3.3 Easy ILP problems and totally unimodular matrices

Size: px
Start display at page:

Download "3.3 Easy ILP problems and totally unimodular matrices"

Transcription

1 3.3 Easy ILP problems and totally unimodular matrices Consider a generic ILP problem expressed in standard form where A Z m n with n m, and b Z m. min{c t x : Ax = b, x Z n +} (1) P(b) = {x R n : Ax = b, x 0 } is the polyhedron of the feasible solutions of the linear relaxation of (1). Assumption: A is of full rank (since n m, rank(a)=m), i.e, there are no redundant constraints. In general, optimal solutions of the linear relaxation are fractional and far away from optimal solutions of the ILP problem (1). If all the extreme points of P(b) are integer, the formulation is ideal and the ILP problem can be solved by just solving its linear relaxation. Edoardo Amaldi (PoliMI) Optimization A.Y / 14

2 According to Linear Programming theory: Any linear program min{c t x : Ax = b, x 0} with a finite optimal solution has an optimal extreme point. To each extreme point of P(b) corresponds (at least) one basic feasible solution x = (x B, x N ) = (B 1 b, 0), where B is a basis of A, i.e., an m m non singular submatrix of A. Consider any basis B. By partitioning the columns of A in basic and non basic columns, the system Ax = b, x 0 can be written as and can be expressed in canonical form: Bx B + Nx N = b with x B 0 and x N 0, x B = B 1 b B 1 Nx N with x B 0 and x N 0, which emphasizes the basic feasible solution (x B, x N ) = (B 1 b, 0). Edoardo Amaldi (PoliMI) Optimization A.Y / 14

3 Observation: If an optimal basis B has det(b) = ±1, the optimal solution of the linear relaxation of (1) is integer and hence is also an optimal solution for the ILP problem (1). Proof: We know that x B = B 1 b and B 1 = 1 det(b) C t, where C is the matrix of cofactors α ij = ( 1) i+j det(b ij ) with B ij the squared submatrix obtained from B by deleting the i-th row and the j-th column. If all the components of B are integer, all the cofactors α ij are integer. If det(b) = ±1, B 1 is integer and therefore, since b is integer, also the basic feasible solution (x B, x N ) = (B 1 b, 0) is integer. Since B 1 b is integer also if det(b) = 2 and all components of b are even integers, det(b) = ±1 is only a sufficient condition for the integrality of (x B, x N ) = (B 1 b, 0). Edoardo Amaldi (PoliMI) Optimization A.Y / 14

4 3.3.1 Totally unimodular matrices and optimal integer solutions Definition: An m n matrix A with integer coefficients is totally unimodular (TU) if every squared submatrix has a determinant 1, 0 or 1. Clearly, if A is TU, a ij { 1, 0, 1} for all i and j. Examples of matrices that are TU and not TU: Recall: The Laplace expansion of the determinant of an m m matrix B along any row i, with 1 i m, is given by det(b) = m j=1 b ijα ij, where α ij = ( 1) i+j det(b ij ) are the cofactors of B. Analogously the expansion along any column j, with 1 j m, is det(b) = m i=1 a ijα ij. Proposition: A is TU if and only if A t is TU. A is TU if and only if (A I m) is TU. The matrix A obtained from A by permuting and/or changing the sign of some columns and/or rows is TU if and only if A is TU. Edoardo Amaldi (PoliMI) Optimization A.Y / 14

5 Theorem 1: Consider an m n TU matrix A and an integer vector b such that P(b) = {x R n : Ax = b, x 0} =, then all the extreme points (vertices) of the polyhedron P(b) are integer. Proof: As for the previous observation. The result also holds for ILP problems in inequality form. Edoardo Amaldi (PoliMI) Optimization A.Y / 14

6 Corollary: Consider an m n TU matrix A and an integer vector b such that P(b) = {x R n : Ax b, x 0} =, then all the extreme points of the polyhedron P(b) are integer. Proof: Let x be any arbitrary extreme point of P(b). First we show that ( x, s) with s := A x b is an extreme of P (b) := {(x, s) R n+m : Ax s = b, (x, s) 0}. If it were false, there would exist two distinct points (x 1, s 1 ) and (x 2, s 2 ) of P (b) such that ( x, s) = α(x 1, s 1 ) + (1 α)(x 2, s 2 ) for some α with 0 < α < 1. Since s 1 = Ax 1 b 0 and s 2 = Ax 2 b 0, x 1 and x 2 belong to P(b). Moreover, (x 1, s 1 ) (x 2, s 2 ) would imply x 1 x 2 and hence x = αx 1 + (1 α)x 2 could not be a vertex of P(b). Since A is TU, also (A I m) is TU. According to Theorem 1 for P (b), ( x, s) is integer, hence also x is integer. Edoardo Amaldi (PoliMI) Optimization A.Y / 14

7 Sufficient conditions for a matrix to be TU Proposition: An m n matrix A with integer coefficients is TU if i) a ij { 1, 0, +1} for all i and j, ii) each column of A contains at most two nonzero coefficients, iii) the set I of the indices of the rows of A can be partitioned into two subsets I 1 and I 2 such that, for each column j with two nonzero coefficients, we have i I 1 a ij i I 2 a ij = 0. N.B.: If a column has two nonzero coefficients of the same (different) sign, their rows must belong to different subsets (same subset) of indices I 1 and I 2. Proof: Suppose A is not TU, and let Q be a smallest squared submatrix of A among those with det(q) / { 1, 0, 1}. Q cannot contain a column with a single nonzero coefficient, otherwise Q would not be minimal. Thus, each column of Q contains exactly two nonzero coefficients. According to (iii), i I 1 a ij i I 2 a ij = 0 and hence, since a ij = 0 for every column j of Q and every row i not in Q, we would have det(q) = 0, which is a contradiction. Examples of TU matrices (not) satisfying these conditions: Edoardo Amaldi (PoliMI) Optimization A.Y / 14

8 Characterization of TU matrices Theorem 2: An m n matrix A with integer coefficients is TU if and only if each subset I I = {1,..., m} of indices of the rows of A can be partitioned into two subsets I 1 and I 2 such that the sum of the rows in I 1 minus the sum of the rows in I 2 is a vector with all components in { 1, 0, 1}, that is i I 1 a ij i I 2 a ij 1 for every column j. From the ILP point of view, when the matrix A is TU it suffices to solve the linear relaxation. In some sense also the converse is true: Proposition: The linear program min{c t x : Ax = b, x R n +} has an optimal integer solution for any integer vector b for which it admits a finite optimal solution if and only if A is TU. It can be shown that, given A and a basis B with det(b) 1, there always exists a LP min{c t x : Ax = b, x R n +}, for appropriate c and b, with a fractional optimal solution. Edoardo Amaldi (PoliMI) Optimization A.Y / 14

9 3.3.2 Examples of natural formulations that are ideal 1) Assignment problem Given n jobs and n machines with costs c ij for all i, j {1,..., n}, decide which job to assign to which machine so as to minimize the total cost to complete all the jobs. ILP formulation: min n i=1 n j=1 c ijx ij n i=1 x ij = 1 j (2) n j=1 x ij = 1 i (3) x ij {0, 1} i, j where x ij = 1 if i-th job is assigned to j-th machine, with 1 i, j n. N.B.: In the linear relaxation, x ij 0 i, j suffice, because (2)-(3) imply x ij 1 i, j. Property: The matrix of constraints (2)-(3) is TU. In the sufficient conditions for TU, just take I 1 = {1,..., n} and I 2 = {n + 1,..., 2n}. Consequence: Each extreme point of the feasible region of the linear relaxation is integer, and the formulation is ideal. Edoardo Amaldi (PoliMI) Optimization A.Y / 14

10 2) Transportation problem Suppose there is a single type of product. Given m production plants (1 i m) n clients (1 j n) c ij = transportation cost of one unit of product from plant i to client j p i = maximum amount that can be produced (capacity) at plant i d j = demand of client j q ij = maximum amount that can be transported from plant i to client j determine a transportation plan so as to minimize total transportation costs while satisfying all client demands and all the plant capacity constraints. Assumption: m i=1 p i n j=1 d j (to guarantee feasibility) Without loss of generality, we can assume that m i=1 p i = n j=1 d j. Edoardo Amaldi (PoliMI) Optimization A.Y / 14

11 Natural ILP formulation: Decision variables: x ij = amount of product transported from plant i to client j, with 1 i m and 1 j n min m i=1 n j=1 c ijx ij n j=1 x ij p i i (4) m i=1 x ij d j j (5) x ij q ij i, j (6) x ij 0 integer i, j Property: The matrix of constraints (4)-(6) is TU. Proof: Put the problem in consistent inequality form by multiplying all constraints (4) and (6) by 1. Since the matrix of the demand and capacity constraints is TU, we just apply the Corollary. Consequence: If all p i, d j and q ij are integer, every extreme point of the feasible region of the linear relaxation is integer and, hence, the formulation is ideal. Edoardo Amaldi (PoliMI) Optimization A.Y / 14

12 3) Minimum cost flow problem Given a directed graph G = (V, A) with a capacity u ij and a unit cost c ij for each arc (i, j) A, and a demand/availability b i for each node i V (b i < 0 for sources, b i > 0 for destinations, and i V b i = 0), determine a feasible flow of minimum total cost which satisfies all the demands. Natutal ILP formulation: min (i,j) A c ijx ij (h,i) δ (i) x hi (i,j) δ + (i) x ij = b i i V (7) x ij u ij (i, j) A (8) x ij 0 integer (i, j) A where δ + (i) = {(i, j) A : j V } and δ (i) = {(h, i) A : h V }. Property: The matrix of constraints (7)-(8) is TU. Proof: The n A node-arc incidence matrix of the directed graph is TU since it contains exactly one 1 and one 1 per column (take I 1 = I and I 2 = ). The matrix obtained by juxtaposing to A the identity matrix I A A of the capacity constraints (8) is also TU. Edoardo Amaldi (PoliMI) Optimization A.Y / 14

13 Consequence: If the demands/availabilities b i and the capacities u ij are integer, every extreme point is integer and, thus, the formulation is ideal. Exercise: Show that the following well-known combinatorial optimization problems are special cases of the Min cost flow problem. - Shortest path: Given a directed graph G = (V, A) with a cost c ij associated to each arc (i, j) A, and two prescribed nodes s and t, determine a minimum cost path from s to t. - Maximum flow: Given a directed graph G = (V, A) with a capacity u ij associated to each arc (i, j) A, a source s and a sink t, determine a feasible flow of maximum value from s to t. Edoardo Amaldi (PoliMI) Optimization A.Y / 14

14 Rounding optimal solutions of the linear relaxation In general, when the constraint matrix of an ILP is not TU, the optimal solution x PL of its linear relaxation is fractional. As a first attempt to derive a sub-optimal solution of the ILP, we may try to round each fractional component of x PL to the closest integer value. However the rounded solution are often infeasible for the ILP, even if the rounded solution is feasible for the ILP, the error with respect to an optimal ILP solution may be arbitrarily large. In general, by rounding x PL we obtain a good approximation of the optimal solution only when the components of x PL have large values. Illustration of the different cases: Edoardo Amaldi (PoliMI) Optimization A.Y / 14

. P. 4.3 Basic feasible solutions and vertices of polyhedra. x 1. x 2

. P. 4.3 Basic feasible solutions and vertices of polyhedra. x 1. x 2 4. Basic feasible solutions and vertices of polyhedra Due to the fundamental theorem of Linear Programming, to solve any LP it suffices to consider the vertices (finitely many) of the polyhedron P of the

More information

Lecture 11: Integer Linear Programs

Lecture 11: Integer Linear Programs Lecture 11: Integer Linear Programs 1 Integer Linear Programs Consider the following version of Woody s problem: x 1 = chairs per day, x 2 = tables per day max z = 35x 1 + 60x 2 profits 14x 1 + 26x 2 190

More information

5 INTEGER LINEAR PROGRAMMING (ILP) E. Amaldi Fondamenti di R.O. Politecnico di Milano 1

5 INTEGER LINEAR PROGRAMMING (ILP) E. Amaldi Fondamenti di R.O. Politecnico di Milano 1 5 INTEGER LINEAR PROGRAMMING (ILP) E. Amaldi Fondamenti di R.O. Politecnico di Milano 1 General Integer Linear Program: (ILP) min c T x Ax b x 0 integer Assumption: A, b integer The integrality condition

More information

Chapter 2 Integer Programming. Paragraph 1 Total Unimodularity

Chapter 2 Integer Programming. Paragraph 1 Total Unimodularity Chapter 2 Integer Programming Paragraph 1 Total Unimodularity What we did so far We studied algorithms for solving linear programs Simplex (primal, dual, and primal-dual) Ellipsoid Method (proving LP is

More information

Integer Programming Theory

Integer Programming Theory Integer Programming Theory Laura Galli November 11, 2014 In the following we assume all functions are linear, hence we often drop the term linear. In discrete optimization, we seek to find a solution x

More information

1 Bipartite matching and vertex covers

1 Bipartite matching and vertex covers ORF 523 Lecture 6 Spring 2016, Princeton University Instructor: A.A. Ahmadi Scribe: G. Hall Thursday, February 25, 2016 When in doubt on the accuracy of these notes, please cross check with the instructor

More information

Week 5 Integral Polyhedra

Week 5 Integral Polyhedra Week 5 Integral Polyhedra We have seen some examples 1 of linear programming formulation that are integral, meaning that every basic feasible solution is an integral vector. This week we develop a theory

More information

Good luck, veel succes!

Good luck, veel succes! Final exam Advanced Linear Programming, May 7, 13.00-16.00 Switch off your mobile phone, PDA and any other mobile device and put it far away. No books or other reading materials are allowed. This exam

More information

Totally Unimodular Matrices

Totally Unimodular Matrices Totally Unimodular Matrices m constraints, n variables Vertex solution: unique solution of n linearly independent tight inequalities Can be rewritten as: That is: Totally Unimodular Matrices Assuming all

More information

Min-cost flow problems and network simplex algorithm

Min-cost flow problems and network simplex algorithm Min-cost flow problems and network simplex algorithm The particular structure of some LP problems can be sometimes used for the design of solution techniques more efficient than the simplex algorithm.

More information

1 The Minimum Spanning Tree (MST) Problem

1 The Minimum Spanning Tree (MST) Problem IEOR269: Lecture 3 1 Integer Programming and Combinatorial Optimization Lecture #3 IEOR 269 Feb 5, 2010 Lecturer: Dorit Hochbaum Scribe: Dogan Bilguvar Handouts: Game of Fiver and Fire Station Problems

More information

Geometry of Linear Programming

Geometry of Linear Programming Chapter 2 Geometry of Linear Programming The intent of this chapter is to provide a geometric interpretation of linear programming problems. To conceive fundamental concepts and validity of different algorithms

More information

Extreme Points, Corners, and Basic Feasible Solutions

Extreme Points, Corners, and Basic Feasible Solutions Extreme Points, Corners, and Basic Feasible Solutions CS 149 Staff September 25, 2008 1 Extreme Points and Convexity Example Claim 1. ǫ({5}) = {5} Proof sketch: since there are no other points in the set,

More information

Sample Final Examination Questions IE406 Introduction to Mathematical Programming Dr. Ralphs

Sample Final Examination Questions IE406 Introduction to Mathematical Programming Dr. Ralphs Sample Final Examination Questions IE0 Introduction to Mathematical Programming Dr. Ralphs. Consider the following linear programming problem and its optimal final tableau. min x x + x x + x + x 8 x +

More information

Linear Programming and Network Optimization

Linear Programming and Network Optimization Linear Programming and Network Optimization Jonathan Turner March 1, 1 Many of the problem we have been studying can be viewed as special cases of the more general linear programming problem (LP). In the

More information

Lecture 3: Linear Programming Relaxations and Rounding

Lecture 3: Linear Programming Relaxations and Rounding Lecture 3: Linear Programming Relaxations and Rounding 1 Approximation Algorithms and Linear Relaxations For the time being, suppose we have a minimization problem. Many times, the problem at hand can

More information

Quadratic programming on graphs without long odd cycles

Quadratic programming on graphs without long odd cycles Quadratic programming on graphs without long odd cycles Marcin Kamiński Department of Computer Science Université Libre de Bruxelles, Belgium Marcin.Kaminski@ulb.ac.be Abstract We introduce a quadratic

More information

3. Linear Programming and Polyhedral Combinatorics

3. Linear Programming and Polyhedral Combinatorics Massachusetts Institute of Technology Handout 6 18.433: Combinatorial Optimization February 20th, 2009 Michel X. Goemans 3. Linear Programming and Polyhedral Combinatorics Summary of what was seen in the

More information

Determinants. Dr. Doreen De Leon Math 152, Fall 2015

Determinants. Dr. Doreen De Leon Math 152, Fall 2015 Determinants Dr. Doreen De Leon Math 52, Fall 205 Determinant of a Matrix Elementary Matrices We will first discuss matrices that can be used to produce an elementary row operation on a given matrix A.

More information

Math 240 Calculus III

Math 240 Calculus III The Calculus III Summer 2013, Session II Tuesday, July 16, 2013 Agenda 1. of the determinant 2. determinants 3. of determinants What is the determinant? Yesterday: Ax = b has a unique solution when A is

More information

1. Lecture notes on bipartite matching

1. Lecture notes on bipartite matching Massachusetts Institute of Technology Handout 3 18.433: Combinatorial Optimization February 9th, 2009 Michel X. Goemans 1. Lecture notes on bipartite matching Matching problems are among the fundamental

More information

(where C, X R and are column vectors)

(where C, X R and are column vectors) CS 05: Algorithms (Grad) What is a Linear Programming Problem? A linear program (LP) is a minimization problem where we are asked to minimize a given linear function subject to one or more linear inequality

More information

Minimally Infeasible Set Partitioning Problems with Balanced Constraints

Minimally Infeasible Set Partitioning Problems with Balanced Constraints Minimally Infeasible Set Partitioning Problems with alanced Constraints Michele Conforti, Marco Di Summa, Giacomo Zambelli January, 2005 Revised February, 2006 Abstract We study properties of systems of

More information

THEORY OF SIMPLEX METHOD

THEORY OF SIMPLEX METHOD Chapter THEORY OF SIMPLEX METHOD Mathematical Programming Problems A mathematical programming problem is an optimization problem of finding the values of the unknown variables x, x,, x n that maximize

More information

Minimal spanning and maximal independent sets, Basis and Dimension

Minimal spanning and maximal independent sets, Basis and Dimension Minimal spanning and maximal independent sets, Basis and Dimension February 23, 2010 Let S be a set of real n-vectors. In particular, S can be a linear space, or its subspace. Yet, S can also be just a

More information

SUBSPACES. Chapter Introduction. 3.2 Subspaces of F n

SUBSPACES. Chapter Introduction. 3.2 Subspaces of F n Chapter 3 SUBSPACES 3. Introduction Throughout this chapter, we will be studying F n, the set of all n dimensional column vectors with components from a field F. We continue our study of matrices by considering

More information

1 Polyhedra and Linear Programming

1 Polyhedra and Linear Programming CS 598CSC: Combinatorial Optimization Lecture date: January 21, 2009 Instructor: Chandra Chekuri Scribe: Sungjin Im 1 Polyhedra and Linear Programming In this lecture, we will cover some basic material

More information

Week 10. c T x + d T y. Ax + Dy = b. x integer.

Week 10. c T x + d T y. Ax + Dy = b. x integer. Week 10 1 Integer Linear Programg So far we considered LP s with continuous variables. However sometimes decisions are intrinsically restricted to integer numbers. E.g., in the long term planning, decision

More information

Chapter 2 Incidence Matrix

Chapter 2 Incidence Matrix Chapter 2 Incidence Matrix Let G be a graph with V (G) ={1,...,n} and E(G) ={e 1,...,e m }. Suppose each edge of G is assigned an orientation, which is arbitrary but fixed. The (vertex-edge) incidence

More information

MATRIX ALGEBRA AND SYSTEMS OF EQUATIONS. + + x 2. x n. a 11 a 12 a 1n b 1 a 21 a 22 a 2n b 2 a 31 a 32 a 3n b 3. a m1 a m2 a mn b m

MATRIX ALGEBRA AND SYSTEMS OF EQUATIONS. + + x 2. x n. a 11 a 12 a 1n b 1 a 21 a 22 a 2n b 2 a 31 a 32 a 3n b 3. a m1 a m2 a mn b m MATRIX ALGEBRA AND SYSTEMS OF EQUATIONS 1. SYSTEMS OF EQUATIONS AND MATRICES 1.1. Representation of a linear system. The general system of m equations in n unknowns can be written a 11 x 1 + a 12 x 2 +

More information

Math 1180, Hastings. Notes, part 9

Math 1180, Hastings. Notes, part 9 Math 8, Hastings Notes, part 9 First I want to recall the following very important theorem, which only concerns square matrices. We will need to use parts of this frequently. Theorem Suppose that A is

More information

10. Graph Matrices Incidence Matrix

10. Graph Matrices Incidence Matrix 10 Graph Matrices Since a graph is completely determined by specifying either its adjacency structure or its incidence structure, these specifications provide far more efficient ways of representing a

More information

6. Mixed Integer Linear Programming

6. Mixed Integer Linear Programming 6. Mixed Integer Linear Programming Javier Larrosa Albert Oliveras Enric Rodríguez-Carbonell Problem Solving and Constraint Programming (RPAR) Session 6 p.1/40 Mixed Integer Linear Programming A mixed

More information

Math 5593 Linear Programming Weeks 12/13

Math 5593 Linear Programming Weeks 12/13 Math 5593 Linear Programming Weeks 12/13 University of Colorado Denver, Fall 2013, Prof. Engau 1 Introduction 2 Polyhedral Theory 3 LP and Lagrangean Relaxation 4 Computational Methods and Algorithms Integer

More information

Approximation Algorithms

Approximation Algorithms Approximation Algorithms or: How I Learned to Stop Worrying and Deal with NP-Completeness Ong Jit Sheng, Jonathan (A0073924B) March, 2012 Overview Key Results (I) General techniques: Greedy algorithms

More information

Duality. Uri Feige. November 17, 2011

Duality. Uri Feige. November 17, 2011 Duality Uri Feige November 17, 2011 1 Linear programming duality 1.1 The diet problem revisited Recall the diet problem from Lecture 1. There are n foods, m nutrients, and a person (the buyer) is required

More information

2.6 The Inverse of a Square Matrix

2.6 The Inverse of a Square Matrix 200/2/6 page 62 62 CHAPTER 2 Matrices and Systems of Linear Equations 0 0 2 + i i 2i 5 A = 0 9 0 54 A = i i 4 + i 2 0 60 i + i + 5i 26 The Inverse of a Square Matrix In this section we investigate the

More information

We have the following important fact whose proof we leave as an exercise for the reader.

We have the following important fact whose proof we leave as an exercise for the reader. LP Geometry We now briefly turn to a discussion of LP geometry extending the geometric ideas developed in Section for dimensional LPs to n dimensions. In this regard, the key geometric idea is the notion

More information

2.3 Convex Constrained Optimization Problems

2.3 Convex Constrained Optimization Problems 42 CHAPTER 2. FUNDAMENTAL CONCEPTS IN CONVEX OPTIMIZATION Theorem 15 Let f : R n R and h : R R. Consider g(x) = h(f(x)) for all x R n. The function g is convex if either of the following two conditions

More information

10.1 Integer Programming and LP relaxation

10.1 Integer Programming and LP relaxation CS787: Advanced Algorithms Lecture 10: LP Relaxation and Rounding In this lecture we will design approximation algorithms using linear programming. The key insight behind this approach is that the closely

More information

Approximation Algorithms: LP Relaxation, Rounding, and Randomized Rounding Techniques. My T. Thai

Approximation Algorithms: LP Relaxation, Rounding, and Randomized Rounding Techniques. My T. Thai Approximation Algorithms: LP Relaxation, Rounding, and Randomized Rounding Techniques My T. Thai 1 Overview An overview of LP relaxation and rounding method is as follows: 1. Formulate an optimization

More information

4.6 Linear Programming duality

4.6 Linear Programming duality 4.6 Linear Programming duality To any minimization (maximization) LP we can associate a closely related maximization (minimization) LP. Different spaces and objective functions but in general same optimal

More information

(a) The transpose of a lower triangular matrix is upper triangular, and the transpose of an upper triangular matrix is lower triangular.

(a) The transpose of a lower triangular matrix is upper triangular, and the transpose of an upper triangular matrix is lower triangular. Theorem.7.: (Properties of Triangular Matrices) (a) The transpose of a lower triangular matrix is upper triangular, and the transpose of an upper triangular matrix is lower triangular. (b) The product

More information

MATHEMATICAL BACKGROUND

MATHEMATICAL BACKGROUND Chapter 1 MATHEMATICAL BACKGROUND This chapter discusses the mathematics that is necessary for the development of the theory of linear programming. We are particularly interested in the solutions of a

More information

13.1 Semi-Definite Programming

13.1 Semi-Definite Programming CS787: Advanced Algorithms Lecture 13: Semi-definite programming In this lecture we will study an extension of linear programming, namely semi-definite programming. Semi-definite programs are much more

More information

Integer Programming: Large Scale Methods Part I (Chapter 16)

Integer Programming: Large Scale Methods Part I (Chapter 16) Integer Programming: Large Scale Methods Part I (Chapter 16) University of Chicago Booth School of Business Kipp Martin November 9, 2016 1 / 55 Outline List of Files Key Concepts Column Generation/Decomposition

More information

Algorithms for Integer Programming

Algorithms for Integer Programming Algorithms for Integer Programming Laura Galli December 18, 2014 Unlike linear programming problems, integer programming problems are very difficult to solve. In fact, no efficient general algorithm is

More information

6.854 Advanced Algorithms. Scribes: Jay Kumar Sundararajan. Duality

6.854 Advanced Algorithms. Scribes: Jay Kumar Sundararajan. Duality Lecture 1: 10/13/2004 6.854 Advanced Algorithms Scribes: Jay Kumar Sundararajan Lecturer: David Karger Duality This lecture covers weak and strong duality, and also explains the rules for finding the dual

More information

MATRIX ALGEBRA AND SYSTEMS OF EQUATIONS

MATRIX ALGEBRA AND SYSTEMS OF EQUATIONS MATRIX ALGEBRA AND SYSTEMS OF EQUATIONS Systems of Equations and Matrices Representation of a linear system The general system of m equations in n unknowns can be written a x + a 2 x 2 + + a n x n b a

More information

Linear Programming Applications. Assignment Problem

Linear Programming Applications. Assignment Problem Linear Programming Applications Assignment Problem 1 Introduction Assignment problem is a particular class of transportation linear programming problems Supplies and demands will be integers (often 1)

More information

Notes on Determinant

Notes on Determinant ENGG2012B Advanced Engineering Mathematics Notes on Determinant Lecturer: Kenneth Shum Lecture 9-18/02/2013 The determinant of a system of linear equations determines whether the solution is unique, without

More information

1 Determinants. Definition 1

1 Determinants. Definition 1 Determinants The determinant of a square matrix is a value in R assigned to the matrix, it characterizes matrices which are invertible (det 0) and is related to the volume of a parallelpiped described

More information

3.1 Simplex Method for Problems in Feasible Canonical Form

3.1 Simplex Method for Problems in Feasible Canonical Form Chapter SIMPLEX METHOD In this chapter, we put the theory developed in the last to practice We develop the simplex method algorithm for LP problems given in feasible canonical form and standard form We

More information

Minimize subject to. x S R

Minimize subject to. x S R Chapter 12 Lagrangian Relaxation This chapter is mostly inspired by Chapter 16 of [1]. In the previous chapters, we have succeeded to find efficient algorithms to solve several important problems such

More information

, we define the determinant of a 21 a 22 A, (also denoted by deta,) to be the scalar. det A = a 11 a 22 a 12 a 21.

, we define the determinant of a 21 a 22 A, (also denoted by deta,) to be the scalar. det A = a 11 a 22 a 12 a 21. 70 Chapter 4 DETERMINANTS [ ] a11 a DEFINITION 401 If A 12, we define the determinant of a 21 a 22 A, (also denoted by deta,) to be the scalar The notation a 11 a 12 a 21 a 22 det A a 11 a 22 a 12 a 21

More information

Lecture 3: The Simplex Method. Reading: Chapter 2

Lecture 3: The Simplex Method. Reading: Chapter 2 Lecture 3: The Simplex Method Reading: Chapter 2 1 The Simplex Tableau We start with the standard equality form LP max z = c 1 x 1 + c 2 x 2 + + c n x n a 11 x 1 + a 12 x 2 + + a 1n x n = b 1 a 21 x 1

More information

Permutation Betting Markets: Singleton Betting with Extra Information

Permutation Betting Markets: Singleton Betting with Extra Information Permutation Betting Markets: Singleton Betting with Extra Information Mohammad Ghodsi Sharif University of Technology ghodsi@sharif.edu Hamid Mahini Sharif University of Technology mahini@ce.sharif.edu

More information

LINEAR PROGRAMMING. Vazirani Chapter 12 Introduction to LP-Duality. George Alexandridis(NTUA)

LINEAR PROGRAMMING. Vazirani Chapter 12 Introduction to LP-Duality. George Alexandridis(NTUA) LINEAR PROGRAMMING Vazirani Chapter 12 Introduction to LP-Duality George Alexandridis(NTUA) gealexan@mail.ntua.gr 1 LINEAR PROGRAMMING What is it? A tool for optimal allocation of scarce resources, among

More information

Unit 19 Properties of Determinants

Unit 19 Properties of Determinants Unit 9 Properties of Determinants Theorem 9.. Suppose A and B are identical n n matrices with the exception that one row (or column) of B is obtained by multiplying the corresponding row (or column) of

More information

Simplex Methods. a ij a ij < 0,

Simplex Methods. a ij a ij < 0, Simplex Methods The main idea of primal cost improvement is to start with a feasible flow vector x and to generate a sequence of other feasible flow vectors, each having a smaller primal cost than its

More information

Linear programming. Lecture 8. Kin Cheong Sou

Linear programming. Lecture 8. Kin Cheong Sou Lecture 8 Linear programming Kin Cheong Sou Department of Mathematical Sciences Chalmers University of Technology and Göteborg University November 17, 2013 Linear programs (LP) Formulation Consider a linear

More information

The Simplex Method. yyye

The Simplex Method.  yyye Yinyu Ye, MS&E, Stanford MS&E310 Lecture Note #05 1 The Simplex Method Yinyu Ye Department of Management Science and Engineering Stanford University Stanford, CA 94305, U.S.A. http://www.stanford.edu/

More information

CSC Linear Programming and Combinatorial Optimization Lecture 12: Approximation Algorithms using Tools from LP

CSC Linear Programming and Combinatorial Optimization Lecture 12: Approximation Algorithms using Tools from LP S2411 - Linear Programming and ombinatorial Optimization Lecture 12: Approximation Algorithms using Tools from LP Notes taken by Matei David April 12, 2005 Summary: In this lecture, we give three example

More information

1 NP-Completeness (continued)

1 NP-Completeness (continued) Mathematical Programming Lecture 27 OR 630 Fall 2006 November 30, 2006 adapted from notes by Omer Hancer 1 NP-Completeness (continued) Before starting on complexity theory, we talked about running times,

More information

1 Introduction. A Partition Problem for Sets of Permutation Matrices

1 Introduction. A Partition Problem for Sets of Permutation Matrices A Partition Problem for Sets of Permutation Matrices Richard A. Brualdi Department of Mathematics, University of Wisconsin, Madison, WI 53706. brualdi@math.wisc.edu Hanley Chiang and Chi-Kwong Li Department

More information

Lecture notes 2 February 1, Convex optimization

Lecture notes 2 February 1, Convex optimization Lecture notes 2 February 1, 2016 Notation Convex optimization Matrices are written in uppercase: A, vectors are written in lowercase: a. A ij denotes the element of A in position (i, j), A i denotes the

More information

Solution. Area(OABC) = Area(OAB) + Area(OBC) = 1 2 det( [ 5 2 1 2. Question 2. Let A = (a) Calculate the nullspace of the matrix A.

Solution. Area(OABC) = Area(OAB) + Area(OBC) = 1 2 det( [ 5 2 1 2. Question 2. Let A = (a) Calculate the nullspace of the matrix A. Solutions to Math 30 Take-home prelim Question. Find the area of the quadrilateral OABC on the figure below, coordinates given in brackets. [See pp. 60 63 of the book.] y C(, 4) B(, ) A(5, ) O x Area(OABC)

More information

Last time we saw that, given a (minimizing) linear program in equational form, one of the following three possibilities is true:

Last time we saw that, given a (minimizing) linear program in equational form, one of the following three possibilities is true: Lecture 2 Geometry of LPs Last time we saw that, given a (minimizing) linear program in equational form, one of the following three possibilities is true: 1. The LP is infeasible. 2. The optimal value

More information

Generalizing Vertex Cover, an Introduction to LPs

Generalizing Vertex Cover, an Introduction to LPs CS5234: Combinatorial and Graph Algorithms Lecture 2 Generalizing Vertex Cover, an Introduction to LPs Lecturer: Seth Gilbert August 19, 2014 Abstract Today we consider two generalizations of vertex cover:

More information

7 Branch and Bound, and Dynamic Programming

7 Branch and Bound, and Dynamic Programming 7 Branch and Bound, and Dynamic Programming 7.1 Knapsack An important combinatorial optimization problem is the Knapsack Problem, which can be defined as follows: Given nonnegative integers n, c 1,...,

More information

Scheduling and (Integer) Linear Programming

Scheduling and (Integer) Linear Programming Scheduling and (Integer) Linear Programming Christian Artigues LAAS - CNRS & Université de Toulouse, France artigues@laas.fr Master Class CPAIOR 2012 - Nantes Christian Artigues Scheduling and (Integer)

More information

CSE 190, Great ideas in algorithms: Matrix multiplication

CSE 190, Great ideas in algorithms: Matrix multiplication CSE 190, Great ideas in algorithms: Matrix multiplication 1 Matrix multiplication Given two n n matrices A, B, compute their product C = AB using as few additions and multiplications as possible. That

More information

Hamming Codes. Chapter Basics

Hamming Codes. Chapter Basics Chapter 4 Hamming Codes In the late 1940 s Claude Shannon was developing information theory and coding as a mathematical model for communication. At the same time, Richard Hamming, a colleague of Shannon

More information

Stanford University CS261: Optimization Handout 7 Luca Trevisan January 25, 2011

Stanford University CS261: Optimization Handout 7 Luca Trevisan January 25, 2011 Stanford University CS261: Optimization Handout 7 Luca Trevisan January 25, 2011 Lecture 7 In which we show how to use linear programming to approximate the vertex cover problem. 1 Linear Programming Relaxations

More information

Discuss the size of the instance for the minimum spanning tree problem.

Discuss the size of the instance for the minimum spanning tree problem. 3.1 Algorithm complexity The algorithms A, B are given. The former has complexity O(n 2 ), the latter O(2 n ), where n is the size of the instance. Let n A 0 be the size of the largest instance that can

More information

Introduction to Linear Programming.

Introduction to Linear Programming. Chapter 1 Introduction to Linear Programming. This chapter introduces notations, terminologies and formulations of linear programming. Examples will be given to show how real-life problems can be modeled

More information

SOLUTIONS TO PROBLEM SET 6

SOLUTIONS TO PROBLEM SET 6 SOLUTIONS TO PROBLEM SET 6 18.6 SPRING 16 Note the difference of conventions: these solutions adopt that the characteristic polynomial of a matrix A is det A xi while the lectures adopt the convention

More information

Inverse Linear Programming

Inverse Linear Programming Inverse Linear Programming S. Dempe 1 and S. Lohse 2 1 Technical University Bergakademie Freiberg, Department of Mathematics and Computer Sciences, Akademiestr. 6, 09596 Freiberg, Germany, corresponding

More information

basic feasible solutions (BFS): a basic solution that is feasible. That is Ax = b, x 0 and x is a basic solution.

basic feasible solutions (BFS): a basic solution that is feasible. That is Ax = b, x 0 and x is a basic solution. EMIS 3360: OR Models The Simplex Method 1 basic solution: For a system of linear equations Ax = b with n variables and m n constraints, set n m non-basic variables equal to zero and solve the remaining

More information

Linear Programming: Geometry, Algebra and the Simplex Method

Linear Programming: Geometry, Algebra and the Simplex Method ISyE Lecture Notes c Shabbir Ahmed Linear Programming: Geometry, Algebra and the Simple Method A linear programming problem (LP) is an optimization problem where all variables are continuous, the objective

More information

THE MIXING SET WITH FLOWS

THE MIXING SET WITH FLOWS THE MIXING SET WITH FLOWS MICHELE CONFORTI, MARCO DI SUMMA, AND LAURENCE A. WOLSEY Abstract. We consider the mixing set with flows: s + x t b t, x t y t for 1 t n; s R 1 +, x R n +, y Z n +. It models

More information

Applied Algorithm Design Lecture 5

Applied Algorithm Design Lecture 5 Applied Algorithm Design Lecture 5 Pietro Michiardi Eurecom Pietro Michiardi (Eurecom) Applied Algorithm Design Lecture 5 1 / 86 Approximation Algorithms Pietro Michiardi (Eurecom) Applied Algorithm Design

More information

If we apply Gaussian elimination then we get to a matrix U in echelon form

If we apply Gaussian elimination then we get to a matrix U in echelon form 5. Gauss Jordan Elimination Gauss Jordan elimination is very similar to Gaussian elimination, except that one keeps going. To apply Gauss Jordan elimination, first apply Gaussian elimination until A is

More information

Discrete Optimization 2010 Lecture 4 Minimum-Cost Flows

Discrete Optimization 2010 Lecture 4 Minimum-Cost Flows Discrete Optimization 2010 Lecture 4 Minimum-Cost Flows Marc Uetz University of Twente m.uetz@utwente.nl Lecture 4: sheet 1 / 31 Marc Uetz Discrete Optimization Outline 1 Remarks on Max-Flow and Min-Cut

More information

Matrix techniques for strongly regular graphs and related geometries

Matrix techniques for strongly regular graphs and related geometries Matrix techniques for strongly regular graphs and related geometries Supplement to the notes of W. Haemers Intensive Course on Finite Geometry and its Applications, University of Ghent, April 3-14, 2000

More information

POL502: Linear Algebra

POL502: Linear Algebra POL502: Linear Algebra Kosuke Imai Department of Politics, Princeton University December 12, 2005 1 Matrix and System of Linear Equations Definition 1 A m n matrix A is a rectangular array of numbers with

More information

On bilevel machine scheduling problems

On bilevel machine scheduling problems Noname manuscript No. (will be inserted by the editor) On bilevel machine scheduling problems Tamás Kis András Kovács Abstract Bilevel scheduling problems constitute a hardly studied area of scheduling

More information

Lecture 9 and 10: Iterative rounding II

Lecture 9 and 10: Iterative rounding II Approximation Algorithms and Hardness of Approximation March 19, 013 Lecture 9 and 10: Iterative rounding II Lecturer: Ola Svensson Scribes: Olivier Blanvillain In the last lecture we saw a framework for

More information

Vector Spaces: Theory and Practice

Vector Spaces: Theory and Practice Chapter 5 Vector Spaces: Theory and Practice So far, we have worked with vectors of length n and performed basic operations on them like scaling and addition Next, we looked at solving linear systems via

More information

Cofactor Expansion: Cramer s Rule

Cofactor Expansion: Cramer s Rule Cofactor Expansion: Cramer s Rule MATH 322, Linear Algebra I J. Robert Buchanan Department of Mathematics Spring 2015 Introduction Today we will focus on developing: an efficient method for calculating

More information

56:272 Integer Programming & Network Flows Final Exam -- Fall 99

56:272 Integer Programming & Network Flows Final Exam -- Fall 99 56:272 Integer Programming & Network Flows Final Exam -- Fall 99 Write your name on the first page, and initial the other pages. Answer all the multiple-choice questions and X of the remaining questions.

More information

M AT H E M AT I C S O F O P - E R AT I O N S R E S E A R C H

M AT H E M AT I C S O F O P - E R AT I O N S R E S E A R C H A N D R E W T U L L O C H M AT H E M AT I C S O F O P - E R AT I O N S R E S E A R C H T R I N I T Y C O L L E G E T H E U N I V E R S I T Y O F C A M B R I D G E Contents 1 Generalities 5 2 Constrained

More information

Rank. Rank. Definition. The set of all linear combination of the row vectors of a matrix A is called the row space of A and is denoted by Row A

Rank. Rank. Definition. The set of all linear combination of the row vectors of a matrix A is called the row space of A and is denoted by Row A Rank Rank he rank of a matrix is the maximum number of independent rows (or the maximum number of independent columns) Definition he set of all linear combination of the row vectors of a matrix A is called

More information

On the distribution of Traveling Salesman Problem solutions among the ranked Linear Assignment alternatives

On the distribution of Traveling Salesman Problem solutions among the ranked Linear Assignment alternatives On the distribution of Traveling Salesman Problem solutions among the ranked Linear Assignment alternatives The Linear Assignment Problem is considered one of the fundamental combinatorial optimization

More information

MATH 2030: EIGENVALUES AND EIGENVECTORS

MATH 2030: EIGENVALUES AND EIGENVECTORS MATH 200: EIGENVALUES AND EIGENVECTORS Eigenvalues and Eigenvectors of n n matrices With the formula for the determinant of a n n matrix, we can extend our discussion on the eigenvalues and eigenvectors

More information

Singular Value Decomposition

Singular Value Decomposition Chapter 4 Singular Value Decomposition In order to solve linear systems with a general rectangular coefficient matrix, we introduce the singular value decomposition. It is one of the most important tools

More information

Systems of Linear Equations

Systems of Linear Equations Systems of Linear Equations Beifang Chen Systems of linear equations Linear systems A linear equation in variables x, x,, x n is an equation of the form a x + a x + + a n x n = b, where a, a,, a n and

More information

MATHEMATICS 694 Midterm Problems

MATHEMATICS 694 Midterm Problems MATHEMATICS 694 Midterm Problems. Let A be and m n matrix, b R m, and K R n a convex set. Prove or disprove that the set {x R n Ax b} K is a convex subset in R n. 2. Let f : R n R satisfies the inequality

More information

Chapter 11. Approximation Algorithms. Slides by Kevin Wayne Pearson-Addison Wesley. All rights reserved.

Chapter 11. Approximation Algorithms. Slides by Kevin Wayne Pearson-Addison Wesley. All rights reserved. Chapter 11 Approximation Algorithms Slides by Kevin Wayne. Copyright @ 2005 Pearson-Addison Wesley. All rights reserved. 1 Approximation Algorithms Q. Suppose I need to solve an NP-hard problem. What should

More information

Definition of a Linear Program

Definition of a Linear Program Definition of a Linear Program Definition: A function f(x 1, x,..., x n ) of x 1, x,..., x n is a linear function if and only if for some set of constants c 1, c,..., c n, f(x 1, x,..., x n ) = c 1 x 1

More information