ee eee 1 VECTOR SPACES _— a We have already defined R® and examined some of its basic propenies | must now study the fundamental structure of R". In many application i mathematics, the sciences, and engineering, the notion of a vector space arg | This idea is merely a carefully constructed generalization of R”. In studi the properties and structure of a vector space, We can study not only 2°, | particular, but many other important vector spaces. In this section we defing | the notion of a vector space in general and in later sections we study thir structure, DEFINITION 1* A real vector space is a set of elements V together with two operations @ a, © satisfying the following properties: ; (a) If wand v are any elements of V, then u @ v is in V (i.c., V is close under the operation @). (@) u@v=vOu, foruand vin V. (b) u@ (V@W) = (UB y) © w, for u, v, and win V. (©) There is an element 0 in V such that u®0=0Gu=u, foralluinV. (d) For each u in V, there is an element —u in V such that u@®-u=0. (6) Ifuwis any element of V and c is any real nuniber, then c @ wisin V V is closed under the operation ©), t © €0 (OW =cONse Oy, forall reat numbers cand aU inv, f VV. is + © 0u=¢Ou@d Ou, forall eal numbers cand 2 | 7 "7 5 ©) ¢© @ Ow) = (cd) Ou, for all real numbers’c and d andall™ (h) 1Ou=u, forall uin Vv. “Although the definitions es tions in this book ‘i ered Oe vil be refered foots i this book are not numbered, this deition is au! 238 ‘oa numberof times in this chapter.Remark See. 6.1 Vector Spaces 239 The clements of V are culled The operation ( is called vector multiplication, ‘The vector 0 in vectors; the real numbers are called sealars, addition; the operation © is called scalar ae "0 in property (c) is called a zero veetor, ‘The Hee ' nye (A) is called w negative of u., It can be shown (see i; ses 1.5 and 1.6) that the vectors 0 and ware unique. Wwe allow the scalars to be complex numbe spuce, § ich Spaces are important in many Applications in mathematics and the Physical sciences. We provide a brief introduction to complex vector spaces in Appendix A. However, in this book we limit our study to real vector spaces, s, we obtain a complex vector Consider the set R" together with the operations of vector addition and scalar multiplication as defined in Section 4,2. ‘Theorem 4.2 in Section 4.2 estab- lished the fact that " is a vector space under the operations of addition and scalar nfultiplication of n-vectors. . Consider the set V of all ordered triples of real numbers of the form (x, y, 0) and define the operations ® and © by ONO YO = tx, y+ yO) €O (x, y,0) = (cx, cy, 0). then not difficult to show (Exercise 7) that V is a vector space, since it fies all the properties of Definition 1. . (x, Consider the set V of all ordered triples of real numbers (x, y, z) and define the operations @ and © by DOO Z=H+x ytyz+z/) €O (x, ¥.2) = (cxyy, 2) It is then easy to verify (Exercise 8) that propertiés (c), (8), (a), (b). (), (A), and (e) of Definition I hold, Here 0 = (0, 0, 0) and the negative of the vector (x, y, 2) is the vector (-x, —y, -z). For example, to verify property (e) we proceed as follows. First, cOl@ DO yA =cO+x yt yz+7z) H=(@+x,y ty ztz). we Also, (cx, ¥, 2) © (cx’, 9,2) (cx tex’,y+y 242) See tx, yt yi zt2). CO DOOM 2) However, we now show that property (f) fails to hold. Thus + (c+d) 0 (x,y,z) = ((e + Ax, ys 2) On the other hand, . COW, ¥.2) Od O (4,92) = (CX. 9,2) 8 XY.) =(ex+dx,y+y242) = ((c + d)x, 2y, 22). .d bperations. Incidentally, . ‘Thus V is not a vector space under the prescribe properties (2) and (h) da hold for this example,240 Chapter 6 Real Vector Spaces > PSUUNIELN conser te set Aly of all 2 x 3 matrices under the usual operations op jon, In addition and scalar multip! n 1.4 (Theorems 1.1 ang My have established that the properties in Definition 1 hold, thereby may) my, the set ofall m x m matrices under ql® My | nd scalar multiplication isa vectgg sa is vector space will be denoted by My \ of all eal-valued functions that are defined on th jy V, we define f ® g by Mer CIES Let V be th a,b). AF f and g are (FOQO =f +80. ~ i If f isin V and c is a scalar, we define ¢@ f by (€O fy) =cf(0). Then V is a vector space (Exercise 9). Similarly, the set of all real val functions defined forall real numbers is a vector space 7 1 Another source of examples of vector spaces willbe sets of polyno therefore, we recall some well-known facts about such functions. 4 pong tial (in t) is a function that is expressible as P(t) = ayt" + ayy”! +++ Fait +a, ) where aq, a), ..., dy are real numbers. MEST | 1- ftiowing functions are polynomials: pult) = 344 - 2 451-1 pat) =21+1 pst) =4. The following functions are not polynomials (explain why): filt) = 2-6 and fs(t) -Uth 2 The polynomial p(1) in (1) is said to have degree m if ay # 0. Thus Aegree of a polynomial is the highest power having a nonzero coeficet. EE Polynomials defined in Example 6 have the following degrees u(t): degree 4 x(t): degree | ‘ a(t): degree 0. The zero polynomial is defined as Or +O 4 FOr +0, Note that, by definition, the zero polynomial has no degree. io ine ME Ot Py be the ECO panonate a gree < n wogeller™ © 7440 polynomial, Thus 21? — 31 +5 is a member of P2. pe a |Sec. 6.1 Vector Spaces 243 by (f) of Definition 1. Adding —0u to both sides of (2), we obtain by (b). (c), and (d) of Definition 1, | Om ous (0 \ (b) Exercise T.1. (Ou + Ou) + (—Ou) = 0u + [Ou + (—Ou)] = Ou + 0 = Ou. (©) Suppose that cu = 0 and c # 0. We have u=tu=(Ze)u 0=0 by (b) of this theorem and (g) and (h) of Definition 1. (@) (-Dutus (-1ut (us (“1+ Iu = Ou = 0, Since —u is unigu® we conclude that (—1)u = determine whether the given set V 1 through 4d sae the operations @ and ©. ee sofa oe pts of real UMBETS (9 here x > Oand y > 05 (OGY a(xtxyyty) cO (x.y) = (exey)- 1 Visthe set of all ordered triples of real numbers of the form (0, 9, 25 2) 6 (O.-y',2') a €O.5. A Visthe set of all polynomials of the form at? + bt +c, Wire, b, and ¢ are real numbers with b = a + 1f (at? + byt + cy) ® (aa? + byt + 02) = (a tag? + Oy + bse + (C1 +2) ad O(a? +bt +e) = (ra)t? + (rb)t + re. Ais ‘She set of all 2 x 2 matrices [4 co a" hee, = 4; ® is matrix addition and @ is scalar "ubipicaion, ‘ hie that R? is a vector space. a, a that R? is a vector space. Ware FS in Example 2 is a vector space. FPP hota Properties of Definition 1, except Be ie set in Example 3. 9. Show that the set in Example 5 is a vector space 10, Show that the space P of all polynomials is a vector space. In Exercises I through 17, determine whether the given set together with the given operations is a vector space. If itis not a vector space, list the properties of Definition | that fail to hold. 11, The set of all ofdered triples of real numbers (x, ys 2) with the operations (uy OO IZ) =O yt vz) and cy ex, CY, €2). 12, The set of all ordered triples of real numbers (x, y. 2 with the operations Gy DBO, and ss (etxyytyictzZ) (x, 1, 2), coy 13. The set of all ordered triples of real numbers of the form (0, 0, z) with the operations (0,0, 2) ® (0,0, z') = 0,0.2 +2’) and € (0,0, 2) = ©, 0, c2). 14, The set of all real numbers with the usual operations of addition and multiplication. 15, The set of all ordered pairs of real numbers (x, y), where x < 0, with the usual operations in R®. 16. The set of all ordered pairs of real numbers (x, ») with the operations (x, y) ® (x', y') = (+x. y +) and cO(x,y) = (0,0).‘Real Vector Spaces positive real mambersu with the operations Yeuanieon TR Le Whe wey space? 19. Ler set of all real numbers; define © by Vand Oby cOu= cu, Is V avector Whe the set consisting ofa single element 0, Let In Exersises T.1 through T, establish the indicated result for 4 real vector space V. Tl. Show that © = 0 for every calc, T2, Show that ~(-u) = u, TA. Show that if u ++ =u + w,then v= w, Men REE eer a JATLAB Exercises The concepts discussed in this section are not easily implemented in MATLAB routines. The items in Definition 1 ‘must hold forall vectors. Just because we demonstrat in MATLAB that a property of Definition J holds for afew Vectors it does not imply tha it holds forall such vectors. You must guard against such faulty reasoning. However if Jor a particular choice of vectors, we show that a property Jails in MATLAB, then we have established that the property does not hold in all possible cases. Hence the property is considered to be false. In this way we might be able to show that a Set is not a vector space. MLL. Let V be the set of all 2x 2.matrices with operations siven by the following MATLAB commands; AGB is AaB KOA is k+A Is V a vector space’ (Hint: Enter some 2 x 2 ‘matrices and experiment with the MATLAB , ‘commands to understand their behavior before checking the conditions in Definition |.) ML.2, Following Example 8, we discuss the vector space P, :? 6.2 SUBSPACES ‘mn this section we begin to analyze the structure of & oe ! Convenient to have a name for a subset of a given vector SPA; pus ‘ ‘ in Yeclor space with respect tothe same operations as those "the following 020=0andeoo=9, Show thar y 20. (a) Visa ve (b) Describe all vectors Theoretical Exercises Ta, Show that if u #0 and a 5, Show tht vector space hs ny oe Oe 2 eg Té Show that a vector win a vector sean ewis “DEFINITION Lev be avecorspce and W anonempy subset! espect to the operations in V, then W is call iva ElOF space that hy many vectors are in yn PACES hay, vectors, ving g Arial ‘i = BU theng F negative —u. Pac he ay of polynomials of degree n or les polynomials can be performed i Software by associating row matixof sary with polynomial p(t) of P,, The row natives of the coefficients of p(t) using the associating PA) = at" + at SS. Operates in Tinear alge, toopatta, [te ar a al Tf any term of p(t) is explicitly missing, ezaois ‘used for its coefficient. Then the addition of polynomials corresponds to matrix addition at multiplication of a polynomial by a scalar 7 corresponds to scalar multiplication of matics MATLAB to perform the following operations © polynomials, using the matrix association above. Let n = 3 and pi) =2P 45° 41-2 qgt)=P4+3t5. &@ pe+q@). SPC (c) 3p(t) -4g(r). ; wa Boat ats ; wot . a Ted a subsP™™Proof Remark Le EXAMPLE 3 | Solution See.62 Subspaces 245 Every vector space has at te camlaieninaniie ast (Wo subspaces, itself and the subspace {0} zero vector [recall that 0.4) 0 = Oand «© 0 = 0 in any vector space (sce Exercise 19 in Section 6,1). ‘The subspace (0) is called the zero subspace, Let W be the subset of #* consisting of all vectors of the form (a, b. 0), where and D are any real numbers, with the usual operations of vector addition and scalar multiplication, ‘To check if W is a subspace of K3, we first see whether Properties (#7) and (f) of Dedinition | hold. ‘Thus Jet’ u = (ay. b).0) and ¥ = (ay.b3,0) be vectors in W. Then u + ¥ = (ay, by, 0) + (a2, be. 0) (a +b ata, by + by, 0) is in W, since the third component is zero. Also, if ¢ is a scalar then eu = c(ay,b),0) = (cay, cby,0) is in W. Thus properties («) and (f) of Definition I hold. We can easily verify that properties (a)-(h) hold. Henee AV is a subspace of KR! = Before listing other subspaces we pause to develop a labor-saving result. We just noted that to verify that a nonempty subset W of a vector space V is a subspace, we must check that (a), (8), and (a)-(h) of Definition 1 hold. However, the following theorem says that it is enough to merely check that (a) and () hold, Property (a) is called the closure property for ©, and (A) is called the closure property for. ¢ - Let V be a vector space with operations ® and © and let W be a nonempty subset of V. Then W is a subspace of V if and only if the following conditions hold: (a) [wand v are any vectors in W, then u@v is in W. (B) Ic is any real number and w is any vector in W, then ¢ © wis in W. Exe ise TA. . Observe that the subspace con: is a nonempty subspace, ing only of the zero vector (see Example 1) Consider the set W consisting of all 2 x 3 matrices of the form ab 0 ve fp caf! where a, b, ¢, and d are arbitrary real numbers. Then W is a subset of the Vector space Mzy defined in Example 4 of Section 6.1. Show that W is a subspace of Mzy. Note that a2 x 3 matrix is in W provided its (1,3) and (2, 1) entries are zero. az by 0 i a) by) 0 ns . come [i % S]oniv= [3% S]ew. re _[ata bith 0 BHVE] 9 ter di tdr so that (x) of Theorem 6.2 is satisfied. Also, if k is a scalar, then a= |"! kb al isin W ] isin W key kdh so that (B) of Theorem 6.2 is satisfied. Hence W is a subspace of Mas. ._ 246 charter Real sor SI eee ye at nl fe de space of if and only Ef ay scala 4 and b (Exe 3 ig J vectors of the form (a. b, 1), wf 4 ao we the sub ar of RCOnSISTT I whether propertics (a) and’ : ES ot any rea nu panies ean ‘Theorem 62 nol mye e Then pes conte not hold, a r nection 61> 8° dent oe Ua of deste’ 4 nan o poly PZ ofall polynomials. I s ° general. that Pa js a subs] ‘ (Exercise 8)- LetV perth et ofall polyno ordegree exactly = 2: V isa.cy pat itis not a subspace © p,, since the Sum ofthe polynomials 27+ 34 and -2e +208 potynomial of AeBTee 1, is notin V. 1 (Calculus Required) Let Cla. b] denote the set of all real-valued conti Fe jefined on the interval fa bh. If f ands arein cla tif tions is continuoase functions that f +g isin Cla, Bh since the sum of two continuous func Ae taly if cisa scalar then cf isinC Si eetor space ofall real-valued functions that are defined on {a, 6), whid was introduced in Example 5 of Section 6. 1, If the functions are defined, Ail real numbers, the vector space is denoted by C(—22, 00)- 1 [a,b]. Hence Ca, b] is a subspacedt] We now come to a very important example of a subspace. ser the homogeneous system Ax = 0, where A i Sines ma ee se f slatns ote homogeneous system. Since AO = 0, We conclude t ia 7 To heck that IV is a subspace of R", we verify Prope orem 6.2. Thus let x and y be solutions. Then Ax=0 and Ay Consic Now A(x +y) = Ax+ Ay =04+0=0. sox-+yisa solution, Also, if c is a scalar, the , then A(cx) = c(Ax) = c0 = 0, so cx is also a soluti ice W is a subspace solution, H ibs : . Hen i space of the homogeneous system, or the null space ull 'spé n, called the sol “A. It should be n 0 where A is m x n, rae to ie set of all solutions to the linear syste space of R" if b £0, ) (Exercise 73). UIE A simple way and y; of construct mba aan ae ai ree ilo : t see Section 13) of ved y and let w be the oe ft es -, that is, W cons! i slSec.6.2 Subspaces 247 in V of the forma that W is a subsp: Thus let WYt+ ava, where a 1 and ay are any real numbers. To show ace of V, we verify Properties (a) and (#) of Theorem 6.2. Wi Savi bay, and wy = by, + b2v2 be vectors in W. Then Wit wa = (ivi + azva) + (bivy + byve) = (ay + by)vy + (a2 + ba)v2, which is in W. Also, if cis a scalar, then ew = (cay)vy + (caz)v2 is in W. Hence W is a subspace of V s The construction carried out in Example 9 for two vectors can easily be performed for more than two vectors. We now give a formal definition. Let V1, V2, ..., vy be vectors in a vector space V. A vector v in V is called a linear combination of v,, sve if v Wi cava boob eavE for some real numbers cy, c2, ..., cy. (See also Section 1.3.) { In Figure 6.1 we show the vector v in R? or R? as a linear combination of the vectors v; and v2. Figure 6.1 > on Linear combination A ofp vectors veam tes i °. | | ” am a) In R? let yy = (12,1), v2=(1,0,2), and v3 = (1, 1,0). The vector v=(1,5) is a linear combination of v), ¥2, and v3 if we can find real numbers cy, c2, and 3 80 that CLV) + C2V2 + 633 = Substituting for v, v1, V2, and v3, we have cr(1y 2, 1) +62(1, 0,2) + ex(1, 1,0) = (2, 1,5). Combining terms on the left and equating corresponding entries lends to the i verify) linear system (' at akan 2c, toast oy + 202 =5.248 = Chapter6 Real Vector Spaces m Solving this linear system by the methods of Chapter I gives (rity) cy = 2, and cy = —1, which means that v is a linear combination oy and v3. Thus Mn yp v2 = V3. | Figure 6.2 shows v as a linear combination of ¥), v2, and vy Figure 6.2> z j ch | \ BN\( If S = (v1, Va... Ve) is a set of vectors in a vector space V, then the sett all vectors in V that are linear combinations of the vectors in S is denoted by span Sor span {Vj}, V2.0... Vil- In Figure 6.3 we show a portion of span (vj, ¥2), where vy and ¥; | noncollinear vectors in R¥; span (v1 a] isa plane that pases through origin and contains the vectors v, and v2. Figure 6.3 > ERR Consider the set $ of 2 x 3 matrices given by s={[! 9 o} fo 1 o 09 a} fo 0 of) 0 0 of'[o 0 of*jo 1 of'jo ot ‘Then span S is the set in Mz3 consisting of all vectors of the form 1 omo 010 aed io of] «[s 2 **[ 0 lee[o 1 of*@{o 0 | =o " d + where a, b,c, and d are real numbers-y Sec. 6.2. Subspaces 249 That is, span ‘pan Sis the subset of Mas consisting of all matrices of the form abo Ocd where a, b,c, and dare real numbers. bl Let $= (v1.2 V4) be a set of vectors in a vector s is a subspace of V) of na vector space V. Then span S Proof See Exercise T.. ysrtr+2, ve _ Determine ifthe vector : u belongs to span {v1, V2, V3, Va) a Solution If we can find scalars ¢y, ¢2, ¢3, and cy so that CIM, + e2¥2 + 33 + C4V5 then u belongs to span (v1, V2, V3, V4}. Substituting for u, vi, V2, ¥3, and Vs, + we have - ; asa cult? 1-2) + cg(t? = 21) + e9(51? — St +2) + c4(— = 3 — 2) this =P 4642 egy or (2cy +e + Sey — cat? + (cr — 2er — Sea — 3ea)t + (2ei + 2c3 — 2cs) [fh aPtrt2. Now two polynomials agree for all values of t only if the coefficients of re- spective powers of f agree. Thus we get the linear system 2er+ co+Se3- y= \ cr — 2c) — Se ~ 3e4 = 1 2c} + 2c3 — 2ey =2. To investigate whether or not this system of linear equations is consistent, we form the augmented matrix and transform it to reduced {row echelon form,\, m obtaining (verify) 1D) ov 3-0 os) Hj WB 1 fo fe Oo WO which indicates that the system is ingonsistent, that is, it has no solution. Hence u does not belong to span (V1, V2. V3. ¥s}- . Remark In general, to determine if a specific vector v belongs to span S, we investigate the consistency of an appropriate linear system. 4 ...sof AAWhich of the following subsets of RS are subspaces RY The set of all vectors of the form WV (a. b,2). MObY (a. b,c), where c= a+b (a, b,c), where c > 0. Lame ofthe following subsets of R? are subspaces of R°? The set of all vectors of the form (a) (a,b,c), where a = ¢ = 0. (b) (a,b, c), where a =~ (©) (a,b, 0). where b = 2a +1 A. Which of the following subsets of Ré are subspaces of #2 The set of all vectors of the form (arb, c,d), where a = b = 2. \00V (a, bred), where c = a + 2b and d va 4 Awnich ofthe following subsets of R* are subspaces of R*? The set of all vectors of the form (a) (a,b, c,d), where a = b=0. (©) (a,b.c,d),wherea = 1,b=O,anda += 1. (©) (a,b, c,d), where a > Oandb <0. 3b, c,d), where a = Oand b= “Which of the following subsets of P are: subspaces? ee ‘of all polynomials of the form bp gut? + at + a0, where dp = 0. ates tag, where dy = 2 (6) ant? + ayt + ao, where a2 + 41 = a0. Which of the folloviing subsets of P, are subspaces? ‘The set of all polynomials of the form (a) ast? +ayt + a9, where a; = 0 and ay (©) axt? + ayt +a, where a; = 2a he + ayt +9, where a; + a) + dp = 2. af show that P, isa subspace of Ps ‘Show that P, isa subspace of Pys1- 8. Show that P, isa subspace of P. 9. Show that P is a subspace of the vector space defined in Example 5 of Section 6.1. Let = (1,2, -3) and v = (2,3, 0) be two vectors in R® and let W be the subset of R? consisting of all vectors of the form au + by, where a and b are any real ‘numbers. Give an argument to show that W is a subspact of R?, M1. Letu = (2,0,3,—4) and v = (4,2, ~5, 1) be two vectors in R* and let W be the subset of R* istit all vectors of the form au + by, where a inueay a real numbers. Give an argument to show that W is. 4 subspace of R°. ; Je ‘Which of the following subsets of the vector 15. ‘Which ofthe following subsets ofthe yy Jefined in Example 4 of Section 6,1 apg 0 hcg set ofall matrices ofthe form subg Ree 78 a i (a) é ¢ ppwhereb =a +, of where c > 0, ‘I where a = —2 sf = Pend fae, zs asc es oc a © i Mefned in Example 4 of Section 6, ce serie all matrices of the form Shon abe lg 2 fmt = 2041 ola o [: ¢ ‘I a be (c) |, wherea +e=0 olf eo i es Oandb +d ja5 ‘Which of the following subsets ofthe vector spc are subspaces? tM (a) The set of all n x n symmetric matrices / (b) The set of all n x n nonsingular matrices} (©) The set of all m x 1 diagonal matrices. / Which of the following subsets of the vector space, are subspaces? (a) The set of all n x n singular matrices. (6) The set ofall n x n upper triangular matics (c) The set of all n x n matrices whose deternita's! 16. (Calculus Required) Which of the following sts 17. (Calculus Required) 4 are subspaces of the vector space C(—20.) efalt Example 7? (@) All nonnegative functions, (b) All constant functions. (©) All functions f such that (0) = 0 (@) All functions f such that £0) = 5: (€) All differentiable function. j swing Which of the fot. the vector space C(—00, 00) defined in Bxat subspaces? (a) All integrable functions. (b) All bounded functions. (©) All functions that are integrable on [a,b] (@) All functions that are bounded on [a6] aed 4Consider te differential equation Uy +2y=0. Jed faetion f satisfying the real-2i0%rof all solutions to the given eNorne © and © asin Example Sin that Wis asubspace of the vector # functions defined on (20, 00). Gal eavalved stot eGo 2) ee ae ofthe following subsets of R* are c Iie } 2 =: 9 | | | | y [* o * { 1 Dexmine which ofthe following subsets of R° are Syms. ® y = O| u al Exercises [egg mame. tha Saco Ret ofa veto space Visa Str and nly if the following eonition Pesca age mY Vectors in fay, ° any scalars, then au + bY that ag tthe nism Of al Solutions to Ax = b, where A +n, Subspace of R* if b #0. $s. 6.3, Nel be a set of vectors in a vector Sec.6.2 Subspaces 251 (b) x x 21. In each par. determine whether the given vector ¥ belongs to span { }. where W=(1.0.0.1), v= (1,-1,0.0). and w= O1.20. (a) v=(-1.4.2,2). ) v=(1,2,0,1). (©) v= (-1,1,4,3). @ v=(0.1.1,0). 22, Which of the following vectors are linear combinations of og 3-2 © [ 3} @ [: HI 23, Ineach par, determine whether the given vector p(t) belongs to span (pu(0).pa(®). pst} where pit) =P 1 pW) =P 2+ pe) () pi)=38-3+L pe (© pi)=t+l @ p= space Vand let W bea subspace of V containing S. ‘Show that W’ contains span S. mé IF A isa nonsingular matrix, whats the nll space of ‘A? Justify your answer, bea fixed vectorina vector space V. Show that TY, Lets i the set WY consisting of all scalar multiples exo of % is a subspace of V. TB Leta beanm x n matrix, I the set of all vectors ¥ in R* such that Ax # Oa subspace of R"? Justify your answer.252. Chayer Real We SENT et ane (0) aunt rn sr9, Show thatthe ony spaces oF to) inset ; seem gce VP Eel TA Ler Ay and HE Be STTNTN Wee We the sto ATE STS an 1 ey ii aa 2 2 We is 90S ny Show that Wy woot Y eV with enn Lee Wy aa Ws De goes oF A NETT gue W witl MATLAB Exercises ML Let Rand be the subset of V of vectors of the foem (2.2. 5). where and Pane any wal Famers. Is Wa stypace of V7 Use the following fare as commands to help you determine the ancwer. al = fix(10 + randn)s fix(10» randn)s bi = fix(10+ randn); 2 = fix(10+ randn) 2 al b1) w= [2 a2 b2] vew Bey MLZ Let V be P: and let I" be the subset of V of vectors cof the form ax? + bx +5, where @ and b are arbitrary real numbers. With each such polynomial in W we ‘associate a vector (a, b.5) in R*, Construct ‘commands like those in Exercise ML.1 to show that W is not a subspace of V. Before solving the following MATLAB exercises, you should hhave read Section ML. Use MATLAB to determine if vector v is a linear ‘combination of the members of set S. Ws. Va. ¥3) 1.1), (0,1,1,0), (1, 1,1, D) 0.1.1.1) (>) $= burns) ‘GME 8 MILA, Use Matin i detemine vs ater combination of the members of i wre Ge meee HSS ieee (@ Ws, Ya. ¥5) (1,2,1),8,0,1), (0,8 sd 3) Wy AW LOE Let yyy saris THO. Si Wh Hrservise THO Suppose tha ye Steg every vector HV cam bo amiquay tt HN Wely a whore a 481 WY a iy long sta awrite V = Wy WS and : PW a say thay 9 ig at the subspaces Wy ad Vy isthe ey Hd, Show thal the set oF all point ‘ Mts in the Dreb by bes = isa subspace a Use MATLAM to determine if vis combination of the members of set § th yin terms ofthe members ofS. Va) tinny ) S= (pil), Pr) =QP-1th v=p()=4eti-5. determine whether ble0B® ML6, In each par, where S= (vi. ¥2 Vb = {(1, 1,0, .( 1,-1.0, pet (a) v= (2.3.2.3) (b) v= 2,-3,-2.9) () v=(0,1,2.3). MLZ7, In each part, determine W . S, where if pesher poo getter rnc i yet he a (a) piypar eats (b) pit) =2P 40-2 (©) py =-2P +!EXAMPLE 1 Solution once 253 See. 6.) Linear tndep DEPENDENCE Thus far we have detined a mathematicul system called a real vector space and noted some of its properties, We further observe that the only real vector space having a finite number of vectors in it is the veetor space whose only veetor is 0, for itv #0 is in a vector space V, then by Exercise T in Section vy Ac'v, where ¢ and ¢" are distinct real numbers, ancl so V has infinitely many vectors in it, However, in this section and the following one we show that most vector spaces 1’ studied here have a set composed of a finite number of vectors that completely describe 1’, It should be noted thi poral, there is more than one stich set deseribing V. We now turn to a formulation of these id The vectors ¥y.¥2,...,¥_ in a vector space V are si ery vector in V is a linear combination of vj, ¥2.....¥. Moreover, if $ +s ¥afathen we also say that the set S spans Vor that (Vi. ¥2s V, or that V is spanned by $, or in the language of Section 6.2, span $ = V. The procedure to chec V isas follows the vectors v1, the vector space Step 1. Choose an arbitrary veetor v in V. Step 2. Determine if v is linear combination of the given vectors. If it iss then the given vectors span V. If i is not, they do not span V. Again we investigate the consistency of a linear system, but this time for aright side that represents an arbitrary vector in a vector space V. Let V be the vector space R? and let vp =(L20, v2= (10,2), and v3 = (1, 1,0). Dov, id ‘pan V? Step 1. Let v = (a,b,c) be any vector in R3, where a, b, and c are arbitrary real numbers. Step 2. We must find out whether there are constants ¢), €2, and c3 such that civ) teva + ea = This leads to the linear system (verify) cyt cp tea 2c Hey cr 2e2 A solution is (verify) —2a+2b+e a-b+e fab = 2 3 7 atte) ae A ae Since we have obtained a solution for every choice of a, b, and c, we conclude that v,, vp, v3 span R®. This is equivalent to saying that span {¥1, V2, Va} = R s a=254 Chayrer Real Vevtor Saces Solution CO Solution eof DE IE 9 of pons te sahgace of My eosin Fal ym yy sm acy step Bait synmetic mat sth form ab a-[f ‘h: where a,b, and c are any real numbers, ‘Step 2. We mus find constants dh dy and dy such thy 10 ot 00 als alt o}*]0 1 ==} A which leads toa linear systemn whose solution is (verify) d=a, b=), dae Since we have found a solution for every choice of a, b and¢ : Weng that S spans the given subspace. ki Let V be the vector space P. Let § = {p1(t), p2(t))}, where p,(t and px(t) 51° +2. Does $ span P;? hy Step 1. Let p(t) = at? + bt $ c.be any polynomial in P,, where a, al are any real numbers. Step 2 We must find out whether there are constants ¢, and cp suhtht P(t) = cipi(t) + copr(t) or af torts ei +2t +1) +exlt +2) Thus (ci +a)? + Qey)t + (c1 + 2c2) = at? +b +e . cies? Since two polynomials agree forall values of r only ifthe coef spective powers of r agree, we obtain the linear system at a= 2c, c+ 2c. 5c. , a it 2 ti Using clementary row operations on the augmented mattix tem, we abn (verify) 1 0} 2c O 1 c-a |. 0 0j)b-4a+2¢ 0% Itb~ a erant and Ga 2c oO, then the system is inconsisteM 2 te, we (P10), p2(t)} does not span Pp. For &% (o) : 1 ~ 1cannot be written asa linear combination ° 3rSe0.6.3 Linear Independence 24 1,0) and ey The vectors = was observall = Vectors insenaraclsir = (0, 1) span R®, for as was observel in 1.42) is any vector in R®, then u = Wey + Ss Was noted in Secti ‘ Section 4.2, every vee 7 conbinnion erie ever ‘tor uw in R* can be written as a linear k= 0.0). There 1.0.0), e2 (0.1.0), and e; (.0,....0),¢. al he ez and ey spin’. Similarly the vectors ane 62 = (0.1.0... Ohms r= (0,0, .... 1) span R®, since any / vector u = WS (Hy, wo. ett, in R" can be written as SMe) + ures FoF Uney . The set $= (rm 4, the form UN beep nat + any Jements in S. roi 0 2 Lt 2-2 1-5 By a ieenie ieee N 4 4-1 9. 4 From Example 8 in Section 6.2, the set of all solutions to Ax = 0 forms a subspace of R*. To determine a spanning set for the solution space of this homogeneous system, we find that the reduced row echelon form of the aug ‘mented matrix is (verify) coos ooo ooo 1 0 0 0 % ‘The general solution is then given by m= mer B mS, where r and s are any real numbers. In matrix form we have that any mem of the solution space is given by 1 : ee Hence the vectors |g | 2 0.MF 256 Chapter Real Vector S ae ee ce es Linear Independence Jr space V are said to be lineart y, in a vector ‘ 4, not all zero, stich that i ’: a <3 rhe vectors Vis 82s" ‘ | SSSRRNREDEETTTON FP re exis const __ eve teava bees Fave =O (W are called linearly independent. Thatis, yy Jent if whenever c1¥1 EVA FOV = 0, wena st ¥ depen, Otherwise Vie V2.2" xy are linearly independ have near combination of Vi, V2 +++ ¥s that yields the zero veto tit the coefficients are 7er0. 1S = (Vi. ¥2...., Vil theme Tinearly dependent or linearly independent i ye ty defined above, Thatis, the only iq is that in which ¢ also say that the set 5 is Tin vectors have the corresponding propert It should be emphasized that for any vectors v1, V2y+..4 V4. Equation (1) always holds if we choose all the scalars ¢1,€2,..-.¢k @qual to zero, ‘The important point in this definition is whether or not itis possible to satisty (1) with at least one of the scalars different from zero. The procedure to determine if the vectors Vi, V2. » Ve are linearly depen- dent or linearly independent is as follows. Step 1. Form Equation (1), which leads to a homogeneous system. Step 2. If the homogeneous system obtained in Step | has only the trivial solution, then the given vectors are linearly independent; if it has a nontriv- ial solution, then the vectors are linearly dependent. IAEA Determine whether the vectors =I 2 1 0 o| and ; 0. 1 found in Example 6 as spanni i dependent or linearly independent igen ree Solution Forming Equation (1), = 2 0" i 0. I 0. obtain the homogeneous system —C1 = 2c. = 0 C1 +0 =0 Or + 20 ! oe - %+ a=0, : S Whose only soluti : ion is cy = independent, a °2 = 0. Hence the given vectors are linea'ysam solution ee Solution SOON anen Iodependence 257 Ate the vectors Ys (1,0, linearly de V2 Vy ow Pendent or line, vam (Ot AND and vy ay in ee ily independengy ME YY = (1,1 1,3) in te We for Lqnation (1), OME OMA EW =O, and sOWVE HOF EL. andes. The Fesulting homapencous syste “ fos0 2 a+ 20 + + Ged 2ey + 2c) + ey = 0, which has as its only solution ¢, given vectors (2 = Cy = 0 (verify), showing that ae lincarly independent, . Consider the vectors, WEG, vee (L—21), > y= (3,21), and Va = (2,00) in R?, Is Se {V1 ¥2.-¥4, va) linearly dependent or linearly independent’? Setting up Equation (1), we are led to the homogeneous system (verify) Cr+ 0 -3ey +2 =0 2ey — 2c; + ey =0 -at a7 4 =0, homogeneous system of three equations in four unknowns. By Theorem 1.8, ction 1.5, we are assured of the existence of a nontrivial solution. Hence, $ is linearly dependent. In fact, two of the infinitely many solutions are a G=l c=0; o=0 c fe vectors ean ein 2, defined in Example 4, are linearly independent, since . c1(1.0) +.62(0, 1) = (0,0) can hold only if cy = cz =-0. Similarly, the vectors e1, €2, and e3 in R’, and more generally, the vectors €),€2.-...€, in R” are linearly independert (Exercise T-1). ‘ i v. gives another way of testin lary 6.4 in Section 6.6, 10 follow. gives another wa wtih even vectors in R” are linearly dependent or linearly independent We form the matrix A, whose columns are the given n vectors. Then the given vectors are linearly independent if and only if det(A) # 0. Thus. in Example i 4 and det(A) = 1 so that e; and e2 are linearly independent.258 Chapter6 Real Vector Spaces OS Consider the vectors artit2 2 (pil. To find out whether S = (P10 we set up Equation independent we Se.UP EQUA pins2r th p= 3 422 prlts poh is linearly dependent or linearly (1) and solve for ¢1. ¢2, and ¢3. The resulting pit) homogeneous system # cy + 2c) +303 = 0 qt ot 23=0 2e1 ‘+23 =0, lutions (verify). A particular solution is c which has infinitely many s 1,50 o=ho= =0. (Cpilt) + prt) ~ Ps Hence $ is linearly dependent. : re k vectors in any vector space and yj is the zero vector, O for j # i. Thus § sry set of vectors containing Tf vy Vee Me al Equation (1) holds by leting ¢) = 1 and J is linearly dependent. Henc {Vp Va 202 Mh the zero vector is linearly dependent. 7 Let 5; and be finite subsets of a vector space and! let Sy be a subset of Se, Then (a) if Sis linearly dependent, so is Sz: and (b) if Sis linearly independent, so is 5; (Exercise T.2) in R? and R°. Sup ‘We consider next the meaning of linear independen pose that vy and v3 ate linearly dependent in R°, Then there exist scars + and ¢2, not both zero, such that cy Fa If; #0, then If cy #0, then ‘Thus one of the vectors is a scalar multiple of th that vy = ev, Then iple of the other. Conversely, suppose “ Iv) = cv: = 0, and at the coeficints of v1 and v2 are not both zero, it follows tha v and ®t ney dependent. Ths van are linear dependent in R? if and ar lineaty dope eee multiple of the other. Hence two vectors in R? ident if and only if they both li ‘ame line passing through the origin (Figure Caan 'y both lie on the same line Pp: Suppose E ; ago Shpvos: now that Yi fa. and vsiare linearly dependent in RD, Then we : C1¥y + C2 + 03¥3 = 0, Where cy, ¢2, and c3 are not all zero, say cy # 0. Then(a) Linearly dependent vectors in R? which means that ¥> is in the subspace Sec. 6.3 Linear Independence £97 a (b) Lineatly independent vectors in R° W spanned by v; and v3. : Now W is either a plane through the origin (when v; and v3 are linearly independent), or a line through the origin (when v; and v3 are linearly depen- dent), oF the origin (when vy = v2 = vy = 0), Since a line through the origin always lies in a plane through the origin, we conclude that vj, v2, and v3 all lie in the same plane through the origin, Conversely, suppose that v1, V2, and ¥3 all lie in the same plane through the origin. Then either all three vectors are the zero vector, or all three vectors lie on the same line through the origin, or all three vectors lie in a plane through the origin spanned by two vectors, say vy, and v3. Thus, in all these cases, ¥> is a linear combination of v1 and vs: vy = ayvy +.aV3. Then ayy, — Iv) +.43v3 = 0, which means that y), ¥.and vs are linearly dependent. Hence three vectors in Rare linearly dependent if and only if they all lie in the same plane passing through the origin [Figure 6.5(a)]. Figure 6.5 > (a) Linearly dependent vectors in R “ a ¥ Be.0. (b) Linearly independent vectors in R’ More generally, let u and v be nonzero vectors ina vector space V. We can show (Exercise T.13) that u and v are linearly dependent if and only if there je a scalar k such that v = ku. Equivalently, w and v are linearly independent if and only if neither vector is a multiple of the other. This approach will not ‘work with sets having three or more vectors. Instead, we use the result given by the following theorem. MOTRIN | The nonzero vectors V1. ¥2 vnina vector space V are linearly dependent if and only if one of the vectors ¥j, J = 2 is a linear combination of the preceding vectors ¥1.¥2,-+-+¥)-1 Proof if; is. linear combination of ¥1.¥: then qatar Fea + Uy too tag ae \ ' 4 Vj cay Hegvn bee FEIN160 Chapter 6 Real Vector Spaces Since at least one coefficient, ~1, 1s NONZETO, We con, east one coefficient Since at least that ¥, ly dependent. ie is. are linearly a suppose tat V1,V2+--++Vn Ae linear, . & conve tags cy, C3, +++ ns HOt all 26F0, Such thay ead there exist seals C126 +++ ; cy FOV2 bo FAV, = 0, Xy Now let j be tangs! subscript for which cj 0.16 jy (e-em), Ai ' ' yhich implies that vy = 9, g j = 1, then cv) = 0, W 2 conten Hd thesis that none of the vectors are the zero Vector. Thus om "ction ree car abubinaton ofthe preceding vectors ve ohh jis a'linear cdmbinatio ctatio \ MRS EREN 165, v2.5, and vy are as in Example 9, then we find (sett hy vj ty¥2+0v3 — vy =0, 0,¥1,¥o,¥3,and vy are linearly dependent. We then have Vesvi tye. 1 Remarks 1. We observe that Theorem 6.4 does not sa that every vectar yy combination ofthe receding vectors. Thus, in Example 9, ye tat the equation v; + 2v2 + v3 + Ova = 0. We cannot solve, in this equine for vs as a linear combination of v1, ¥2, and v5, since in coef zero. i 2 Wecan also prove that if S = (v1. v2,..-, vi) isa set of vector ina ey space V, then $ is linearly dependent if and only if one of the ecto Sis. linear combination of all the other vectors in $ (see Exercise, For instance, in Example 13, WS=v)-0vy 45 and 3. Observe that if v1, vay... space, then they must be d Aare linearly independent vectors inayece istinet and none can be the zero vector. ‘ The following result will be used in Section 6.4 ‘as well as in several abet Places. Suppose that $ = {¥1, 3, .. + Yn} Spans a vector space V andy ist preceding vectors in S. Then the set SLE Wis ¥2s.-. Vit, Vjgte ree Vale itvisaneceeith¥s deleted, also spans V. To show this res ote suah an) Yorn Vs thn, since $ pans V, we en find Seas such that linear combination of the = A oo aie, Hajyy tayyavjni tote Now if bv + bevy teh DAY-L, then Yay tay 4.. Faivigy p.. byt’ Haj) 4 aj(byyy + bevy $20 tPF +ay¥p, E cane Sa beyyivjyn bee teot Sec.6.3 Linear Independence 261 Consider the set of vectors $ = (v1, v2, v3, va) in R4 where and let W = span S. Since vy = vj + v2, we conclude that W = span S 0 2 1 wel) 1 0 1 if) v=] yf. and 0, 0. 0, gine following ses of weetors span R? a 1.2.01 DE 1,2). (2.-5.4)). fg 12.236 0) (1.0.0 01,0), (0.0. DADE 4h ofthe following vectors span RY ¥.0,00.0. 0,1,0,0), 1, 11D. 1 10). (6) (1.2.1.0). (1,1, 1,0), 0.00, 1). 4 (1.0.0), (1,2. 1,20). ‘Liich ofthe following sets of polynomials span we+Le tert y) OPH Leta. +22 Let Vete say. Mee, =1,1), 00,1. $ Dot polynomials ° + 2t + 1,7 = 1 +-2.0° +2, WHE St42 span Be? Fa } a ‘of Vectors spanning the solution space of [ =0.where | H-ne 0 1 1 1 VFida set ‘of vectors spanning the null space of ea oe Az | 24134. 6-2 8 % 10. UL. 2 Let z m=|-1]. 1 belong to the solution space of Ax = 0. Is (x1, ¥2,%3} linearly independent? Let 1 2 malo, 2 oo 1 1 0. belong tothe null space of A. Is ($1, X2,%3} Finearly independent? Which of the following sets of vectors in R? are linearly dependent? For those that are, express one vector as a linear combination of the rest. Qa (a) (1,2. -1), GB, 2, 5))- (b) (4, 2, 1), (2,6, —5), , -2, 3))- ea (e) (1, 1,0), (0, 2, 3), (1, 2, 3). GB, 6, 6)}. Ov, a (@) (01.2.3). 0,1,0.0,0, D) a Consider the vector space R*, Follow the directions of Exercise 10. (a) {(1, 1,2, 1), (1,0,0, 2). (4, 6, 8, 6), (0.3.2, DE 2 (b) (1,-2.3, “1, (2,4, 6 2))- (© {G11 1D, 2,3.1,2). 8.1.2. De 2.21, DE @ (4,2,-1.3). 6.5,-5. 1, 2,-1.3, Consider the vector space P3. Follow the directions of Exercise 10. (@ (P40 - 2143). ) QP +12 +301 © Brt13F +1 +641 (@) (8 —4,5P — 516,37 — 51 +2) . Consider the vector space Mz2. Follow the directions of Exercise 10.262 Chapter6 Real Vector Spaces o} fo a, p2 & oll bi a6 11) ft o} fo r oll LG 3-6 3 11) f2 3) f3 1) p2 ill rb apt ap apf a 14. Let V be the vector space of al real-valued continuous functions. Follow the directions of Exercise 10. (@) {cosr, sins, e'). (©) (tee', sing). © Wane). (d) {cos*, sin’ t, cos 2), 15, For what values of athe recto Q2.1,2)and (1, 1c) in RY Nearly epee a 16, For what values of are the vector 2r+22 +2 in PLneanly dependents” 32M Theoretical Exercises TL. Show that the vectors e;, e, independent. T2, Let S; and Ss be finite subsets of a vector space and let Si be a subset of Ss. Show that: (a) If S, is linearly dependent, so is Sp. (b) IF Sy is linearly independent, sos 5, TS, Let $= (vi, ¥2,...,¥1) bea set of vectors ina vector Space. Show that Sis linearly dependent if and only if ‘one of the vectors in 5 is a linear combination of all the other vectors in S, TA. Suppose that $ = {y;, v2, v3} isa linearly independent Set of vectors in a vector space V. Show that T = (Wi, w2, Wa) is also linearly independent, where WIS VEY V5, Wa = V5 + ¥5,and Wy = Wy TS. Suppose that $= (¥,, v2, v3) isa linearly independent Set of vectors in a vector space V. Is T= (01, W2, Ws} where Wy = ¥, + ¥9, Wy Ws = vo +s, linearly dependent or linearly independent? Justify your answer. T6. Suppose that S = (vs, v2, va} i a linearly dependent Set of vectors in a yector space V. Is T = (1, Wo, 3), where wi = v4, wy Ms = Vi + ¥2-+ v5, linearly dependent or independent? Justify your answer, TF. Let, Yost ects faa Sector Spite th ha {¥1, va} is linearly independent. Show tha not belong to span {v, v9), then {v1, v2, vs} is linearly independent. TS. Let A be an m xn mat form. Show that the n 2 in RY are linearly WN, Mtv, linearly ix in reduced row echelon vonzero rows of A, viewed as MATLAB Exercises ML.1. Determine dependent, ‘Sis linearly independent or linearly @ S=(1,0,0,1,,1,1,0),(,1, Ly). mah JL a vectors in R", vectors. TO, Let $ = {uy.us....,u} space, and let T = {¥y, vs, .. 2 i m, is a linear coy in S. Show that form a linearly indepen den se tof bea set of vec Vim), wi bination of ge WH bv + bev. “tbayy is a linear combination of thé vecto Suppose that (V1. V2... ¥q) is alin setof vectors in R”. Show that if 4 ‘nonsingular matrix, then {Ay,, Avy linearly independent. ay inde isannxy ANDi TIL Let S\ and 5: be finite subsets ofa vector ga Vet let Si be a subset of Ss. If Ss lineary depen show by examples that S, may be either’ linearly dependent or linearly independent, TAZ, Let Sand 5: be finite subsets of a vetorspue ade Si be a subset of $s. IF 5 is linearly independex show by examples that Sy may be ether line dependent or linearly independent TAB. Let w and v be nonzero vectors in vector pu ‘Show, that {u, v) is Tinearly dependent if and ony there i a scalar k such that v= ku, Equi . (u, ¥) is linearly independent if and only if one Vectors is not a multiple ofthe other. TA. (Uses material from Section 5.2) aed linearly independent vectors in R?. Show th "I and u x v form a basis for R?. (Hint: Form (1) and take the dot product with u x ¥) 1 o xn? 2 1 : I ©s=f}i],] 2],] of. 1 o} |-1 oT 1 Wy L-1 a ae ce ML2. Finda spanning set of the solution 5°Sec. 6.4 Basis and Dimension 263 5 DIMENSION ~ In this section we continue our study of the structure of a vector space V by determining a smallest set of vectors in V that completely describes V. 5 The vectors ¥1, ¥2,.... V4 ina vector space V are said to form a basis for V if (a) Vy. V2... Ve span V and (b) v1, ¥3, .... Vg are linearly independent. i Remark If ¥).¥2.....¥s form a.basis for a vector space V, then they must be distinct and nonzero, so We write them asa set (¥1. ¥3...-. Ya) The vectors ¢) = (1.0) and e = (0, 1) form a basis for R?, the vectors e1, ez, and ey form a basis for R? and, in general, the vectors e), €2,..-,€n form a basis for R". Each of these sets of vectors is called the natural basis or standard basis for R, R’, and R", respectively. . Re Show that the set $= (vj.¥2.¥3.¥a)y Where vy) = (1,0,1,0), v2 = (0, 1. =1.2).¥9 = (0.2.2, D),and va = (1,0, 0, 1) isa basis for Rt. Solution To show that S is linearly independent, we form the equation t civ Fea Fea + eaVy = and solve for ¢1, 3, €3. and cs, Substituting for vi, ¥2..¥3, and vs, we obtain the linear system (verify) a +c=0 en $20 cy — ¢2 +203 = 2+ Qta= Which has as its only solution cy = ¢2 = ¢s = €s = 0 (verify), showing that $ is linearly independent. : To show that § spans R¥. we let v = (a, b, c,d) be any vector in RY We now seek constants ky, Az. k3. and ky such that cc [am Foyt b+ a nd a solution (verify) for ki, %2, Kx. Substituting for vy, v2. V3. ¥4- and v, we find a solution (ver? ; and ky to the resulting linear system for any a,b,c, and d: Hence S spans x and is a basis for R*.264 Chapter 6 Real Vector Spaces as) Solution f the P_EXAMPLE| Find a basis for the subspace V of P,, consisting ofall vectors of" ‘Show that the set § +21 isa basis for the vector, TP, Todo this, we must show that § spans V” and is Tinearly independeny 3° that it spans V7, we take any vector in V, that is, a polynomial gy bey and must find constants a), a3, and ay such that Hey at thttesale tl) tat — 1) +2142) = ay $ (ay $+ 2ay)t + (a ~ a) +24), Since two polynomials agree forall values oft only if the coetcien spective powers off agree, we get the linear system * a a a + 2a js | Solving, we have a Hence S spans V. r To illustrate this result, suppose that we are given the vector 27°-+6r4.18 Here a = 2, b = 6, and c = 13, Substituting in the foregoing expressions fr 4}, a», and 3, we find that a =2, Hence 24613 =2P +--+ Lore). To show that S is linearly independent, we form a(t? +1) +a 1) +as2r +2) Then aut” + (a) +2a3)1 + (a, — a) +2a3) = 0. Again, this can hold for all values of ¢ only if a 0 a) + 2a, =0 a = ay + 2a; = 0. The only solution to this homogeneous system is a implies that $is linearly independent. Thus S is a basis for P2 The set of vectors (7,1... 151) forms a basis forthe vet P, called the natural, or standard basis, for Py. It has already be in Example 5 of Section 6.3 that this set of vectors spans P,. Lineat dence is left as an exercise (Exercise 7.15). i at’ + bt +c, where c =a—b. 4 : 4THEOREM 6.5 Proof See 64° Hasisvand Dimension 265 a? + bt--a—b which can be written as a(? +1) + b0 = 1), , 24a so the ie +1 and 1 — 1 span V. Moreover, these vectors are linearly independent because neither one is a multiple of the other, ‘This conclusion could also be reached (with more work) by writing the equation a(? +1) +ax¢-1) =0 or Pay +1az + (ay —a) = 0. Since this equation is to hold for all values of ¢, we must have a = 0 and ow) 1. A vector space V is called finite-dimensional if there is a finite subset of V that is a basis for V. If there is no such finite subset of V, then V is called infinite-dimensional, Almost all the vector spaces considered in this book are finite-dimen- sional. However, we point out that there are many infinite-dimensional vector spaces that are extremely important in mathematics and physics; their study lies beyond the scope of this book. The vector space P, consisting of all poly- nomials, and the vector space C(—o0, 00) defined in Example 7 of Section 6.2 are not finite-dimensional. We now establish some results about finite-dimensional vector spaces that will tell about the number of vectors in a basis, compare two different bases, and give properties of bases, First, we observe that if (Vv), v2)... Vi) isa basis for a vector space V. then (evs, ¥2 is also a basis when ¢ # 0 (Exercise T.9). Thus a basis for a nonzero vector space is never unique. Sa (vv Va) is a basis for a vector space V, then every vector in V can be written in one and only one way as a linear combination of the vectors in. First, every vector v in V can be written as a linear combination of the vectors in S because S spans V. Now let Very) FOV H+ + CnVn qd) v= div + dav +--+ dnVn- 2) Subtracting (2) from (1), we obtain 0 = (ey — diy + (c2 — da)¥2 4 + (Gn — And ¥oe Since S is linearly independent, it follows that ¢; ~ di = 0.1 = i <7. so ci: = dj, 1 m. We take the vectos in Sas m x 1 matrices and form Equation (3) above, This equation leads to a homogeneous system in the m unknowns ¢1, ¢2, «+4 Gy; the columns of is m xn coefficient matrix A are v1, ¥2..... Vy. We now transform A toa matt, B in reduced row echelon form, having r nonzero rows, | z 4 Tis the solution space to Ax = 0. Th is the set of all solutions to Ax = b. ‘The following result, which is important in the study of differential equa- tions, was presented in Section 1.5 and its proof left to Exercise T.13 of that section. If xp is a particular solution to the nonhomogeneous system Ax = b, b £0, and x; is a solution to the associated homogeneous system Ax = 0, then xp + x, is a solution to the given system Ax = b. Moreover, every solution x to the nonhomogeneous linear system Ax = b can be written as Xp-+Xq, where x, isa particular solution tothe given nonhomogeneous system and Xj is a solution to the associated homogeneous system Ax = 0. Gide | Consider the linear system xy+2x, — Baa 3s =2 xy + 2x. +43 — 3x4 + 5 + 2x6 =3 x) + 2x2 —3xyt 2xs+ 16 =4 3xy + Oxy + x3 — 9x4 + 445 + 3x6 =9, which was defined in Example 9 in Section 1.5. There we found that a solution to the given linear system is given by r43s—2] PO] [r+3s—20 t 0 t 1-2 us| t -2r 13 s = olds: se |p 2-1 2 r 0. r , where r, s, and ¢ are any real numbers. Let 3s 21 % ; ar and =| 5 Xp= -r r i jcular solution to the given system Then x = xp + Xs. Moreover, xp is a particull! » a eous system. (verify) and x, is a solution to the associated homogeneous sys jSneo-o91. Let 2-1 A=|-4 2 4]. 8 4 8 (a) Find the set of all solutions to Ax = 0. (b) Express each solution as a linear combination of (wo vectors in R*. (©) Sketch these vectors in a three-dimensional coordinate system to show that the solution space is aplane through the origin. 2 Let 11 -2 A=]-2 -20 4]. -1 -1 0 2 (a) Find the set of all solutions to Ax = 0. (b) Express each solution as a linear combination of two vectors in R°. (©) Sketch these vectors in a three-dimensional coordinate system to show that the solution space is a plane through the origin. In Exercises 3 through 10, find a basis for and the dimension of the solution space of the given homogeneous system. 3 ntmtutey= a afi ch 2 2 1/2] _fo “[3 -3 2 0 2)/8)= of x 8. mt+2n- 5 4+3m=0 2x, + 2xp— x3 + Dee x + 3x5 $35 =0. 6m mt t3u +415 =0 =x) + 2g + 3xy + 4g + 5x5 = 0 y= x $325 + Sty + 645 =0 3x) — 4x9 + 53 + Day + Bas = 0. 1 2 og qi[e a. |¢ 2 gobo 2} 12} _|o P12 4 Bp 3s 3h ol oy i} | 9 oi 1-1/3) Lo 1 0 2] fa 0 a]2 1 3} }nl=]o]. 3 1 2) bs. 0 x 1 2 2) -i 1 0 oO pae2 -2 -1|12]|_|o % 16 2 4 IIE I fol 1 x40. 3 OL 0. 1 23) see 0 wo. |-2 -4 6 4 -3 0 0 254 1 2-3 -2 0 merase given matrix A. 1 2 3-1 2) 34,2 0 WAa|y 4. t roi-lo H-1 201 2 0 1-1 3 ma=|5 -1 3 0 3). 4-25 1 3 1 3-4 -5 6, In Exercises 13 through 16, find a basis for the solu ‘space of the homogeneous systent (il, — A)x = 0 fo sgiven scalar 2 and given matrix A. 3 2 Bastaq[} Ht 4 Wie 4=[ 2 O30 1 .A=1,A=/{1 0 3]. ORS E 3 11 -2 W6.4=3,A=|-1 2 1 o 1-1 the homogeneous system (Al, — A)X = has a nontri solution, _f2 3 wan[} 3} 0 0 0 -2 0 %A=/0 1 -1/. 20A=] 0 -2 1 030 o 4 In Exercises 21 and 22, solve the given linear system @ write the solution x as X = Xp + Xp, where Xp is a parn® solution to the given system and x, is a solution 10 the associated homogeneous system. 3 -2 We xt2y- 2- we x yt3z42w 2e— ytdet aw 2x — 2y +82 + 6w = —4, ox yt+2e+2w Te 4 2y 432 4 2w Ax+2vt zSec.6.6 The Rank of a Matrix and Applications 283 oretical Exercises $= [xieX2.---- 4] be a set of solutions toa geneous system AX = 0, Show that every vector ‘Sis solution to Ax = 0. (b) Let A be a nonzero 3 x 3 matrix and suppose that ‘Ax = O has a nontrivial solution. Show that the dimension of the null space of A is either 1 or 2. that if the n x n coefficient matrix A of the T.4, Matrices A and B are m x and their reduced row Fomogencous system AX = 0 has a row or column of ‘echelon forms are the same. What is the relationship ~ zeros, then AX = 0 has a nontrivial solution, between the null space of A and the null space of B? (2) Show that the zero matrix isthe only 3 x 3 matrix TS. (0) Face null space has dimension 3. _Marias Exercises jeBsercses MLL] through ML, use MATLAB's rref MLA, For the matrix se rpand 10 aid in finding a basis for the null space of A. a may also use routine homsoln. For directions, use help. A= [2 | and A = 3, the homogeneous system (A/ ~ A)x = 0 has a nontrivial solution, Find such a solution using MATLAB commands. MLS. For the matrix 1 2 3 A=||3 ag2pes 201 3 and A = 6, the homogeneous linear system (Als — A)x = 0 has a nontrivial solution. Find such a solution using MATLAB commands, In this section we obtain another effective method for finding a basis for a vector space V spanned by a given set of vectors S = (V1, V2, .++.Vq). In the proof of Theorem 6.6 we developed a technique for choosing a basis for V that is a subset of S, The method to be developed in this section produces a basis for V that is nof guaranteed to be a subset of S. We shall also attach a unique number to a matrix A that we later show gives us information about the dimension of the solution space of a homogeneous system with coefficient matrix A. Let au ay et 22) azz sts A=|, Am\ my ++ be an m x n matrix. The rows of A, WS ai... ain) V2 = (anda Ay) Vin = (ont indy os Ginn)284 Chapter 6 Real Vector Spaces Bi seta Proof Remark Solution considered as vectors in R", span a subspace OF RY called the row sag ‘A. Similarly, the columns of A, ‘au a yy’ a az 2 o mel. |> w=]: We = | He Am). m2. Ayn. considered as vectors in R™, span a subspace of R™, called the column pq of A. IFA and B are rwo m xn row equivalent matrices then the row spaces of and B are equal. IFA and B are row equivalent. then the rows of B are obtained from those ‘A by finite number ofthe three elementary row operations. Thus each of B isa linear combination of the rows of A. Hence the row space of 8 contained in the row space of A. Similarly, A can be obtained from B by finite number of the three elementary row operations, so the row space of A contained in the row space of B. Hence the row spaces of A and B are equ Fora related result sce Theorem 1.7 in Section 1.5. It follows from Theorem 6.10 that if we take a given matrix A and tan form it to reduced row echelon form B, then the row spaces of A and B a equal. Furthermore, recall from Exercise T.8 in Section 6.3 that the nom rows of a matrix that is in reduced row echelon form are linearly indepen nd thus form a basis for its row space. We can use this method to find ab for a vector space spanned by a given set of vectors in R", as ‘iMustrated i Example 1. Let S = {¥1, V2, V3, Va}, Where v, = (1, -2, 0,3, 4), v2 = (3,2,8,1,4), vy = (2, 3,7, 2,3),° v4 = (-1,2,0,4,-3), and let V be the subspace of R® given by V = span S. Find a basis for V- Note that V is the row space of the matrix A whose rows are the given 1-2 0) 3 3°92 8 1 4 lg? 3 2 -1 2 0 4 -3 ‘ vealeo Using elementary row operations, we find that A is row equivaleat © trix (verify) 1 I B= a coro corn o-oo 1 0 0 0p See. 6.6 which i tical, and Hence ‘The Rank of a Matrix and Applications 285 Diese row echelon form. The row spaces of A and B are iden- a basis for the row space of B consists of the nonzero rows of B. wi = (1,0,2,0,1), wa = (0,1,1,0,1), and ws = (0,0,0,1,—1) form a basis for V. Ld A We may now summarize the method used in Example 1 to find a basis for the subspace V of R" given by V = span S, where S is a set of vectors in R". 2 ee The procedure for finding a basis for the subspace V of R” given by V = span S, where S = (vj, V2,..., Ve) is a set of vectors in R" that are given | in row form, is as follows. | Step 1. Form the matrix whose rows are the given vectors in S. Step 2. Transform A to reduced row echelon form, obtaining the matrix B. Step 3, The nonzero rows of B form a basis for V. Of course, the basis that has been obtained by the procedure used in Ex- ample 1 produced a basis that is not a subset of the spanning set. However, | the basis for-a subspace V of R” that is obtained in this manner is analogous, in its simplicity, to the standard basis for R". Thus if v = (a1, 42, -.- dx) is a vector in V and {¥j, V2, ..-. Vk) is a basis for V obtained by the method of Example 1 where the leading 1’s occur in columns ji, ja...» je, then it can be shown that Vs aj Vi +a),V2t-- +a, %E- v= (5,4, 14,6, 3) is in V, write v as a linear combination of the basis determined in Example 1. Solution We have ji = 1, j2 = 2, and js = 4,80 v = 5w, + 4w2 + 6ws . Remark The solution to Example | also yields a basis for the row space of matrix A in that example. Observe that the vectors in this basis are not rows of matrix A jimensit is \d the EDENTON ension of the row space of A is called the row rank of A, an ee f A is called the column rank of A. dimension of the column space o| TEI i ‘A defined in the solution of Ex- .d.a basis for the row space of the matrix Sa F that contains only row vectors from A, Also compute the row rank of the matrix A.286 Chapter Real Vector Spaces Solution AIMEE Find a basis for the column space of the matrix A defined in the solution et Solution 1 Using the procedure inthe altemative proof of Theorem 6.6, ye g equation q(1,-2.0,3, 4) +28. 2.8.1.4) $052.3, 7. 2.3) + cx(-1. 2.0.4, -3) = whose augmented matrix is = [aT 0]; wre ww te be owe ooooc that is, the coefficient matrix is A’. Transforming the augmented [AT £0] in (1) to reduced row echelon form. we obtain (verify) Mat 1 0 1 ooo Co - ZS 0 0 0 0 o 0 oO: Since the leading 1’s in (2) occur in columns 1, 2, and 3, we conclude that te first three rows of A form a basis for the row space of A. That is, (C1, -2, 0,3, -4), (3, 2,8, 1.4), 72,3) is.a basis for the row space of A. The row rank of A is 3. 1 Example 1 and compute the column rank of A. Writing the columns of A as row vectors, we obtain the matrix AT, shi when transformed to reduced row echelon form is (as we saw in Example3) 1 0 0 o 1 0 oo 1 0 0 0 oo Thus the vectors (1,0,0, #4), (0,1, 0, — the row space of AT. Hence the vectors 1 0 0 1 o}" 0 HJ L-8 on ea for the column space of A and we conclude that the °°Solution 2 edit kL See.6.6 The Rank ofa Matrix and Applications 287 If wi i pawels eae ie find a basis for the column space of A that contains only the eo fectors from A, we follow the procedure developed in the proof of jeorem 6.6, forming the equation [ 2) 0" 3 4" 0" 2 8 1 4 a} a]te] Zfte|Sl+eals}re| 3]=19 1 2 0. 4 -3! Lo whose augmented matrix is [A : 0]. Transforming this matrix to reduced row echelon form, we obtain (as in Example 1) 1 0 2 0 o 1 1 0 oo 0 1 0 0 0 0 Since the leading 1's occur in columns 1, 2, and 4, we conclude that the first, second, and fourth columns of A form a basis for the column space of A. That is, 1) 7-27 73 3 2] fa 2 3] ]2 -1 2} La. is a basis for the column space of A. The column rank of A is 3. . From Examples 3 and 4 we observe that the row and column ranks of A are equal. This is always true, and is a very important result in linear algebra. We now tum to the proof of this theorem. The row rank and column rank of the m xn matrix A = [aij] are equal. Let wy, W2,--+.Wn denote the columns of A. To determine the dimension of the column space of A, we use the procedure in the alternative proof of ‘Theorem 6.6. Thus we consider the equation yw, + 02W2 $0 + CW =0. We now transform the augmented matrix, [A : 0], of this homogeneous sys tem to reduced row echelon form. The vectors corresponding to the columns Containing the leading 1's form a basis for the column space of A. Thus the Column rank of A is the number of leading 1's. But this number is also the number of nonzera rows in the reduced row echelon form matrix that is fow equivalent to A, soit is the row rank of A. Thus row rank A olumn rank A. . Since row rank A = column rank A, we now merely refer to the rank of an m x n matrix and write rank A. We next summarize the procedure for computing the rank of a matrix. ‘The procedure for computing the rank of the matrix A is as follows. Step 1. Using elementary row operations, transform A t reduced row echelon form. Step 2. Rank A = the number of nonzero fOWS of B. oa matrix B in