MTH 311: Advanced Linear Algebra Semester 1, 2020-2021 Dr. Prahlad VaidyanathanContents |. Prel 1 2 3. 4. inaries Fields Matrices and Elementary Row Operations Matrix Multiplication Invertible Matrices Il, Vector Spaces Bee Definition and Examples Subspaces Basos and Dimension Coordinates Summary of Row Equivalence IIL. Linear Transformations eRene Linear Transformations ‘The Algebra of Linear Transformations Isomorphisi Representation of Transformations by Matrices Linear Functionals ‘The Double Dual The Transpose of a Linear Transformation 1V. Polynomials ween Algebras Algebra of Polynomials Lagrange Interpolation Polynomial Ideals Prime Factorization of a Polynomial V. Determinants 1 2 3. 4 Commutative Rings Determinant Functions Permutations and Uniqueness of Determinants Additional Properties of Determinants VI.Elementary Canonical Forms 1 2 Introduction . Characteristic Values ae 18. 22 27 35 40 44 44 48 ST 58 67 72 83 83 85 88 90 99 104 104 105 i 116 123 123 1233. Annihilating Polynomials 4. Invariant Subspaces 5. Direct-Sum Decomposition 6. Invariant Direct Sums 7. Simultaneous Triangulation; Simultaneous Diagonalization 8. The Primary Decomposition Theorem ViIThe Rational and Jordan Forms 1. Cyclic Subspaces and Annihilators 2. Cyclic Decompositions and the Rational Form, 3. The Jordan Form Vilinner Product Spaces 1. Inner Products 2 Inner Product Spaces 3. Linear Functionals and Adjoints 4 Unitary Operators 5. Normal Operators IX. Instructor Notes 132 ul 151 155 161 164 170 170 175 189 196 196 201 212 218 226 231|. Preliminaries 1. Ids ‘Throughout this course, we will be talking about “Vector spaces”, and “Fields”, The definition of a vector space depends on that of a field, so we begin with that Example 1.1. Consider F = R, the set of all real numbers. It comes equipped with two operations: Addition and multiplication, which have the following properties: (i) Addition is commutative zty for all ,y €F (ii) Addition is associative rt(ytz)=(e+y) +2 for all x,y,z € F. (iii) There is an additive identity, 0 (zero) with the property that r+0=0+n=2 for all x € F (iv) For each x € F, there is an additive inverse (—x) € F which satisfies x+(-2)=(-2)42= (v) Multiplication is commutative ry = ye for all c,y € F (vi) Multiplication is associative (yz) = (xy)2 for all x,y,z F (vii) There is a multiplicative identity, 1 (one) with the property that tl=lr=x for all x € F(viii) To each non-zero x € F, there is an multiplicative inverse x7! € F which satisfies (ix) Finally, multiplication distributes over addition aly +2) =ay=22 for all x,y,z € F. Definition 1.2. A field is a set F together with two operations Addition : (x,y) 2+ y Multiplication : (2, y) 4 2y which satisfy all the conditions 1.1-1.9 above. Elements of a field will be termed scalars. Example 1.3. (i) F = Risa field (ii) F =C isa field with the usual operations Addition : (a + ib) + (c+ id) = (a+c) +i6+4d), and Multiplication : (a + #b)(c-+ id) := (ae — bd) + i(ad + be) (ai) F = Q, the set of all rational numbers, is also a field. In fact, Q is a subfield of R (in the sense that it is a subset of IR which also inherits the operations of addition and multiplication from R). Also, R is a subfield of C. (iv) F is not a field, because 2 € Z does not have a multiplicative inverse. Standing Assumption: will either be R or C, unless stated otherwise For the rest of this course, all fields will be denoted by F, and 2. Matrices and Elementary Row Operations Definition 2.1. Let F be a field and n,m € N be fixed integers. Given m scalars Um) € F™ and nm elements {a,j :1< i n. But R has n rows, so r (@i): If A is row-equivalent to the identity matrix, then R = J in the above equation. ‘Thus, A = P~!, But the inverse of an elementary matrix is again an clementary matrix, Thus, by Theorem 4.3, P 1 is also a product of elementary matrices. (it) = (): This follows from Theorem 4.4 and Theorem 4.3. ‘The next corollary follows from Theorem 4.6 and Corollary 3.7. Corollary 4.7. Let A and B be m xn matrices. Then B is row-equivalent to A if and only if B= PA for some invertible matrix P Theorem 4.8. For ann xn matrix A, the following are equivalent (i) Ais invertible. (it) The homogeneous system AX = 0 has only the trivial solution X = 0. (iit) For every vector Y € F*, the system of equations AX = has a solution, Proof. Onco again, we prove (i) => (it) = (i), and () = (i) > (7) (i) = (@): Let B= A“ and X be a solution to the homogeneous system AX = X =X =(BA)X = B(AX) = B(0) =0 , then Hence, X = 0 is the only solution, (is) = (®): Suppose AX = 0 has only the trivial solution, then A is rov-equivalent to the identity matrix by Theorem 2.11, Hence, A is invertible by Theorem 4.6 (i) = (Wi): Given a vector Y, consider X == ATI, then AX = Y by associativity of matrix multiplication. (iis) > (i): Let R be a row-reduced echelon matrix that is row-equivalent to A. By Theo- rem 4.6, it suffices to show that R= J. Since R is a row-reduced echelon matrix, it suffices to show that the n‘® row of R is non-zero. So set Y =(0,0,...,1) ‘Then the equation RX = Y has a solution, which must necessarily be non-zero (since ¥ # 0). Thus, the last row of R cannot be zero. Hence, R = I, whence A is invertible 16Corollary 4.9. A square matrix which is either left or right invertible is invertible. Proof. Suppose A is left-invertible, then there exists a matrix B so that BA = J. If X is a vector so that AX = 0, then X = B(AX) = (BA)X = 0. Hence, the equation AX =0 has only the trivial solution. By Theorem 4.8, A is invertible Now suppose A is right-invertible, then there exists a matrix B so that AB =I. If is any veetor, then X := B(Y) has the property that AX = Y. Hence, by Theorem 4.8, Ais invertible. a Corollary 4.10. Let A = AjAz... Ay where the A; aren xn matri invertible if and only if each A, is invertible. Then, A is Proof. If each A; is invertible, then A is invertible by ‘Theorem 4.3. Conversely, suppose Ais invertible and X is a vector such that A,X = 0, then AX = (AAz... An a)ARX = Since A is invertible, this forces X = 0. Hence, the only solution to the equation AtX = 0 is the trivial solution. By Theorem 4.8, it follows that Ay, is invertible, Hence, AjAy.. Ay, = AAG? is invertible, Now, by induction on k, each A, is invertible for 1 V given by (a, 8) Hat (Scalar Multiplication): F x V > V given by (ca) +9 ca with the following properties: (i) Addition is commutative at+s=Bta for alla, BEV (ii) Addition is associative at (G49) =(0 +8) +9 for all a, 8,7 €V (ii) There is a unique zero vector 0 € V which satisfies the equation a+0=0+a=0 for alla eV vector a € V, there is a unique vector (—a) € V such that (iv) For a+(-a) =(-a) +a=0 (v) For each ae V, (vi) For every c1,c2 € Fand a €V, (crea) = ea(era) (vii) For every c € F and a, 8 € +8) =ca+ep (viii) For every ¢1,¢2 € F anda eV, (ata)a = qa + oa 18.An element of the set V is called a vector, while an element of F is called a scalar. Technically, a vector space is a tuple (V, F, +,-), but usually, we simply say that V is a vector space over F, when the operations + and - are implicit. Example 1.2. (i) The n-tuple space F": Let F be any field and V be the set of all n-tuples (1,22 =) whose entries x, are in F. If 3 = (y1, y2,-..,Yn) EV. and c € F, we define addition by + B= (ey +My F2 + as En + Yn) and scalar multiplication by 6: (cx1, €%2,..-, Cty) One can then verify that V = F* satisfies all the conditions of Definition 1.1 (ii) The space of mx n matrices P™": Let F be a field and m,n € N be positive integers. Let F*" be the set of all m x n matrices with entries in F. For matrices A,B & F™, we dofine addition by A (A+B), a5 + Bas and sealar multiplication by (cA)ig = cig for any c € F. [Observe that FO = F” from the previous example] (ai) The space of functions from a set to a field: Let F be a field and Sa non-empty set. Let V denote the set of all functions from § taking values in F. For f,g € V, define (f + a)(s) = S(s) + 9s) whore the addition on the right-hand-side is the addition in F, Similarly, scalar multiplication is defined pointwise by (cf\(s) = ef(s) which the multiplication on the right-hand-side is that of F. Once again, it is easy to verify the axioms (note that zero vector here is zero function) If S = (1,2,....n}, then the fmnetion f : S + F may be identified with a tuple (f(1), ((2),---. f(n)). Conversely, any n-tuple (23, 29,...,2) may be thought of as a function. This identification shows that the first example is a special case of this example. ¢ Similarly, if S$ = {(i,j) :1 F may be identified with a matrix A € FP" where Ajj -= f(i,j). This identification is a bijection between the sot of functions from S'—» F and the space F”*". ‘Thus, the second example is also a special case of this one 19(iv) The space of polynomial functions over a field: Let F be a field, and V be the set of all functions f: F + F which are of the form F(z) = for some scalars ¢9,¢1,-..,¢, € F. Such a function is called a polynomial function. With addition and scalar multiplication defined exactly as in the previous example, V forms a vector space. (v) Let C denote the set of all complex numbers and F = R, Then C may be thought of as a vector space over IR. In fact, C may be identified with R?. cg tart... tenn" Lemma 1.3. (i) For any F, a =0 where 0 € V denotes the zero vector. (it) Ife € F is a non-zero scalar and a € V such that ca =0 Then « (iit) For any a €V (De =-a Proof. (i) For any c € F, 040=c0=c0+0)=0 40 Hence, 0 = 0 (i) If € F is non-zero and a € V such that ‘Then, But Hence, a = 0 fii) For any a €V, R Vy, a+ (-Ia= la + (la = (1+ (-1))a=0a=0 But a + (a) = 0 and (~a) is the unique vector with this property. Hence, (“1a = (-a) 20Remark 1.4. Since vector space addition is V, we have sociative, for any vectors a1, a2, 03,04 € ay + (a2 + (a3 + @41)) can be written in many different ways by moving the parentheses around. For instance, (oa + a2) + (a3 +04) denotes the same vector. Hence, we simply drop all parentheses, and write this vector as Gy > Og + 03 + Oy ‘The same is true for any finite number of vectors a1, a2,...,0% € V, so the expression ay + a2 + + on denotes the common vector associated to all possible re-arrangements of parentheses. The next definition is the most fundamental operation in a vector space, and is the reason for defining our axioms the way we have done. Definition 1.5. Lot V be a vector space over a field F, and ay,02,...,0 € V. A vector § € V is said to be a linear combination of a1,a2,...,aq if there exist scalars G1, €y--4€q © F such that e104 + 202 +... + Cn When this happens, we write Note that, by the distributivity properties (Properties (vit) and (viii) of Definition 1.1), we have eas + Y das = Wlce + deo a mi ‘ (So) = Desde a r= Exercise: Read the end of {Ilofiman-Kunze, Seotion 2.1] concerning the geometric in- terpretation of vector spaces, addition, and scalar multiplication 212. Subspaces Definition 2.1. Let V be a vector space over a field F. A subspace of subset WoC V which is itself a vector space with the addition and scalar multiplication operations inherited from V. Remark 2.2. What this definition means is that WC V should have the following properties (i) Ifa, 8 € W, then (a+ 8) must be in W. (i) fa € W and ce F, then ca must be in W. We say that W is closed under the operations of addition and scalar multiplication. Theorem 2.3. Let V be a vector space over a field F and W CV be a non-emply set. Then W is a subspace of V if and only if, for any a,8 € W and c & F, the vector (ca + 8) lies in W Proof. Suppose W is a subspace of V, then W is closed under the operations of scalar multiplication and addition as mentioned above. Hence, if a, 8 € W and c € F, then ca € W, so (ca + 8) € W as well. Conversely, suppose W satisfies this condition, and we wish to show that W is subspace In other words, we wish to show that W satisfies the conditions of Definition 1.1. By hypothesis, the addition map + maps W x W — W, and the scalar multiplication map maps F x W to W (i) Addition is commutative because it is commutative in V. (ii) Addition is associative because it is associative in V (ii) V has a zero clement 0 € V. To see that this vector lies in 1, observe that W is non-empty, 50 it contains some vector a € W. Then 0a = 0 € W by Lemma 1.3. (iv) If € W, then @ € V, so there is a unique vector (~a) € V so that But we know that 0 € W, so by Lemma 1.3 once again, (-a) =0- (a) =0+4(-l)aeW (v) For each a € W, we have 1-a@ = @ in V. But the scalar multiplication on W is the same as that of V, so the same property holds in W’ as well. (vi) The remaining three properties of a vector space are satisfied in W because they atisfied in V (Check!) are 22Example 2.4. (i) Let V be any vector space, then W = {0} is a subspace of V. Similarly, W := V is a subspace of V. These are both called the trivial subspaces of. (i) Let V = F as in Example 1.2. Let {(2,225+-+43) €V ex, = OF Note that if a = (03, 22,...,2n),9 = (Yt. Ya-+sYn) € W and c€ P, then n= =0Sen+y=0 Hence, (ca +8) € W. Thus W is a subspace by Theorem 2.3. Note: If F = R and n= 2, then W defines a line passing through the origin (ii) Let V =F as before, and let W = {(tayta.-0yt5) € Vern = 1} ‘Then W is not a subspace Note: If F =R and n =2, thon W defines a line that does not pass through the origin: (iv) Let V denote the set of all functions from F to F, and let W denote the set of all polynomial functions from F to F. Then W is subspace of V (v) Let V = F™" denote the set of all n x n matrices over a field F. A matrix A € V is said to be symmetric if Ais = Ay for all 1 < i,j 8; a = Thus, ca + 8 is also of the form in Equation I1.7, and so ca + 8 € W. So, by Theorem 2.3, W is a subspace of V a(ii) If L is any other subspace of V containing S, then Wc L. Proof. If 8 € W, then there exists ¢, € F and a; € S such that Yaa Since L is a subspace containing S, a; € L for all 1 < i 0, define f, € V by In(z) = ‘Then, W is the subspace spanned by the set { fo, fir fs ---} Definition 2.10. Let 51, $2,...,S, be k subsets of a vector space V. Define Sit Soto. +S to be the set consisting of all vectors of the form ay Fag bo HOE where a € S; for all 1 0, define fn €V by JIn(z) = 2" ‘Then, as we saw in Example 2.9, § >= {fo, fi, fay---} is a spanning set. Also, if €0,€2;--+4¢e € F are scalars such that Srau-o ‘=o ‘Then, it follows that the polynomial co + ye + cpa? +... + cyt is the zero polynomial. Since a non-zero polynomial can only have finitely many roots, it follows that c; = 0 for all 0 m. Since {(;,42,-.., 8m} is a spanning set, there exist sealars {A,, > 1 Ere) But the {a;} are a basis, so we conclude that a 1o:k=j PQ = iQ {; kAG Thus, if Q = (Q,,), then we conclude that PQ=1 Hence the matrix P chosen above is invertible and Q = P~!, The following theorem is the conclusion of this discussion 37Theorem 4.5. Let V be an n-dimensional vector space and B and B' be two ordered bases of V. Then, there is a unique n x n invertible marie P such that, for any ainV, we have lols = Plalsr and [ols = Pols Furthermore, the columns of P are given by P, 3 = (Ble Definition 4.6. The matrix P constructed in the above theorem is called a change of basis matrix. (End of Week 2) The next theorem is a converse to Theorem 4.5. Theorem 4.7. Let P be ann xn invertible matrix over F. Let V be an n-dimensional vector space over F and let B be an ordered basis of V. Then there is a unique ordered basis B' of V such that, for any vector a €V, we have [ale = Ploler and lel = P~ [al Proof. We write B= {ay,02,...,a,} and set P = (P,;). We define (11.6) ‘Then we claim that B = {(3;, 3,...,8,} is a basis for V. wealars ¢; € F such that (i) B'is linearly independent: If we have ‘Then we get Rewriting the above expression, and using the linear independence of B, we con- clude that » DY Pisces a for cach 1 }, then we have PQ = I, so ba Qu= {' ify ij ‘Thus, from Equation I1.6, we get YS ih = OY ns Pines rat ijt => (5 0) ij ya \iet a Hence, if a € V as above, we have o= Stee PY ada =F (Heeu,) Br (Ls) a ae ma ‘Thus, every vector a € V is in the subspace spanned by B’, whence B’ is a basis for V. Finally, if @ € V and suppose [ole = and [ale = dn, &n, so that Equation IL-7 holds, then by Equation IL8, we ce = Yo Quid a Hence, [als = Qla|s — Pals By symmetry, it follows that [als = Play This completes the proof a 39Example 4.8. Let F = R and V = R? and B = {e1,¢} the standard ordered basis. For a fixed @ € R, let cos(0) —sin(0) sin(9) cos(0) Then P is an invertible matrix and po ( cos(0) sn) —sin(0) cos(0) for V. It is, geometrically, the for «= (1,22) € V, we have Then B' = {(cos(0), sin(@)), (— sin(0), cos(0))} is a bi usual pair of axes rotated by an angle 9. In this basis la sry cos(9) + 2 sin() s —2 sin(9) + 22 cos(0) 5. Summary of Row Equivalence Definition 5.1. Let A be an m x n matrix over a field F, and write its rows as vectors {011,042 .5.,4m} CF (i) The row space of A is the subspace of F” spanned by this set. (i) The row rank of A is the dimension of the row space of A. Theorem 5.2. Row equivalent matrices have the same row space. Proof. If A and B are two row-equivalent m x n matrices, then there is an invertible matrix P such that B=PA If {a1,09,... cq} are the row vectors A and {9),82,..., 8yn} are the row vectors of B, then m DY Paes If W, and Wp are the row spaces of A and B respectively, then we see that {Bt Bays Bm} C Wa Since Wg is the smallest subspace containing this set, we conclude that We c Wa Since row equivalence is an equivalen ‘elation, we have WC We as well. a Theorem 5.3. Let R be a row-reduced echelon matrix, then the non-zero rows of R form a basis for the row space of R. 40Proof. Let pi, p2,--.,pr be the non-zero rows of R and write Ria, Riss Riga) By definition, the set {p1, p2,...,-} spans the row space Wp of R. Therefore, it suffices to check that this set is linearly independent. Since R is a row-reduced echelon matrix, there are positive integers ky < ky <... < ky such that, for all i 2 Hence, m = p:. Thus proceeding, we may conclude that m, = p; for all 1, whence S=R. 42ao Corollary 5.5. Everymxn matrix A is row-equivalent to one and only one row-reduced echelon matric. Proof. We know that A is row-equivalent to one row-reduced echelon matrix from The- orem 1.2.9. If A is row-equivalent to two row-reduced echelon matrices K and S, then by Theorem 5.3, both R and S have the same row space. By Theorem 5.4, R=S. O Corollary 5.6. Let A and B be two m xn matrices over a field F, Then A is row equivalent to B if and only if they have the same row space Proof. We know from Theorem 5.2 that if A and B are row-equivalent, then they have the same row space. Conversely, suppose A and B have the same row space. By ‘Theorem 1.2.9, A and B are both row-equivalent to row-reduced echelon matrices R and S respectively. By Theorem 5.2, R and S have the same row space. By Theorem 5.4, R = $. Hence, A and B are row-equivalent to each other. a 43WM. Linear Transformations 1. ear Transformations Definition 1.1. Let V and W be two vector spaces over a common field F. A function T : V + W is called a linear transformation if, for any two vectors a, 8 € V and any scalar ¢ € F, we have T(ca + 8) = Ta) +T(8) Example 1.2. (i) Let V be any vector space and I: V—+ V be the identity map. Then J is linear. (ii) Similarly, the zero map 0: V + V is a linear map (aii) Let V = F" and W = F™, and A € F™*" be an m x n matrix with entries in F. ‘Then T: V > W given by T(X) := AX is a linear transformation by Lemma II.2.5. (iv) Let V be the space of all polynomials over F. Define D : V > V be the ‘derivative’ map, defined by the rule: If L( cot eae ten +...+ ene" Then (D(x) = er + ear +... nega” (v) Let F = R and V be the space of all functions f : R - R that are continu. ous (Note that V is, indeed, a vector space with the point-wise operations as in Example IT.1.2). Define T : V+ V by rene) = [" seoae With V as in the previous example and by R, we may also define T: V+ W Remark 1.3. If 7’: V > W is a linear transformation 44(i) T(0) =0 because if a == T(0), then Qa =a+a=T(0) +10) =T(0 +0) =70) =a Hence, a = 0 by Lemma I1.1.3 (ii) If 8 is a linear combination of vectors {a7,02,...,}, then we may write B= Yeas oa for some scalars ¢1,¢2,..., Cy € F. Then it follows that 708) = Darl) Theorem 1.4. Let V be a finite dimensional vector space over a field F and let a, a9, be an ordered basis of V. Let W be another vector space over F and (1, 2,...,8n} be any set of n vectors in W. Then, there is a unique linear transformation T : V > W such that Tla)=8 VWi W by Since the above expression is uniquely associated to a, this map is well-defined. Now we check linearity: If and c € F a scalar, then we have cat f= Vleet dias ‘= So by definition T(ca + 8) = Lec + d,)B; cotNow consider > =) + z 48 = Dlea says, { =T(ca + 8) Ta) +T(8)=e Hence, T’is linear as required. (di) Uniqueness: If $: V + W is another linear transformation such that S(a;) =f Vi R® such that T(ax) = (3,2,1) and T(a2) = (6,5,4) We find T(e2): To do that, we write & = C404 + C20 = (cy + Beg, 2eq + deg) = (1,0) Hence, = -2,e) = So that T(c) = -2T (ax) + T(a2) = -2(3,2, 1) + (6,5,4) = (0,1,2) (i) IT; F* + F* is a linear transformation, then Tis uniquely determined by the vectors B=Tle),1 sei) =05 > cia € ker(T) Sen Sea Hence, there exist scalars dy, d2,-..,dy € F such that k YS cai = Yoda; Sm Sat Since the set B is lincarly independent, we conclude that «0-4; for alll W be two linear transformations, and ¢ € F a scalar (i) Define (T +U):V + W by (T+ U)(a) = T(a) + U(a) 48(ti) Define (eT): V > W by (Tle) = T(0) Then (T +U) and cT are both linear transformations. Proof. We prove that (T | U) is a linear transformation. The proof for (cT) is similar. Fix a, 8 € V and d € F a scalar, and consider (I + U)(da + 8) =T(de + 8) + U(de + 8) = dT (a) +7(8) + dU(a) + U(3) = d(T(a) + U(a)) + (T(8) + U(8)) = d(T + U)a) + (T+ U)(8) Hence, (1’ + U) is linear. a Definition 2.2. Let V and W be two vector spaces over a common field F. Let L(V,W) be the space of all linear transformations from V to W. Theorem 2.3. Under the operations defined in Lemma 2.1, L(V,W) is a vector space. Proof. By Lemma 2.1, the operations + LV,W) x L(V,W) > L(V, W) and Fx L(V,W) + L(V,W) are well-defined operations. We now need to verify all the axioms of Definition II1.1 For convenience, we simply verify a few of them, and leave the rest for you (i) Addition is commutative: If T,U € L(V,1W), we need to check that (7+ U) = (U+T). Hence, we need to check that, for any a € V, (U+T)(a) =(T + UY(a) But this follows from the fact that addition in W is commutative, and so (T +U)(a) = T(a) + U(a) = U(a) + Ta) = (U + T)(a) (ii) Observe that the zero linear transformation 0: V + W is the zero element in LIV,W). (ai) Lot d € F and T,U € L(V,W), then we verify that d(T +U)=dT + dU So fix a € V, then [d(T + U)] (a) = d(T + Ua) =d(T(a) + U(a)) = dT (a) + dU (a) = (dT + aU)(a) This is true for every ainV, so d(T +U) = dT +d. 49The other axioms are verified in a similar fashion. o Theorem 2.4. Let V and W be two finite dimensional vector spaces over F. Then L(V,W) is finite dimensional, and dim(L(V, W)) = dim(V) dim(WV) Proof. Let B= {ay,03,-..,0n} and BY = (81, 82, be bases of V and W respectively. Then, we wish to show that dim(L(V,W)) = mn For each 1

V be the linear transformation T(f)(2) = xf (2) Then, if f(x) = 2" and n > 1, then (DT —TD)(fa)(z) = DT(fn)(e) = TDUn)(2) = D(x fn(2)) — T(nz"™) = D(x") —ne® (n+ 12" = n2* By Theorem 1.4, DT-TD=1 In particular, DT # TD. Hence, composition of operators is not necessarily a commutative operation. (di) Lot B = {a1,02,....,a,} be an ordered basis of a vector space V. For 1 < p,q AB (End of Week 3) Definition 2.9. A linear transformation T : V + W is said to be invertible if there is a linear transformtion $ : W > V such that ST =Iy and TS = Iw Definition 2.10. A function f : $+ T between two sets is said to be (a) injective if f is one-to-one, In other words, if x,y € Sand f(x) = f(y), then x =y (ii) surjective if f is onto. In other words, for any 2 € T, there exists x € S such that f(z) = 2, (ai) biyective if f is both injective and surjective. Theorem 2.11. Let T: V+ W be a linear transformation. Then T is invertible if and only if T is bijective. Proof. (i) If Tis invertible, then there is a linear transformation S': V > W as above (i) Sis injective: If a, 8 € V are such that T(a) = 7(8), Then ST( But ST = Iy, soa