Linear transformations Definition, Let V and W be vector spaces. Alinear transformation L_ of V into W isa function assigning a unique vector Lu) in W toeach w in V such that: a) L(w+v) = Lu) +L(v), forevery u,veV, b) L(ku) = kL(u), forevery uw € V andevery scalar k. If V=W, the linear transformation L_ is also called a linear operator on V. Example 1. Let L:R? —> R? be defined by L(x) = (x,,%2 — x4). x = (X23). Y= (Wn Yara) a). Le +) = LO + Yar X2 + Yar %s + Ys) = (1 + Yar 2 + Ye) — (a + ¥3)) = (21 + Yn, 2 — 3) + O2 — ¥5)) = (tn Ce — 49) + (v1.02 - ys) = L@) +L). b). L(kx) = L(kxy, kx2, kx) = (kx, kxg — kx) =k (24,X2 — X3) =k L(x). Thus L is linear transformation. Example 2. Let L:R? —+ R? be defined by L(x) = (4; +1,2). Is L a linear transformation? a). LQ +y) = Ga +1 + 1,2 + ¥2) and LOX) + LO) = Ga +122) + On + Liye) = Gi + 1 +2,%2 + ye): Since L(x + y) # L(x) + L(y), we conclude that L is not a linear transformation. Definition. The linear transformation 1), L:R? — R? defined by L(@x,y,z) =(x,y) is called projection, 2). L:R? — R* defined by L(w) =ru,r> 1 is called dilation,3). L:R? — R# defined by L(u) = ru, 0< r <1 iscalled contraction, 4). L:R? — R* defined by L(x, y) =(x,—y) is called reflection, cosp —sing sinp cosp 5). L:R? — R? defined by L(w wis called rotation. Example 3. Let L:P, — P, defined by L(at+b) = t(at+b). Show that L is linear transformation. Solution. Let at+b and ct+d be vectors in P, and let k is a scalar. Then L[{(at+b) + (ct+d )] = t{(at+b) + (ct+d )] =t(at+b) + t(ct+d) =L(at+b)+L(ct+d) and L[k(at+b)] = t[k(at+b)] = k[t(at+b)] = kL(at +b). Hence L is linear transformation. Example 3. Let L: Mypy, > Myyn defined by L(A) = AT Is La linear transformation? Solution. Let A and B be vectors in Myyp Then L(A +B) = (A+B)! = Al + Bland ifk isa scalar L[kA] =(kA)" = kA? = kL(A). Hence L is linear transformation Theorem 1. If L:V — W isalinear transformation, then (CV, + CyVq + + CyVpy = CL (Vy) + Cb (Vg) +o + CyLy)For any vectors v,,V2,..,q in Vand any scalars ¢1,€2 Cn Theorem 2. Let L:V — W is linear transformation, Then a). L(y) = Ow b). L(@u—v) = LQ) —L(v), Definition, A linear transformation L:V — W is said to be one-to-one if for all vy, in V, vy # vz implies L(y) # L(v2). An equivalent statement is that Lis one-to-one if forall L(v,) = L(v2) implies v, = v2. Example 3. Let L:R? — R? be defined by L(x,y) =(x+y,x-y). To determine whether L is one-to-one, we let Y, = (@,4;) and v2 = (by,by)- Then L(y) = L@®) we have a, +a, =b, + by ay — a) = by — by Adding these equations, we obtain 2a, = 2b,, or ay = by, similarly ay = bo. Hence, v; = vz and L is one-to-one. Example 4. Let L:R? — R? is linear transformation defined by LGe,y,z) = Gy). Since (1,3,3) #(1,3,—2) but L(1,3,3) = L(1,3,—2) =(1,3), we conclude that Lis not one-to-one.Definition. Let L:V — W isalineartransformation. The kernel of L, denoted by kerl, isthe subset of V consisting of all vectors v such that L(v) = Oy. Example 5. Let L:R? — R? is linear transformation defined by L(x,¥,2) = (%Y)- The vector (0,0,2) is in kerL, since L(0,0,2) = (0,0). To find kerL, we must determine all u in R? so that L(w) = 0. That is, we seek u = (x,y,z) so that L(w) = L(x,y,z) = 0 = (0,0). However, L(x,y,z) = (x,y). Thus (x,y) = (0,0), sox =0,y=0 andz can be any real number. Hence, kerZ consists of all vectors in R* of the form (0,0,r), where r is any real number. Theorem 3. If L:V — W isa linear transformation. Then kerl isa subspace of V. Example 6. Let L:R? — R? is linear transformation defined by Lx,y,2z) = (sy). Then (0,0,1) isa basis for kerL and dim( kerL)=1 Theorem 4. A linear transformation L:V — W is one-to-one ifand only if kerL= (0y}. Definition. Let L:V — W bea linear transformation, then the range of L, denoted by rangel, is the set of vectors in W that are images, under L, of vectors V. Thus a vector w is in rangel. ifand only if there exists some vector v inV such that L(») = w. IfrangeL=W, we say that L isonto, Thatis, L is onto ifand only if, given any w in W, there is a vector v in V such that L(v) =w.Theorem S. If L:V — W isalinear transformation then rangel isa subspace of W. Theorem 6. If L:V — W isa linear transformation of an n —dimensional vector space V into vector space W, then dim(kerL) +dim(rangeL) =dimv. Example 7. Let L: 8? — Ris linear transformation defined by @) ft 0 ape Llag}=]1 1 2] }a2 as} 12 1 al las Is L onto? Solution. a) Given any - in R*, where a, b,c are any real numbers, can we find ay v= |az as, so that L(v) = w? We seck solution to the linear system 10 1ypay pa F 1 | [| -[| 2 1 alla} te Use the elimination method we have, that a; +a, =b-a | a +a,=a 0 =c-b-a Thus the solution exists only when 0 =c—b—a.Hence Lis not onto.That is, exist values a,b,c for which is no vector v in R? such that a Lv) = [. c b). To find a basis for range , we note that a) fl 0 tym a, +03 1 0 1 L}42}=]1 1 2/]@2}=] a1 + az + 2a3\=a,} 1] + az] 1] + a, 2]. a3} l2 1 3}las) l2a, +a 43a 2. 1 3. This means that vectors 1] poy pt } *} [| spans rangeL. Thatis, rangel is the subspace of R°. al ul Is The first two vectors are linearly independent, the third vector is the sum of first two. Therefore, the first two vectors form a basis for rangeL, and dim( rangeL)=2. Therefore, we have dim(kerL) =3-—2=1.