SCHAUM’S SOLVED PROBLEMS SERIES 3000 SOLVED PROBLEMS IN LINEAR ALGEBRA by Seymour Lipschutz, Ph.D. Temple University SCHAUM’S OUTLINE SERIES McGraw-Hill New York San Francisco Washington, D.C. Auckland Bogota Caracas Lishon London Madrid Mexico City Milan Montreal New Delhi San Juan Singapore Sydney Tokyo Toronto' eee Lipschutz, Ph... Professor of Mathematics at Temple iniversity, ‘Dr. Lipschutz eared the Ph.D. st the Courant Institute of Mathematical Sci- ences of New York University. He has authored a number of volumes in the Schaum's Outline Series, including the internationally used LINEAR: ALGE- BRA Project supervision was done by The Total Book. Library of Congress Cataloging-in-Publication Data Lipschutz, Seymour. '3000 solved problerns in linear algebra. (Schaumn’s solved problems series) 1, Algebras, Linear—Problems, exercises, ete, 1. Title. II, Title: Three thousand solved ia linear algebra. I. Series. QA184.5.156 1988 512'.9 —B8-B94d, ISBN 0-07-038023-6 67.89 10 VHG 05 04 03 02 Q-07-038023-b Copyright © 1989 by The McGraw-Hill Companies, Inc. All rights reserved. Printed in the United States of America. Except as permitted under the United States Copyright Act of 1976, no part of this publication may be reproduced or distributed in any form or by any means, or stored in a data base or retrieval system, without the prior written permission of the publisher, McGraw-Hill mud ‘A Division of The MoGrave- Fill Companf & {Ohapter 1 Chapter 2 Chapter 3 ‘Chapter 4 AY Chapter 5 Chapter 6 Chapter 7 Chapter 8 CONTENTS VECTORS IN R" AND C” 1.1 Vectors in R" / 1.2 Vector Addition and Sealar Multiplication / 1.3 Summation Symbol / 1.4 Dot (Inner) Product / 1.5 Norm (Length) in R" / 1.6 Distance, Angles, Projections / 1.7 Orthogonality / 1.8 Hyperplanes and Lines in R" / 1.9 Complex ‘Numbers ( 1.10 Vectors in C” / 1.11 Dot (Inner) Products in C* / 1.12 Cross Product MATRIX ALGEBRA 2.1 Matrices. / 2.2 Matrix Addition and Scalar Muitiptieation / 2.3 Matrix Mulkiplication / 2.4 Transpose of a Matrix / 2.5 Elementary Row Operations; Pivots / 2.6 Echelon Matrices, Row Reduction, Pivoting | 2.7 Row Canonical Form, Gauss Elimination ( 2.8 Block Matrices SYSTEMS OF LINEAR EQUATIONS 3.1 Linearity, Solutions / 3.2 Linear Equations in One Unknown / 3.3 Linear Equations in Two Unknowns / 3.4 One Equation in Many Unknowns J 3.5m Equations in.» Unkniowns / 3.6 Systems in Triangular and Echelon Form / 3.7 Gaussian Elimination ( 3.8 Systems of Linear Equations in Matrix Form ( 3.9 Homogeneous System / 3.10 Nonhomogencous and Associated Homogeneous Systems / 3.11 Systems of Linear Equations as Vector Equations SQUARE MATRICES 4,1 Diagonal, Trace / 4.2 Idemtity, Scalar, and Diagonal Matrices / 4.3 Algebra of Square Matrices. Commuting Matrices / 4.4 Powers of Matrices / 4.3 Square Matrices as Functions / 4.6 Inventible Matrices / 4,7 Elementary Matrices / 4.8 Column Operations. Matrix Equivalence / 4.9 Upper Triangular and Other Special Matrices | 4.10 Symmetric Matrices / 4.11 Complex Matrices / 4,12 Hermitian Matrices / 4.13 Orthogonal Matrices / 4.14 Unitary Matrices / 4.15 Normal Matrices / 4,16 Square Block Matrices DETERMINANTS. 5.1 Determinant Function, Determinants af Order One | 5.2 Determinants of Order Two | 5.3 Determinants of Order Three / 5.4 Permutations / 5.5 Determinants of Arbitrary ‘Onder / 5.6 Evaluation of Determinants; The Laplace Expansion / 5.7 Clagsical Adjoint / 5.8 Volume as a Determinant J 5.9 Cramer's Rule. Block Matrices / 5.10 Submatrices, General Minors, Principal Minors / 5.11 Miseellaneaus Problems: ALGEBRAIC STRUCTURES 6.1 Sets, Mathematical Induction, Produet Sets / 6.2 Relations | 6.3 Partitions and Equivalence Relations / 6.4 Operations and Semigroups / 6.5 Groups and Subgroups / 6,6 Normal Groups, Factor Groups. Group Homomomphisms / 6.7 Rings and Ideals / 6.8 Integral Domains, PID, UFD / 6.9 Fields VECTOR SPACES AND SUBSPACES 4.1 Vector Spaces / 7.2 Subspaces of Vector Spaces | 7.3 Linear Combinations, Linear Spans / 7.4 Spanning Sets, Generators / 7.5 Row Space of a Matrix / 7.6 Sums and Direct Sums LINEAR DEPENDENCE, BASIS, DIMENSION 8.1 Elementary Properties of Linear Dependence and Independence / 8.2 Linear Dependence of Vectors / 8.3 Theorems on Bases and Dimension / 8.4 Bases and Dimension / 8.5 Dimension and Subspaces / 8.6 Rank of a Matrix / 8.7 Applications to Linear Equations / 8.8 Sums, Direct Sums, Intersections / 8.9 Coordinates Bw 51 80 ua 143, 173 196 iitiv G CONTENTS Chapter 9 Chapter 10 Chapter 11 Chapter 12 Chapter 13 Chapter 14 Chapter 15 Chapter 16 Chapter Chapter 18 Chapter 19 MAPPINGS 9. Mappings, Functions / 9.2 Real-Valued Functions / 93 Vector-Valued Mappings / 9.4 Composition of Mappings / 9.5 One-to-One, Onto, and Invert LINEAR MAPPINGS 10.1 Linear Mappings / 10,2 Properties of Linear Mappings / 10.3. Kemel snd Image of a Linear Mapping 10.4 Computing the Kemel and Image of Linear Mappings / 10,8 Singular and Nonsingular Linear Mappings, Isomorphisms / 110.6 Applications to Geometry, Conver Sets SPACES OF LINEAR MAPPINGS 11.1 Operations with Linear Mappings / 11.2. Vectar Space of Linear Mappings / 11.3 Algebra of Linear Mappings (11.4 Invertible Operators / 11.5. Linear Mappings and Systems of Linear Equations MATRICES AND LINEAR MAPPINGS 12.1. Mairix Representation of a Linear Operator (12.2. Matrices and Linear Operators on Ry 12.3 Matrices and Linear Mapping Operations / 12.4 Matrices and Linear Mappings CHANGE OF BASIS, SIMILARITY 13.1. Change-of-Basis (Transition) Matrix / 13.2 Change of Basis and Linear Operators 13.3 Similarity and Similarity Transformations / 13.4 Trace and Determinant of Linear Gperaiors / 13.5 Change of Basis and Linear Mappings INNER PRODUCT SPACES, ORTHOGONALITY 141 Inner Product Spaces / 14.2 Properties of Inner Products and Normy 14.3 Cauchy- Schwara Inequality and Applications | 14.4 Onhogonality, Orthogonal Complements 14.5 Orthogonal Sets and Buses / 14.6 Onhogonal Matrices / 14,7 Projections, Gram Sehumidt Algorithm, Applications / 14.8 Inner Products and Positive Definite Maitices / 14,9 Complex Inner Product Spaces / 14.10 Nomed Vector Spaces POLYNOMIALS OVER A FIELD 15:1 Ring of Polynomials / 15.2 Euclidean Algorithm, Roots af Polynomials / 15.3 Principal Ideat Domain, Unique Factorization Domain / 15,4 Polynomials with Matrices and Linear Operators EIGENVALUES AND EIGENVECTORS, DIAGONALIZATION 16.1. Characteristic Polynomial, Cayley-Hamilton Theorem / 16.2 Eigenvalues and Eigenvectors / 16.3 Computing Eigenvalues and Eigenvectors, Diagonalizing Matrices / 16.4 Minimum Polynomial / 16.5 Further Properties of Eigenvalues and Eigenvectors CANONICAL FORMS 17.1 Invariant Swbspaces / 17.2 Direet Sums, Projections / 17.3 Invariant Direst-Sum Decompositions / 17.4 Nilpotent Operators and Matrices / 17.5 Jordon Canonical Form / 17.6 Quotient Spuces and Triangular Farm / 17,7 Cyclic Subspaces, Zy,T) | 17.8 Rational Canonical Form LINEAR FUNCTIONALS AND THE DUAL SPACE 18.1 Linear Functionals and the Dual Space / 18.2 Dual Basis / 18.3 Second Dual Space, Natural Mapping / 18.4 Annihilators / 18.5 Transpose of a Linear Mapping BILINEAR, QUADRATIC, AND HERMITIAN FORMS 19.1 Bilinear Forms / 19.2 Bilinear Forms and Matrices | 19.3 Alternating Bilinear Forms { 19.4 Symmetric Bilinear Forms / 19,5 Quadratic Forms / 19.6 Real Symmetric Bilinear and Quadratic Forms, Law of Inertia / 19.7 Orthogonal Diagonalization of Real Quadratic Forms 19.8 Hermitian Forms / 19,9 Multilinearity and Determinants 227 247 267 282 302 319, 355 364 416 a6Chapter 20 Chapter 21 CONTENTS 0 v LINEAR OPERATORS ON INNER PRODUCT SPACES 451 20.1 Adjoint Operators / 20.2 Self-Adjoint Operators, Symmetric Operators / 20.3 Orthogonal and Unitary Operators { 20.4 Positive and Positive Definite Operators / 20.5 Normal Operators. / 20.6 Spectral Theorem APPLICATIONS TO GEOMETRY AND CALCULUS 467 21.1 Vector Notation in R? / 21.2 Planes. Lines. Curves. and Surfaces in R' / 21.3 Scalar und Vector Fields / 21.4 Del Operator V and Gradient, Divergence and Curl / 21.5 Differential EquationsTo the Student This collection of thousands of solved problems covers almost every type of problem which may appear in any course in linear algebra. Moreover. our collection includes both computational problems and theoretical problems (which involve proofs). Each section begins with very elementary problems and their difficulty usually increases as the section progresses, Furthermore, the theoretical problems involving proofs normally appear afler the computational problems, which can thus preview the theory. (Most students have more difficulty with proofs.) Normally, students will be assigned a textbook for their linear algebra course, The sequence of our chapters follows the customary order found in most textbooks (although there may be some discrepancies). However, whenever possible, our chapters and sections have been written so that their order can be changed without difficulty and without loss of continuity. The solution to each problem immediately follows the statement of the problem. However, you may wish to try to solve the problem yourself before reading the given solution. In fact, even after reading the solution, ‘you should try to resolve the problem without consulting the text. Used thus, 3000 Solved Problems in Linear Algebra can serve as 4 supplement to any course in linear algebra, or even as an independent refresher course. vii1 Aa 12 13 14 1s Lr CHAPTER 1 Vectors in R” and C” VECTORS IN R" ‘A vector 1 in the vector space R is an ordered set of n real numbers: =(a,,4;,---,@,)- ‘The real number a, is called the kth component or coordinate of 1. Compare this with the definition of a vector in physies I Physics. defines a vector u to be a quantity with magnitude: and direction, represented by means of an arrow or direeted line segment emanating from a reference point O. In Fig. 1-I a planar vector w is identified with the coordinates of its endpoint, P(4,2). That is, w=(4,2)—in accord with the above definition of @ vector in R° Fig. 1 State the difference between a row vector and a column vector, FA cotumn vecior wis @ vector whose components are arranged vertically A. row vector is a vector whose components are arranged horizontally, as in Problem 1.1. [In this chapter, vectors will normally be written as row yectors.] To which vector space R" does each vector belong? (a) (3,-2:5,8) — (B) (B,6+2i) (wr, 2, Say FT (@) RB since there are four components. (6) None, since not all the components are reat numbers, (0) R’ [and Sa are real numbers} For vectors u and v in R", when is w=? I The vectors u and v are equal if and only ifthe corresponding components afe equal Let 1,2,3), u,= (2.3.1), wy=(1,3,2), w= (2,3,1) be vectors in R® Which of the vectors, if any, are equal? FT Only w, and u, are componenwise equal.2 0 CHAPTER 1 16 Find x and y if (x,3)= £49), I Since the two vectors are equal, the corresponding components are equal to each other: ¥=2, 3=x+y. By subtraction, y=1 1.7 Define the zero vector in R”. FT 0=(,0,...,0) 1 Find x and y if u=(x+y,x—3) is the zero vector. 0 and x=3=0. The 1 First set each component of u equal 10 0 to obtain the equations x +) second equation gives x=3; then the first equation gives y= —3 19 Define the negative, ~u, of a vector w= (a,,4,e---4 a4) im RY ty Age) 1.40 Find the negatives of (a) u= (3, =8,4), (@) v=(-42m0), (@) 0=(0,0,0,0), T Take the negative of each component of the vector, (a) —u=(~3, ~(~5), ~(-8), ~4)= 3, 5,8, 4); ) v= (4, 2, 0); @ -0=(-0, ~0)=(0, 0,0, 0)=0. Li Show that —(-u)=u for any vector w in R". HW Let w=(a,,@....,4,). Then, because —(-a)=a for any scalar a in R, = (20) = =a, Bag.) = Cy tgs oo) 1,2 VECTOR ADDITION AND SCALAR MULTIPLICATION LAD Let w= (uj.t,..-.u,) and v=(0,,0,,...,0,) be any veetors in R’. Then the sun, w+v, is the vector Ut OR (U, A Diy tease thy FQ) Show how this definition conforms to the addition of veetors in physics I in physics, vector w+y is the diagonal of the parallelogram formed by yectors w and y, as pictured in Fig. 1-2(a). Now let @ be chosen as the origin of a coardinate system (of R7), and suppose (a, 5) and (6d) are the endpoints of u and v, respectively, as pictured in Fig. 1-2(b). ‘Then, using geometry, one can show that (a+c.6+d) will be the endpoint of u+v, On the other hand, adding corresponding components, we obtain (a,b) + (.d)=(a+ cb +a) Both definitions of addition give the same result, (ate,b+a) @ ) Fig. 1-2La 1.18. VECTORS INR’AND GC" [3 Compute: (a) (3, -4, 5, -6) + (1, 1,-2.4), () (1.2, -3) + (4. -5). F(a) Add corresponding components: (8,-4.5, -6)+ (1,1, -2,4)=(B+ 1, -441, 5-2, 6+ 4) =(4, -3,3,-2) (®) The sum is not defined, since the vectors have different numbers of components 7 -3 V Compute: (a) (-)-(-1), @) (3)« = s}+{-1}. @ (3]+(%) 1 (@) Add corresponding components: rT =3) Toe 4 parla) a Vs) Vaas) V9 Bmoatigd er eat cating Ue erec eoete Be eye) ary vers cali eee ony eee Pe need a ety [scalar] product ku is the vector fem Chi, tds, Rt.) Show how this definition conforms to the scalar multiplication of vectors in physics I Physies defines the product of a real number k and a vector [arrow] u, say with reference point Q, to be the vector (i) whose magnitude is equal to the magnitude of u multiplied by kl; (fi) whose direction is that of wit &20, but is opposite to wif, k<0. This is pictured in Fig. 1-3(a). Now fet O be chosen as the origin of a coordinate system [of R°], and suppose (a, 6) is the endpoint of u, as pictured in Fig 1-3(6). Then, using geometry, one can easily show that the endpoint of ku is (Ka,kb). On the other hand, our definition gives (a,b) = (ka, kb), the same result. ee) s (a,b) o (6) Fig. 1-3 Compute: (a) —3(4,-5,-6), (6) 46,7, -8) I (a) Muktipty each component by the scalar: —3(4, ~5, -6)=(-12, 15, 18). (6) Either multiply each component by ~1 or take the negative of each component; either way we obtain (6, ~7, 8). = Compute: (a) { 3). ) -2(_2). 4 1 Multiply each component by the scalar; =3) 7-10 1) (718 @ { 3)-( i) 6) -2(_$)= (49): )-( (0-8 Define the difference, u—v, of the vectors w and v in R”. I ‘The difference is obtained by adding the negative: u—v = 0+ (~v).119 1.20 121 1.22 123 1.25 CHAPTER 1 Compute: (a) (3, -5.6,8)~ (1,-7.9). @ (_$)-(_3) I First find the negative of the second vector and then add. (@)_ G,5,6,8)- (4,1, -7,9)=(3, ~5,6,8) + (-4, =1,7, -9) = (1, -6, 13, <1) 6)_( 2)_( 6), (-2\_(4 w (3)-(2)=(4)+(9)-@) Let M1) v= (-3,0,4), w= (0,5, -8). Find (a) 3u—4u, (b) 2u+30=5w. # First perform the scalar multiplication and then the veetor addition. (a) 3u~4v =3(2, -7, 1) ~4(—3, 0, 4) = (6, 21,3) + (12, 0, -16) = (18, -21, -13) (6) 2u+3v —5w =2(2, ~7,1) +3(-3,0, 4) -9(0, 5, 8) 4, —14, 2) + (-9.0, 12) + (0, —25, 40) 4-940, Id +0 25,2412 + 40)= 5, ~39, 54) Let w= 1,4) and 0 =(7,1,-3,6) in‘ Find (a) +0, (6) 4u,(c) 2u-3v, Fa) ut v=(347,-241,1-3,4+6)= (10, -1,-2,10); (6) 4u=(4-3,4-(-2),4:1,4-4) = (12, -8,4, 16); (c) 2u ~4u = (6, 4, 2,8) + (-21, -3,9, —18) =(-15, —7, 11, -10). Find xx and y if (4, y) = x(2,3), 4G) Multiply by the scalar x to obtain (4, y)=a(2, 3)=(2a, 3x). (ii) Equate correspond components: 4=2x, y=3z, (ii) Solve linear equations for x and y: x=2 and y=6. Write w=(1,9) asa linear combination of the vectors. w= (1,2) and v=(@,—1). 4 We want to find scalars x andy such that w= aut ye; ie. (1,9) = x(1, 2) # y(3, =1) = (, 28) + (By, -y} = (6+ 3y,28- y) Equality of corresponding components gives the two equations etBy=1 (Qe y=9 To solve the system of equations, multiply the first equation by 2 and then add it to the second equation to obtain. ~7y=7, or 1. ‘Then substitute y=—1 in the first equation to obtain x —3 or x=4. Accordingly, w=4u—v. Write v= (2,-3.4) as a linear combination of vectors, = (1,11), = (1.1.0), ue (1,0,0). I Proceed as in Problem 1.23, this time using column representations. (2)-4i)-Ai)-G-0)-0)-G)-(2) Now set corresponding components equal to each other gtytre2 xty=-30 xed To solve the system of equations, substitute x=4 into the second equation to obtain 44+ y= or y=~7. Then substitute into the first equation to find 2=5. Thus v=4u,—Tu,+5u, Write w=(1,2,-5) as a linear combination of 4, = (1.—! Wy w= (21,4) and wy= (11,3). 1 First multiply by scalars x, y,z and then add: aa) 1g Components equal to each other to obtain the system: xt2yte=l sxtytr=2 9 -e4+day+321.26 127 1.28 1.29 1.30 1.31 132 VECTORS INR"AND CG” J 5 Add the first equation to the second equation to obtain (i) 3y+22=3. Now add the frst equation to the equation to obtain 6y+4z=—4, or (il) 3y+22=~-2, Equations (i) and (ii) are obviously inconsistent. In other words, w is not a Kinear combination of e,,#,,4,- In Problems 1261.36, u,v, w denote vectors in R"; kk” denote scalars in R; u,,0, and w, denote the ith components of u, v and w, respectively Prove that (1+ 0) + w=u + (0+ w). 1 By definition [Problem 1.12], u, +0, is the ith component of +0 and so (u,+u,)+W, is the ith component of (u+v)+¥, On the other hand, v,+w, is the ‘th component of v+w and so u,+(¥,+w,) is the ith component of u+(u+%). But u,,v,, and w, are real numbers for which the associative law holds; that is, (a +o) +m =m tle +m) an) Accordingly, (u+v)+w=w+(v+w), since their corresponding components are equal Prove that u+0= Fut 0=(u,u,,...,u,)+(0,0,-..,0)=(u, +0,u,40,....0,)=(u, soy) Prove that w+ (—w) =0. Be (-u) = (Ups ttyv een) + Attys te (Uy = ty My =U, Prove that ut+u=vtu J By definition [Problem 1.12]. «, +0, is the ith component of u+o, and v,+u, is the ith component of v+u, But u, and v, are real numbers for which the commutative law holds, that is, utusutu, Gls ay Hence u+u=v+u, since their corresponding components are equal. Prove that k(w +p) = ku ko. I Since u, +0, is the ith component of uv, k(u+2,) is the ith component of Au +0). Since Jeu, and kv, are the ith components of ku and ku respectively, ku, + ku, is the ith component of ku + ko. But &, «, and v are real numbers; hence Ku; + 0) = kag + ho, (EA) Thus ku tv) = ku + kv, as corresponding components are equal Prove that (K+ k')u = ku + k'u, FF By definition [Problem 1.15], (&+'}u, is the ith component of the vector (k + K')u Since Au, and k’te, are the ith components of ku and k's, respectively, Au, +k’, is the ith component of tut k’u, But k, #’, and u, are real mumbers, hence (e+ kyu, = ut ku, (i n) Thus (K-+A')u= ku + k’u, as corresponding components are equal. Prove that (Kk')u= k(k'u),6 2 CHAPTER F Since k’u is the ith component of k’w, k(k'x,) is the #th component of A(&'u), But (Kk')u, is the dtl component of (kk"}w and, since k, k’ and u, are real numbers, (KR Ju, = K(k’) (E12. 0) Hence (kk')u= k(Kk’u), a8 corresponding components are equal. 1.33 Prove that 1ite= u Bete Maytag) = (ae Neg Mg) = (uy tgs tg) te Show that Ow=0 for any vector w Be Method 1. Que = O(4,. 1435. +5 dy) = (Dit Ott, Ont) = (0,0... +50) =0. Method 2. By Problem 1.31, Ou=(0-+0)u=0u+0u, Adding —(0u) to both sides gives Ov + (—(0v)) = (Ov + Ov) + (-(Ov)) = 00+ (Ov + (—(0v))) {by Problems 1.28, 1.26] 0=0u+0 [by Problem 1.28] 0=0v [by Problem 1.27] [Note that Method 2, although lengthy, docs not explicitly use coordinates, ] 135 Show that 4=@ for any scalar &, FH Method 1. kO= K(0,0,...,0)=(K-0, 40,0... k-0) =(0,0,..., 0)=0. Method 2, By Problem 1.34, 0+0=0. Then, by Problem 1.31, kO=k(0+0)=k04K0. Adding (KO) to both sides leads t0 the required result, as in Problem 1.34. 1.36 Show that (—I)ue= I From the previously proved properties, 2 w+ (-u)= = Ou=(14(=1)e= le + (1u= 0s (De ‘and the result follows upon adding ~w to both sides, 1.3 SUMMATION SYMBOL 1.37 Let fie) be an algebraic expression involving an integer variable k, Define the expression S, where nz1, [Here 1 is called the lower limit, n is called the upper limit, and the Greek letter sigma functions as the summation symbol] TS,=f1) + f2)+-+-+fla-1) + fin). From this definition, itis obvious that, for at flay. 1.38 Suppose n, and n, are any integers such that m,=n,. Define } f(k). VS fk) = fo) + fl, +1) + fl, +2) + +f). For n,, and that the theorem holds for _n-1; ice., S1t+e@1= 10+ E 6) Then [see Problem 1.37], 2000+ 8@l= 2 -[3n0 + nolo Thus the theorem is proved. , fi) + 3B ah + fod + a0) L/h) + g(AD] + Lf) + gt) tt) + an] 3 4) + 3 8 Prove: Ye flk)=c S fl): I The proof is immediate from the distributive law, a(b+e++--)=ab+ac+-+-, for real numbers 1.4 DOT (INNER) PRODUCT 14s Suppose w= (W,, te) -- 5g) and v= (vy, 0,,...,0y) are vectors ia R". The dot [or irmer or scalar] product of w and v, denoted by uv, is defined to be the scalar obtained by multiplying corresponding components and adding the resulting products: wun > wy Show how this definition conforms to that used in physics FL Suppose that u and v are vectors (arrows) in R” emanating from the origin O, as pictured in Fig. 1-4 ‘Moreover, suppose that P(a,. 4s, a,) and Q(b,, Bs, bs) are the endpoints of u and v, respectively, and let 6 denote the angle between u and v, Physies defines the dot product as u-v= [uj |v|cos@1.46 1.47 1.48 149 1.50 181 CHAPTER 1 0Cb,, 8...) Fig. 14 Now ltadta, bi + 62463, and PQ? =(a, — hy + (a, b,)* + (a, 8, = (aj + a3 + a) + (by + 3 + 63) —2(a,b, +a,b, + @,b,) =|ul? + |v? -2D uo, But, by the law of easines, PQ* = |ul’+|¥|?—2|ul|v| cos @: therefore, ful|vjeos@= ¥ u,u,, and the two definitions agree. m Compute wv, where w=(2,-3,6) and v=(8,2,-3) T Multiply corresponding components and add: «+ » = (2)(8) + (—3)(2)+ (6)(—3) Compute w+v, where u=(1,-8,0,5) and v= (3,64) I The dot produet is not defined between vectors with different numbers of components, Compute usu, where u=(3,-5,2,1) and v= (4,1,-2,5). I Multiply cosresponding components and add: w+ v = (3)(4) + (~8)(1) + (2)(-2) + (1)(3)=8. Compute wv, where w=(1.-2,3,-4) and 0 =(6,7.1,-2). I Muliply corresponding components and add: 1-0 =(1)(6) + (—2)(7) + (8)(1) + (-4)(-2) Suppose 1 G21), v=G,-3.4), w=(1,6,-7). Find: (@) (ute)-m, (6) ucw tow, F(a) Fitstcaleulate u+v by adding comresponding components: w+v=(3+5,2—3,1+4)= (8.-1,5). Then compute the dot product (u + 0)+w by multiplying corresponding components and adding: (w+ v)-w = (8)(1) + (—1}(6) + (5)(-7) = —33 (®) First find usw =3412-7=8 and vew=S-18-28=—41. Then usw =8-41=-33. [For an explanation of the agreement between (a) and (b), see Problem 1.52,] Let u 3,4), v=(5.-6,7.8); and k=3. Find: (a) k(u-v), (6) cku)-0, (6) we(kv), HG) First find wiv =5—10421-2=-18. Then Mu-0)=H-18)=—54 (b) First find ku = (3(1),3(2), 3(3), 3(-4)) = (3,6, 9, -12). Then (ku) (3)(5) + (6) 6) + (9)(7) + (=12)(8) = 15 = 36 + 63-96 = —54 (©) First find hv =(15, ~18,21,28). Then wey) = (1)(15) + (2)(-18) + (RL) + (—4)(24) = 15-36 + 63-96 = =54 [See Problems 1.53 and 1.$4.]1.82 1.53 1st 1.85 1.36 4st VECTORS INR”AND GC” 0 Prove that (w+ v)+w=u+w4v-w. 1B Using Probien 1.43, rb yeS oti gene Sam = Se Prove that (ku)+v = k(u+v) Using Problem 1.44, (hes = 3 (eyo, = 3 kewo,) = Eo, = ku 0) = oa ou Prove: u+w=0, and wew=0 iff [if and only if"] u=0. Since uf is nonnegative foreach i, u-wm ub tub toss + tO. Furthermore, w= 0 itt a= 0 for each i; that is, iff u=0. In Problems 1.56 and 1.57, u, stands for the ith vector in a set of vectors (not the ith component of a vector uw), Similarly for w, Let ys tgs yy adv be vertore in Rand let aj 25y>-4 be scalars in Re Prove ) (Sau)-v= 3 aco) > 0(D ug)=D atorus) In words: ‘Taking the dot product with a fixed vector v represents a linear operation om R°. F (@) The proof is by induction on p. ‘The case p=1 is true by Problem 1.53. Suppose p>1 and the theorem true for p—1; Le. (2 aay) -0= 3 atu) ‘Then, sing Protemns 1.37, 1-52, 1.53, and the above inductive hypothesis, we have (2 eats) -2= (3 aang) © + (aug) = 3 aul 0) + aglty 0) = Bagh (®) Using wv =u-u [Problem 1.54] and part (a), we have o-(3 aa)= (Bea) Zatu-o= Eaton) Let ye tye tgs Bas Bay +o By De Vectors R™ and let @,.o,. dps bys Baroy by be any scalars in R. Prove that ( em)-(3 bore) = BB adtey-e0d (bilimearity of she inner product) FT Using Problem 1.56 (Bau) (3 be) = Ba,[a-(3 re)]-S [3 ty-00] = S/S aca) 910 2 CHAPTER 1 1.5 NORM (LENGTH) IN R" 1.88 If root of ut; +1 Hq) is-a vector in R", the norm or lengit of u, denoted ull, is the nonnegative square llull= var Vid oul Fa ‘Show that the above definition of the norm of a veetor conforms to that of the length of a vector [arrow] in physics, 4 Let u be a vector [arrow] in the plane R? with endpoint P(a, 6), as pictured in Fig. 1-5. Then |a| and {| are the lengths of the sides of the right triangle formed by u and the horizontal and vertical directions. By the Pythagorean theorem, the length of wis {ul = Va"+ 6" This value is the same as the norm of u. Fig. 15 1.59’, together with the definitions of vector addition, scalar multiplication, and inner produet, is called real Euclidean n-space. Why? F According to Problem 1.45, two vectors u and © in the inner-product space R” may be termed Perpendicular if u-v=0, For such vectors, we have from Problem 1.57 e+ wl? = (atv) (ut v) = ere + OF OF De v= lull? + oll? which is the Pythagorean theorem, Singe this thearem is a consequence of Euclidean geometry, we call R” a “Euclidean” space. 1.60 Find |lull if w= (3, -12, 4) # First find |jul*=w-n by squaring the components of w and adding: jel? 94 146+ 16= 169. Then, hu = VIGO = 13, B+ (-12)° + (4) = 1.61 Find |jul| if »=(2, ~3,8,-3) Square each component of v and then add to obtain. |[ul/?= v+u: Nol? =2* +(=3) +8? + (-5)* = 4494 64+25= 102. Then |jo||= VIO. 1.62 Find |jwl| if w= (-3, 1, -2,4,-5), A |i? =(-3 + 1? + (-2)' +47 + (-5)7 041444 16+25=55; hence wl] = v3, 1.63 Determine & such that |jul|=V39 where u=(1, k, -2, 5). Flu? = 1° + + (29° +5?= 02490, Now solve k2430=39 and obiain k= 1.64 Give the definition of a unit veetor FA vector u is a unit vector if |ful|=1 or, equivalently, if u-w=1,1,65 1.66 1.67 1.69 1.70 VECTORS INR’ ANDG” 0 11 Let v be an nonzero vector. Show that eye fel Tell is a unit vector in the same direction as v. [The process of finding is called normalizing v.) The vector cis a unit vector, since 8-2=(Te9)- (Ton Tor is in the same ditection as v, since 0 is a positive scalar multiple of v. 1 = pee ‘Moreover, Normalize v=(12, -3.—4), A) First find |u|? =v+=12* #(-3y' + (~4y lel] = VIG5= 130 obtain =104+9+16= 169, Then divide each component of v by Normalize w= (4,-2, ~3,8) 1 First find [jul <9: w= + (2) + (3) 48" = 164 44 9468=93. Divide each component of w by 53 to obtain Normalize v= A Note that v and any positive multiple of v will have the same normalized form [see Problem 1.72]. Hence, first multiply v by 1210 “clear” fractions: 120 =(6,8,—3). Then ||L2v||"= 3646449 = 109. Accordingly, the required unit vector is a 1 fate to (-o oe Wei) Show that lu] =0, and jjuli-0 iff w=0. A Follows at once from Problem 1.55, Prove the Cauehy=Schwarz inequality: |u| (lull lvl, for arbitrary w and v in B® J) We shall prove the following stronger statement: Ju+v|= > [u,v] = lull ull. First, if w= 0 ‘or v=0, then the inequality reduces to 0<0=0 and is therefore true. Hence we need only consider the case in which uO and x0, ie., where |jull#0 and [oll *0. Furthermore, because usp] =[Z ua = [u,v ‘we need only prove the second inequality. ‘Now, for any real numbers x, yER, 0=(x—y)? 2ay ty? or, eq ayy exit yt @ Set x=|i,|/(full and y=|o,|/llel| in (2) to obtain, for any i, \u @42 17 in L733 1.74 CHAPTER 1 But, by definition of the norm of a vector, |lul]= 3 u?= Su? and fjel|=Se?=S |v. Thus summing (2) with respect to / and using |u,v,| = u,||v,|, we have 2 = luo Pll | Zhe pal | dol? Teel = Tha * or * tal * ei that is, Ela Tel etl Multiplying both sides by |Iu\[ Jul], we obtain the required inequality. Prove Minkowski's inequality: |ju-+ v|| wu — (wu AT Here u-w= aye, +ax0,+a5¢3 and v- wm bye, +b,c,+ b,c, Let (uew)u—(u+w)u = (X,,05,04); then B= (@,6, 4-456. + ,6)b, — (bye, + brea + byes)ay faze, + aye,)b, — (bye, + bse. ‘Similarly, 2% (x0, + 0303)bs—(b4e, +B 46s); #5 = (42, + 430), ~ (bye, + Byes)as Write (uxv) w= (yy yo 95): then 5, = a,b, )e, — (a,b, — a,b, )e, = asb,e, - a,b,¢, — a,b,c, + a,b,c, = bilaxc, + 4,0,)~a,(b,¢, + bes) Thus, y,=%,. Similarly, y,=¥, and y,=2, 1.169 Prove: ux is orthogonal to both wand v, i Us (UX 8) = a,(a,b, — ayb,) + 4,(@,b, — a,b.) + ayla,b, 2b) = a,0,b, ~ a,0,h, + agagb, — a,4,b, + 4,040, — 056, =0 Thus, uXv is orthogonal tow. Similarly, Xv is orthogonal to w. 1.170 Show that ux w=0 for any vector u, A By Problem 1.164, uxu=—(uxu). Hence uxu=0. 1.171 Show that two vectors in R? are linearly dependent if and only if their cross product is the zero vector. I Linear dependence of two vectors v and v means that either u= kv, for some scalar k, or v= lu, for some scalar /. Suppose, then, that w by Problems 1.165 and 1.170, ux v= (ky) x K(o x v)= 0-0, with the same result if v= lu. Conversely, suppose that uxv=0. If w w= kv, for k=O. If w#0, set w=ie in Probiem 1.168 to obtain then O=(uwo—rwue or v= ey 1.172 Find a unit vector w orthogonal to y=(1,3,4) and w=(2,-6,5), AP In view of Problem 1,169, first find ux w. ‘The array (5 § 3) shes uw (-15424,8+5,-0-6) 9, 13, =12) Now normalize vxw to get w= (9393, 13/V 5, —12/V39), Problems 1.173-1.176 refer to the points P,(1,2,3), P.(2,5,-1), and P(3, 3,1) in B®. 1.173 Find the directed line segment (vector) u from P, to P,- W a= Py P, = (2,5, 1) ~(1,2,3) = (1,3, 4) 1.174 Find the directed line segment (vector) v from P, to W v= P,P, (5,3, 1)-(1,2,3) =(4,1, -2),LaTs 1.176 aT 1.178 VECTORS INR” AND GC” J 27 Find a vector w normal to the plane H containing the points P,, P,, and P, HH contains the vectors u and v determined above, Hence uXw is normal to H. The array (1 22) ies wun (6, +4, 16424112) =(-2, -14, 10) Give an equation for the plane H of Problem 1.175, FF Use the point P,(1,2,3) and the normal direction w to obtain =2s=1)=1My=2)-11(2-3)=0 or De Fy + Me = 63 Prove Lagrange’s identity, ju oll? = (us u)(u- v) ~ (uy Hof w= (@y,43,03) and by, ))y then Ju x wl]? = (agbs —aba)® + (ashy — abs)* + (a,b — ab y)* @ (w+ W)(v+0)= (esa) = (a + a3 + 05 )(0; +624 .53)~ (ab, + 0,0, + 4,b,)° 2) Expansion of the right-hand sides of (1) and (2) establishes the identity. Show that |/u+u|] =|\ul] |u|] sin@, where @ is the angle between u and v. By Problem 1.79, w+v=juil|lul|cos@ Then, by Problem 1.177, ux olf? = lal lo [luc ol]? cos* @ = |e) lu]? 1. — cos? 8) = ffel|* Jul? sin? & Taking square roots gives us our result.CHAPTER 2 Matrix Algebra 2d 2 22 24 28 2.6 27 28 This chapter uses letters A, B, C,.... to denote matrices and lowercase letters a, b,¢, x, y,... to denote scalars, Unless otherwise stated or implied, scalars will be real numbers; in other words, the matrices will be over B MATRICES Bee ne ee ee me 45 6 A The rows of A are the horizontal lines of numbers; there are two of them: (1 2 3) and (4 5 6), ‘The columns of A are the vertical lines of numbers; there are three of them: yy pe (i) (3) - (2) ‘The size of A is 2x3 [read: 2 by 3], the number of rows by the number of columns. The notation A=(a,),,., [orsimply A=(a,)] is used to denote an mn matrix, What is the significance of the first subseripr, i, and the second subscript, j? H The scalar a, is the clement of A in row i and column j. ‘Thus, the first subscript tells the row of the element and the second subscript tells the column. Given the matrix A=(q,), find the location of (a) ays, (0) es, (€) a ans (@) @yy.-g H (a) a, lies in the third row and fifth column. (6) a, ,, appears in row 4 and column 12, [Note that here we need a comma to distinguish the subscripts.] (cl and (d) The scalars a,,,, and a,, _, cannot be elements of A, since subscripts in a matrix are, by convention, positive integers. 2 Given mattices A and B, when is A= H Two matrices are equal if only if they have the same size and corresponding entries are equal. (fy BH) (8 5) Find e, yz. (UES AEM a(t a) I Bquate comesponaing entries: ‘The solution of the system of equations is Which of the following matrices, if any, are equal? 41 (2 3 d=(5 5) 8244) F Although all four matrices are 2%2 and contain the scalars 1, ‘equal element by element o=( 3) 2, 3,4, no two of the matrices are The m%n zero matrix, denoted by 0, oF simply 0, isthe matrix whose clemicnts are all zero. Find yeti xty 243 Geisetel 1 set atl entries equal 0 zero to obiain the aystem xtym0 743=0 =0 2tw=0 The solution of the system is x yas,28 29 2.10 MATRIX ALGEBRA J 29 The negative of an mn matrix A=(a,) is the mxn matrix —A=(-a,). Find the negatives of 4-() $03) 8-(4 1) F Take the negative of each element: (08) Show that, for any matrix A, we have —(—A)= A. i --A)= SMe = (8 Iriin =A A matrix A with only one row is called 4 row matrix or a row vector and is frequently denoted by A=(a, a, °-> ay); we omit its first subscript since it must be one. Analogously, a matrix B with only one colurmn is called! a cofumn matrix. or a column vector and is frequently denoted by b ; : I Viewed as vectors in R?,u and v may be considered equal. However, as matrices, they cannot be equal, for they have different sizes 2.2 MATRIX ADDITION AND SCALAR MULTIPLICATION 2a 212 243 2.44 If A=(a,),, and B=(b,),,, are matrices of the same size, their sun is defined as A+ B= (a, * Bago Find the sum of 2.3 2 Fey eu 8) F Add corresponding entries _(1#3 -240 342) _( 4 -2 5 atan({") “ihr -eta)-(3 6 2) a) ee Find A+B if A=() 5 and B=(} 3) FH The sum is not defined, since the matrices have different sizes. Faeroe a “1 2) F Ada corresponding etements: Meee a laa d 440 543 64(-3). Pee) os ' oepa(ity WHER CAM MCN) gt ay30 0 245 246 218 219 2.20 2a CHAPTER 2 Redefine the negative of a matrix [Problem 2.8] in terms of matrix addition, T The negative of a given matrix A is the [unique] matrix whose sum with A is the zero matrix, that is, A+(-A)=0. [Note that this way of defining —A avoids reference to the clements of A.) TE A=(4y)y.. and isa scalar, the matrix kA=(ka,),,,, is called the product of A by the scalar k. Find 34’and 5, where T Multiply each entry by the given scal “1 3862) 30a \_ 7s <6 sd 35) sco) (a 15-18, vS1 8-2) -5:3 ) (ee 10 =e) “$4 35 =5+(-0))“\-20 -25. 30, 3a=(} 17 comme: (6) (3 4), w -2 3) 4 24 3-2 3-4) 6 12 iG) (3 (sc sa)-C5 4) TW f2yd (-2)-7 2-14 () =i )-(3 2 (2) p}=(-4 3) 0-1) (E20 -2)--n) Vo 2 The difference, A~B, of two matrices A and B of the same size is defined by A- B= A+(—B), Find A-B if s(t freien eC) ' aqamascny-() yet) 3s G2) Find 24-38, where a=({ 2 4 First perform the sealar multiplications, and then a matrix addition: De ey eae ee Mila og eit sean alay ee) {Note that we multiply B by ~3 and then add, rather than multiplying B by 3 and subtracting. This usually avoids errors] ), tnd 94 +48 -2¢ 4. First perform the scalar multiplications, and then the matrix additions “(7 3 0) sasae-ae-(¢ 12 _2)e(4 8 “Bae 2 8) 9 9 -1)/*\o -4 20)*\-z 2 2 madsen 2 2)=(4 S)(4, 24)2.23 224 2.25 MATRIX ALGEBRA [31 WF First write each side as a single matrix: Gr ak) -(eewai aves) Set corresponding entries equal t0 each other to-obtain the system of four equations, Sreaxtd are4 By=xty +6 2y= 64x Sr=ztw-1 ee 2z=w-1 3w= 2 +3 wed 2el, wes, ana C=(5 _3). Find a=(£ 2) such that 24=38-2c F siethod 1, First compute 38-2C: se? 213 0-04 2 2 6 Then set 2A =3B-2C; Gr ak)=(5 2) 13, 2y==10, 22 Equate corresponding entries: 2x and w=27/2; that is, (32 3 a-(0" ain Method 2. Apply Theorem 2.1 [proved in Problems 2.24-2.31] to obtain directly A=(3/2)8—C. Fing 24+5B, where A=(} _2) and B=(} > —$) I Akthough 2.4 and 5B are defined, the sum 24+ 5B is not defined since 2A and 5B have different sizes. Theorem 2.1: Let M be the collection of all mx matrices over a field K of scalars. Then for any matrices A=(a,), B=(B,), and C=(c,) in M, and any scalars k,, k in K, (i) (A+B)+C=A+(B4C) (WK (AFB)=RATKB (i) A+0= (i) +k A= KA + A (iit) A+(~A)=0 (Wii) Ok) = ky (ka) (is) A+B=B+A (wii) 1A= A Prove (i) of Theorem 2.1 H The jeenuy of A+B ix a,+b,; hence, (a,+5,)+¢, isthe jrentry of (A+B)+C, The jjenuy of B+C is b, +e,: hence, a,+(b, +6,) is the ipentry of A+(B+C), However, by the associative law of addition in K, ay + By) + €y = Gg + (Ou + My) Therefore, (A+B)+C and A+(B+C) have the same ijentries, and hence (A+B)+C=A+ (BO), of Theorem 2.1 Net te desea me riences AA 8 rt Nae: ts care fren eed ee A+O=A Prove (i32 2.36 22 2.28 2.29 2.30 231 2.32 2.33 2.34 2.35 2.36 CHAPTER 2 Prove (iii) of Theorem 2.1 I See Problem 2.15, Prove (iv) of Theorem 2.1 AF The ientry of A+B is a, +b,, and the jrentry of B+A is 6, +a). However, by the commutative law in K, a+b, =, +a). Thus, A+B and B+ A have the same é-entries, and hence A+ B= B+ A. Prove (v) of Theorem 2.1. A The ij-entry of A+B is ay+by5 hence kfa,+b,) is the ifentry of k\(A+B). The irentry of kA is k,a,, and the jj-entry of kB is k,b,; hence k,a,+k,b, is the irentry of k,A+Kk,B. However, by the distributive law in K, kya +6,)~ kya, + kb,» Therefore, k(A+B) and k,A and kB have the same ijentries; and hence (A+ B)=,A+ &,B Prove (vi) of Theorem 2.1 W Asin Probicm 2.28, the proof is by the distributive law in K. Prove (vii) of Theorem 2.1 The entry of (kyKa)A is (Ayka)a,, ‘The irentry of kA is Kya,,, and so k,(k.a,) is the Hpentry of A,(K.A). However, by the associative law of multiplication in K, (k,k,)a,, = ky(ka,)- Therefore, (KE:)A and k,(K,A) have the same jj-entries, and hence (K,A,)A = k,(KsA). Prove (vii) of Theorem 2.1. F The ieentry of 1+ A is 1-0, 1 Since 1+A and A have the same s-entries, they are equal. ‘Comment on the difference, if any, between the + signs in (vi) of Theorem 2.1 On the left, the + sign refers to addition of scalars in K; on the right, to addition of matrices in Prove that 0A=0, for any matrix A, © By (vii), (vi), and (ii) of Theorem 2.1 At0A=1A40A=(+0)A=14=A and the proof follows upon the addition of —A to both sides. Show that (—1)A= BAy(-1)A=14+ (-1)A=(1+(-1))A=04=0, where the last step follows from Probiem 2.33, Now add =A to both sides Show that A+A=2A and AtA4A=3A A Using (vi) and (vii) of Theorem 23, 24=(141)A=14+14= 444A. Similarly, 34=(241)A= BANIA=A+A+A. Prove that, for any positive integer n, SS A=A+ Att A= nA, 41 ‘The proot is by induction on m, The case m=1 appears in Theorem 2.16 the theorem: holds for mf. ‘Then Sa-Sa+a Suppose n>1, and =(n-1)A+1A=[(1-1)+ 1A = nAMATRIX ALGEBRA 0 33 2.3 MATRIX MULTIPLICATION 2.37 2.38 2.39 ‘The product of a row matrix and 4 column matrix with the same number of elements is their inner product a5 defined in Problem 1.45: oy (a,.a,, af a,b, +4,b,¢6-40,0,= > ab, Calculate: ‘ 3 ie ‘ (a) (8, -4, 5)| 3 (6) (6, Sy (c) B.8, -2.(-1) 1 ¢) 3 and (d) (1,8, 3,4)(6.1,-3.5) F(@) Mutiply corresponding entries and add: (8, 2 af 2)- 0) (492) + 3)(-1) = 24-8 -S=11 a (5) Multiply corresponding entries and add: 4 w-rne{3} 2449-21 + 10=22 3 (©) The product is not defined when the row matrix and column matrix have different numbers of elements, (4) The product of a ow matrix and a row matrix is not defined. Let (rs) denote « niat (@) @x3)Gx4) (@ (xDGx1) (@ _Gx4VGx4 @) @xax2) — @) GxIAG —) Bx 2x4) With size Fs. Find the size of each product, when the product is defined: FH An mx p matrix is multpliable om the right by aq Xn matrix only when p=q, and then the product isan mx mn matrix. (a) 2%4; (6) 4X2; (©) not defined; (d) 5x3: (e) not defined; Cf) 2x4, Suppose that A=(a,) isan mp matrixand Be(b,) isa p%n matrix. Then the product AB=(c,) isthe mxn matrix for which > ayb, y= aby + agby Hot apd, that is, the j-entry of AB is the product of the ith row vector of A and the jth column yector of B, Find the product AB for afte 2 0 -# 4a a) 8G 3 4) FE Since A is 22 and Bis 2%3, the product AB is defined asa 2X3 matrix, To obtain the Ieee Be et of AB) eth es tor ar (113) 6A Oy tro tema (7) (0S) ad (ee) fey (ps (3) 0 Pe) (22) + OH CHO + YE-2)_CNK-A+ VOY) zt -2] Le Cae ey34 2.401 241 2.42 2.42 CHAPTER 2 To obtain the entries in the second row of AB, multiply the second row (2,~1) of A by the ‘columns of B, respectively: Ne i -6 14 ) -(@@+ene) (20) + (1-2) @N—4) + (118), Thus ape( 6 4) 1 2- Find the product BA of the matrices A and B in Problem 2.39. # Note that Bis 2X3 and A is 2%2. Since the inner numbers 3/and 2 are not equal, the product BA is not defined. Find the product AB, where A=(2 1) and B hei @ Since Ais 1X2 and Bis 2x3, the product AB is defined as a 1X3 matrix, of row vector with 3 components ‘To obtain the components of 4B, multiply the row of A by each column of B: (3) FE] B))-coans coo, 2+). 094 MEH) =6.1.-3) Find the product 4B, if W Since Ais 3x2 and Bis’ 2X3, the product AB is defined as a 3X3 matrix. To obtain the first Tow of AB, multiply the first row of A by each column of B, respectively | aa am y the second row of A by each columa of B, respectively: ak 8 =10 =1.=8 =10! (+0 -20 =sre)-( Te 5) OOOGART 3 4 oy = ~ —3+12 6+16 1840, 9 15, -8 -10) 2 3) 2B 15 To obtain the To obtain the le Thus. Find the product BA, where A and B are the matrices of Problem 2.42. B Since Bis 2x3 and Ais 3X2, the product RA is defined asa 2x2 matrix. To obtain the first row of BA, multiply the first row of B by each column of A, respectively: Fal il (C238 tenes) “2 ms 3 424d 2.45 2.46 2.47 2.48 MATRIX ALGEBRA 2 4 (= a)(| i] fel) -(the eo) <5 2) 3 (to) 3) Thus BA Find the size of the product AB, where ye) ee a 4 0-2 Oo F Since Ais 2X3 und Bis 3x4, the product ABisa 24 mati Suppose AB=(c,) for the matrices A and B of Problem 2-44, Find: (@) c2ys (6) ejay (€) €21s (d) Crs FF ‘The vlbroeaie), the ment of AB, ir the product of row fof A by calumaf of €. 0 @ S20 af 3-0 +ma+en-2= +0+6=6 2) 1 a) _-h0(-2) =a) -Cn1-b- 40) +140=3 0 a © «=0.0.-92) 0K) +02) +e-9)=140-1 4 (4) The element cy, does not exist, since A, and with it AB, has only two rows Find AB, where 2-1 06 ene | a-(i 3-5 ') 41 22 FF Since Ais 2=3 and Bis 3%4, the product is defined as a 2%4 matrix, Multiply the rows of A Pr pl by the columns of B to obtain: eee eee ee ete eae) 8-2+20 -4-6+5 0+10-10 24-2+10/-\25 -5 0 32 Refer to Problem 2.46, Suppose that only the third column of the product AB were of interest. How could it be computed independently? F_ By the nute for matrix multiplication, the jth column of a product is equal to the frst factor times the jth column vector of the second. Thus, eo SER Greed Similarly, the ith row of a product is equal to the ith row vector of the first factor times the second factor, rnd (5 5)(4) I. The first factor is 2x2. and the second is 2% 1, so the product is defined as a 21 matrix (23 5)(4)= (2555) =a)36 0 CHAPTER 2 6\( 2 1 249 rina (9)(2 2) 1 The product is not defined since the first factor is 2x1 and the second factor ig 2x2 ra ee Gn 5) A The first factor is. 1x2 and the second factor is. 2x2, so the product is defined as a 1X2 (row) matrix @-n(_3 $)=@+21,22-35)=@3,-2) 251 Find (_} $)@2,-7) F The product is not defined, since the first factor is 2X2 and the second factor is 12. 282 Let Abe an mn matrix, with m>1 and n>1, Assuming wand v are vectors, discuss the conditions under which (2) Au, (b) vA is defined. 1 (@) The product Aw is defined only when wis @ column vector with 1 somponents; i.c., an m1. matsix, In such case, Aw is a column vector with m components. (6) The product vA is defined only when » is @ row vector with m components; i.e.. a 1 xm matrix, In such case, vA is a row vector with n components. 2.53 Compute .) ( 3} =4 5) “1 4) The first factor is 3x1 and the second factor is 13, 0 the product is defined as a 3x 3. matrix 2 (Je aes 36) GY-4) (305) WW -12 15 -1 IMG) (1-4) (-1)5)/ \-6 4 =5, 6-4. 9 ) =i 4H The fist factor is 1X3 and the second factor is 3%1, 90 the product is defined as @ 1%1 matrix, which we frequently write as a scalar, (a6 (2-4) oe) (i -8 ") 2.54 Compute 2 (6, 49 :) = (1212-5) =i Problems 2.55-2.58 establish the following theorem, where we assume that all products are defined, Theorem 2.2: Suppose that A, B,C are matrices and & isa scalar. ‘Then () (AB)C=A(BC) _qusociaive law (i) A(B+C)= AB + AC left distributive law (il) (B+ C)A=BA+ CA right distributive law iv) MAB) = (kA)B = A(kB) 2.55 Prove (i) of Theorem 2.2, F Let A=(a,), B=(b,), and C=(c,,), Furthermore, let AB=$ (s.) and BC2.56 287 2.58 2.59 2.60 MATRIX ALGEBRA 2 37 Then tu~Saibe = Due Now multiplying § by C, i.e., (AB) by C, the efement in the ith row and /th column of the matrix (AB)C is YY audew On the other hand, multiplying A by 7; ie., A by BC, the element in the ith row and /th column of the matrix A(BC) is ‘The associative law in the field of scalars implies that the two double sums are equal, proving (i). Prove (ii) of Theorem 2.20, FH Let A=(@,), F=(b,), and C=(c,). [Since AB and AC are defined, we can use the same index k for the columns of A and the rows of Band C] Let D=B+C=(d,), E=AB=(e,), and F=AC=(f,). Then +e t= Deaaby Sym Be day Hence the if-entry of the matric AB + AC. is Cy AH = Zs MuByt D Ainley = Z aby Fala) @ On the other hand, the sj-entry of the matrix AD=A(B+C) is 2 albu tow) 2 ‘The right sides of (1) and (2) are equal by virtue of the distributive law in the scalar field; this proves (ii). Prove (ii) of Theorem 2.2. I The proof is as in Problem 2.56, [There is no distinction between left and right multiplication in the ficld af scalars.) Prove (iv) of Theorem 2.2. ! «(Ea,0,,) =3 ea, 0, =D a, hb, Display two matrices A and B such that A and BA are defined and have the same size, but AB# BA (3 $) sod a=(} 9). Then ae 35 2-1 ap=(5 3) -1)-(4ro 0s pa=(2 3 5)°G35 nos)-G Matrix: multiplication does not obey the commutative law, Show that 04 =O if A is not square, the two zera matrices will be of different sizes) I Each entry of 0.4 is the inner product of 2 zero row of 0-and a column of A, and hence ig the scalar 0. Thus, 0A=0.2.61 2.62 2.63 CHAPTER 2 Show that A0=0. I Each entry of AO is the inner product of a row of and a zero column of 0, and hence is the scalar 0 Therefore, A0~ 0. Show that we ean have A 0, with 40 and 8x0, Hitt a=(} 2) and B=( § 2), then ae) By) S| bya + Bye ++ Bylin a, ‘Thus, (AB)" = BA”, since corresponding entries are equal, ELEMENTARY ROW OPERATIONS; PIVOTS Show that each of the following elementary row operations has an inverse operation of the same type. [Ei]: Interchange the ith row and the jth row: R= R,_ [E,]: Multiply the ith row by a nonzero scalar A: | R,—- kk, k £0, [£J; Replace the ith row by & times the jth row plus the ith row: R,—» kR, + R, F(a) Imterchanging the same two rows twice, we obtain the original matrix; that is, this operation is its own inverse. (b) Multiplying the ith row by k and then by k-', or by k-' and then by k. we obtain the original matrix. In other words, the operations R,->4R, and R,+k'R, are inverses. (e) Applying the operation R,>4R, +R,, and then the operation R,—>—KR, + R,. or applying the operation Ri KR, +R, and then the operation R,—>kR, + R,, we oblain the original matrix. In other words, the operations R,+kR,+R, and Ri —KkR,+R, are inverses Express the following row operation in terms of the elementary row operations of Problem 2.79: [8]: Replace the ith row by A" times the jth row plus & (nonzero) times the ith row: Rj k’R, + ki, ke. FT Eis equivalent to £, [with parameter k] followed by E, [with parameter ky.2.81 2.82 2.83 2.84 2.85 2.86 2.87 2.88 MATRIX ALGEBRA J 41 Apply the operation Rc R, to 34 i 2 ae 1 bso) sot 8) Apply the operation R,-»3R, to the matrix of Problem 2.81. 3 69 12 1 (676 3 4 5 ~6 Apply the operation R,*-3R, + Ry to the matrix of Problem 2.81 Pes ' Doe a 0-10 -4 -18 Matrix 4 is row equivalent to matrix B, written A~@B, if B can be obtained from A by a sequence of elementary row operations. Show that row equivalence is an equivalence relation, That is, show that (@ A~A; (b)if A~B, then B~A; (Cif A~B and B~C, then A~C. TH (@)A can be obtained from A by applying E, with k=1, (b) If B can be obtained from A by a Sequence of elementary row operations, then applying the inverse operations to B in the reverse order will yield A. [By Problem 2.79, the inverse of an elementary row operation is an elementary row ‘operation,] (c) If B can be obtained from A by a sequence of elementary row operations, and C can be obtained from B by a sequence of elementary row operations, then applying the sequences one after the other on A will result in C. Suppose that a, is a nonzero element ina matrix A, Show that each of the following row operations, which change the kth row of A, yields a 0 in the &-position of A. (a) Re Cayla,)R +R, b) Rea, R, + a/R, [The above clement a,, which is used to produce Os above and/or below it is ealled the pivot of the ‘operations.] A The new scalar in the &j-position of A is (a) (ayfag)a, + a; =0 (0) aeay + (a,)ay) = 0 Refer to Problem 2.85. Discuss the advantages, if any, of using operation (b) instead of operation (a). FF Although (2) will involve fewer arithmetic operations (Problem 2.91), fractions may occur even if all a, are integers. Operation (b), which uses only multiplication and addition, will not produce fractions if all the entries a, are integers. Discuss the advantage of using @,—1 as a pivot. In this case, operations (a) and (b) are the same, and no fractions will be introduced if all the a, are integers Produce 0s above and below the (boxed) pivot: eee Oo -F 3 4