminology relative to linear cocycles and the multiplicative ergodic theorem of Oseledets [1]. THEOREM OF OSELEDETS We recall some basic facts and terminology relative to linear cocycles and the multiplicative ergodic theorem of Oseledets []. 0.. Cocycles over maps. Let µ be a probability measure

More information

Combinatorial Types of Tropical Eigenvector

Combinatorial Types of Tropical Eigenvector Combinatorial Types of Tropical Eigenvector arxiv:1105.55504 Ngoc Mai Tran Department of Statistics, UC Berkeley Joint work with Bernd Sturmfels 2 / 13 Tropical eigenvalues and eigenvectors Max-plus: (R,,

More information

TOPOLOGY HW 8 CLAY SHONKWILER

TOPOLOGY HW 8 CLAY SHONKWILER TOPOLOGY HW 8 CLAY SHONKWILER 55.1 Show that if A is a retract of B 2, then every continuous map f : A A has a fixed point. Proof. Suppose r : B 2 A is a retraction. Thenr A is the identity map on A. Let

More information

LINEAR ALGEBRA KNOWLEDGE SURVEY

LINEAR ALGEBRA KNOWLEDGE SURVEY LINEAR ALGEBRA KNOWLEDGE SURVEY Instructions: This is a Knowledge Survey. For this assignment, I am only interested in your level of confidence about your ability to do the tasks on the following pages.

More information

v( x) u( y) dy for any r > 0, B r ( x) Ω, or equivalently u( w) ds for any r > 0, B r ( x) Ω, or ( not really) equivalently if v exists, v 0.

v( x) u( y) dy for any r > 0, B r ( x) Ω, or equivalently u( w) ds for any r > 0, B r ( x) Ω, or ( not really) equivalently if v exists, v 0. Sep. 26 The Perron Method In this lecture we show that one can show existence of solutions using maximum principle alone.. The Perron method. Recall in the last lecture we have shown the existence of solutions

More information

U.C. Berkeley CS294: Beyond Worst-Case Analysis Handout 12 Luca Trevisan October 3, 2017

U.C. Berkeley CS294: Beyond Worst-Case Analysis Handout 12 Luca Trevisan October 3, 2017 U.C. Berkeley CS94: Beyond Worst-Case Analysis Handout 1 Luca Trevisan October 3, 017 Scribed by Maxim Rabinovich Lecture 1 In which we begin to prove that the SDP relaxation exactly recovers communities

More information

MTH 5102 Linear Algebra Practice Final Exam April 26, 2016

MTH 5102 Linear Algebra Practice Final Exam April 26, 2016 Name (Last name, First name): MTH 5 Linear Algebra Practice Final Exam April 6, 6 Exam Instructions: You have hours to complete the exam. There are a total of 9 problems. You must show your work and write

More information

Math 61CM - Solutions to homework 6

Math 61CM - Solutions to homework 6 Math 61CM - Solutions to homework 6 Cédric De Groote November 5 th, 2018 Problem 1: (i) Give an example of a metric space X such that not all Cauchy sequences in X are convergent. (ii) Let X be a metric

More information

4.2. ORTHOGONALITY 161

4.2. ORTHOGONALITY 161 4.2. ORTHOGONALITY 161 Definition 4.2.9 An affine space (E, E ) is a Euclidean affine space iff its underlying vector space E is a Euclidean vector space. Given any two points a, b E, we define the distance

More information

15-850: Advanced Algorithms CMU, Spring 2017 Lecture #17: The Ellipsoid Algorithm March 3, 2017

15-850: Advanced Algorithms CMU, Spring 2017 Lecture #17: The Ellipsoid Algorithm March 3, 2017 15-850: Advanced Algorithms CMU, Spring 2017 Lecture #17: The Ellipsoid Algorithm March 3, 2017 Lecturer: Anupam Gupta Scribe: Benjamin Berg, David K. Isenberg In this lecture, we discuss some polynomial-time

More information

Ph.D. Katarína Bellová Page 1 Mathematics 2 (10-PHY-BIPMA2) EXAM - Solutions, 20 July 2017, 10:00 12:00 All answers to be justified.

Ph.D. Katarína Bellová Page 1 Mathematics 2 (10-PHY-BIPMA2) EXAM - Solutions, 20 July 2017, 10:00 12:00 All answers to be justified. PhD Katarína Bellová Page 1 Mathematics 2 (10-PHY-BIPMA2 EXAM - Solutions, 20 July 2017, 10:00 12:00 All answers to be justified Problem 1 [ points]: For which parameters λ R does the following system

More information

Rational Distance Sets on Conic Sections

Rational Distance Sets on Conic Sections Rational Distance Sets on Conic Sections Megan D Ly Loyola Marymount Shawn E Tsosie UMASS Amherst July 010 Lyda P Urresta Union College Abstract Leonhard Euler noted that there exists an infinite set of

More information

Online Learning Summer School Copenhagen 2015 Lecture 1

Online Learning Summer School Copenhagen 2015 Lecture 1 Online Learning Summer School Copenhagen 2015 Lecture 1 Shai Shalev-Shwartz School of CS and Engineering, The Hebrew University of Jerusalem Online Learning Shai Shalev-Shwartz (Hebrew U) OLSS Lecture

More information

Network Optimization

Network Optimization Network Optimization Fall 2014 - Math 514 Lecturer: Thomas Rothvoss University of Washington, Seattle Last changes: December 4, 2014 2 Contents 1 The MaxCut problem 5 1.1 Semidefinite programs.........................................

More information

Stochastic Design Criteria in Linear Models

Stochastic Design Criteria in Linear Models AUSTRIAN JOURNAL OF STATISTICS Volume 34 (2005), Number 2, 211 223 Stochastic Design Criteria in Linear Models Alexander Zaigraev N. Copernicus University, Toruń, Poland Abstract: Within the framework

More information

Course 311: Michaelmas Term 2005 Part III: Topics in Commutative Algebra

Course 311: Michaelmas Term 2005 Part III: Topics in Commutative Algebra Course 311: Michaelmas Term 2005 Part III: Topics in Commutative Algebra D. R. Wilkins Contents 3 Topics in Commutative Algebra 2 3.1 Rings and Fields......................... 2 3.2 Ideals...............................

More information

CSC 2414 Lattices in Computer Science September 27, Lecture 4. An Efficient Algorithm for Integer Programming in constant dimensions

CSC 2414 Lattices in Computer Science September 27, Lecture 4. An Efficient Algorithm for Integer Programming in constant dimensions CSC 2414 Lattices in Computer Science September 27, 2011 Lecture 4 Lecturer: Vinod Vaikuntanathan Scribe: Wesley George Topics covered this lecture: SV P CV P Approximating CVP: Babai s Nearest Plane Algorithm

More information

8. Geometric problems

8. Geometric problems 8. Geometric problems Convex Optimization Boyd & Vandenberghe extremal volume ellipsoids centering classification placement and facility location 8 Minimum volume ellipsoid around a set Löwner-John ellipsoid

More information

Math 108b: Notes on the Spectral Theorem

Math 108b: Notes on the Spectral Theorem Math 108b: Notes on the Spectral Theorem From section 6.3, we know that every linear operator T on a finite dimensional inner product space V has an adjoint. (T is defined as the unique linear operator

More information

POSITIVE SEMIDEFINITE INTERVALS FOR MATRIX PENCILS

POSITIVE SEMIDEFINITE INTERVALS FOR MATRIX PENCILS POSITIVE SEMIDEFINITE INTERVALS FOR MATRIX PENCILS RICHARD J. CARON, HUIMING SONG, AND TIM TRAYNOR Abstract. Let A and E be real symmetric matrices. In this paper we are concerned with the determination

More information

APPENDIX A. Background Mathematics. A.1 Linear Algebra. Vector algebra. Let x denote the n-dimensional column vector with components x 1 x 2.

APPENDIX A. Background Mathematics. A.1 Linear Algebra. Vector algebra. Let x denote the n-dimensional column vector with components x 1 x 2. APPENDIX A Background Mathematics A. Linear Algebra A.. Vector algebra Let x denote the n-dimensional column vector with components 0 x x 2 B C @. A x n Definition 6 (scalar product). The scalar product

More information

Learning symmetric non-monotone submodular functions

Learning symmetric non-monotone submodular functions Learning symmetric non-monotone submodular functions Maria-Florina Balcan Georgia Institute of Technology ninamf@cc.gatech.edu Nicholas J. A. Harvey University of British Columbia nickhar@cs.ubc.ca Satoru

More information

Technische Universität München, Zentrum Mathematik Lehrstuhl für Angewandte Geometrie und Diskrete Mathematik. Combinatorial Optimization (MA 4502)

Technische Universität München, Zentrum Mathematik Lehrstuhl für Angewandte Geometrie und Diskrete Mathematik. Combinatorial Optimization (MA 4502) Technische Universität München, Zentrum Mathematik Lehrstuhl für Angewandte Geometrie und Diskrete Mathematik Combinatorial Optimization (MA 4502) Dr. Michael Ritter Problem Sheet 1 Homework Problems Exercise

More information

Convex Geometry. Carsten Schütt

Convex Geometry. Carsten Schütt Convex Geometry Carsten Schütt November 25, 2006 2 Contents 0.1 Convex sets... 4 0.2 Separation.... 9 0.3 Extreme points..... 15 0.4 Blaschke selection principle... 18 0.5 Polytopes and polyhedra.... 23

More information

University of Colorado Denver Department of Mathematical and Statistical Sciences Applied Linear Algebra Ph.D. Preliminary Exam June 10, 2011

University of Colorado Denver Department of Mathematical and Statistical Sciences Applied Linear Algebra Ph.D. Preliminary Exam June 10, 2011 University of Colorado Denver Department of Mathematical and Statistical Sciences Applied Linear Algebra PhD Preliminary Exam June 2 Name: Exam Rules: This is a closed book exam Once the exam begins you

More information

MATRICES ARE SIMILAR TO TRIANGULAR MATRICES

MATRICES ARE SIMILAR TO TRIANGULAR MATRICES MATRICES ARE SIMILAR TO TRIANGULAR MATRICES 1 Complex matrices Recall that the complex numbers are given by a + ib where a and b are real and i is the imaginary unity, ie, i 2 = 1 In what we describe below,

More information

DIAGONALIZATION. In order to see the implications of this definition, let us consider the following example Example 1. Consider the matrix

DIAGONALIZATION. In order to see the implications of this definition, let us consider the following example Example 1. Consider the matrix DIAGONALIZATION Definition We say that a matrix A of size n n is diagonalizable if there is a basis of R n consisting of eigenvectors of A ie if there are n linearly independent vectors v v n such that

More information

VAR Model. (k-variate) VAR(p) model (in the Reduced Form): Y t-2. Y t-1 = A + B 1. Y t + B 2. Y t-p. + ε t. + + B p. where:

VAR Model. (k-variate) VAR(p) model (in the Reduced Form): Y t-2. Y t-1 = A + B 1. Y t + B 2. Y t-p. + ε t. + + B p. where: VAR Model (k-variate VAR(p model (in the Reduced Form: where: Y t = A + B 1 Y t-1 + B 2 Y t-2 + + B p Y t-p + ε t Y t = (y 1t, y 2t,, y kt : a (k x 1 vector of time series variables A: a (k x 1 vector

More information

Eigenvalues, random walks and Ramanujan graphs

Eigenvalues, random walks and Ramanujan graphs Eigenvalues, random walks and Ramanujan graphs David Ellis 1 The Expander Mixing lemma We have seen that a bounded-degree graph is a good edge-expander if and only if if has large spectral gap If G = (V,

More information

1 Perfect Matching and Matching Polytopes

1 Perfect Matching and Matching Polytopes CS 598CSC: Combinatorial Optimization Lecture date: /16/009 Instructor: Chandra Chekuri Scribe: Vivek Srikumar 1 Perfect Matching and Matching Polytopes Let G = (V, E be a graph. For a set E E, let χ E

More information

Final Review Sheet. B = (1, 1 + 3x, 1 + x 2 ) then 2 + 3x + 6x 2

Final Review Sheet. B = (1, 1 + 3x, 1 + x 2 ) then 2 + 3x + 6x 2 Final Review Sheet The final will cover Sections Chapters 1,2,3 and 4, as well as sections 5.1-5.4, 6.1-6.2 and 7.1-7.3 from chapters 5,6 and 7. This is essentially all material covered this term. Watch

More information

Course Notes for EE227C (Spring 2018): Convex Optimization and Approximation

Course Notes for EE227C (Spring 2018): Convex Optimization and Approximation Course Notes for EE7C (Spring 018): Convex Optimization and Approximation Instructor: Moritz Hardt Email: hardt+ee7c@berkeley.edu Graduate Instructor: Max Simchowitz Email: msimchow+ee7c@berkeley.edu February

More information

Matrix decompositions

Matrix decompositions Matrix decompositions Zdeněk Dvořák May 19, 2015 Lemma 1 (Schur decomposition). If A is a symmetric real matrix, then there exists an orthogonal matrix Q and a diagonal matrix D such that A = QDQ T. The

More information

Online Learning, Mistake Bounds, Perceptron Algorithm

Online Learning, Mistake Bounds, Perceptron Algorithm Online Learning, Mistake Bounds, Perceptron Algorithm 1 Online Learning So far the focus of the course has been on batch learning, where algorithms are presented with a sample of training data, from which

More information

3.8 Strong valid inequalities

3.8 Strong valid inequalities 3.8 Strong valid inequalities By studying the problem structure, we can derive strong valid inequalities which lead to better approximations of the ideal formulation conv(x ) and hence to tighter bounds.

More information