Math 317: Linear Algebra Homework 10 Solutions Due: November 20, 2015 The following problems are for additional practice and are not to be turned in: (All problems come from Linear Algebra: A Geometric Approach, 2nd Edition by ShifrinAdams.) Exercises: Section 4.3: 5a, 6, 7, 11, 13, 16, 17 Section 5.1 : 1, 3, 10, 11 Turn in the following problems. 1 1. Continuing the example from class, let V = span {v1 , v2 } where v1 = 0 , v2 = 1 1 1 1 . Recall that we found a basis for R3 by taking v3 = −2 ∈ V ⊥ . −1 −1 (a) Find the standard matrix of the linear transformation T : R3 → R3 , if T (x) = projV ⊥ (x). That is, calculate [projV ⊥ ]stand . Hint: Recall that for any x ∈ R3 , we have that x = projV (x) + projV ⊥ (x). Use the fact that Ix = x where I is the identity transformation and your knowledge of [projV ]stand to calculate [projV ⊥ ]stand . 5/6 2/6 1/6 From class, we recall that [projV ]stand = 2/6 2/6 −2/6. Using 1/6 −2/6 5/6 3 our knowledge that for x ∈ R we have that x = projV (x) + projV ⊥ (x). This implies that projV ⊥ (x) = x − projV (x). Using the fact the Ix = x where I is the identity transformation (so that [I]stand = 1 0 0 I), we have that [projV ⊥ ]stand = I − [projV ]stand = 0 1 0 − 0 0 1 5/6 2/6 1/6 1/6 −2/6 −1/6 2/6 2/6 −2/6 = −2/6 4/6 2/6 1/6 −2/6 5/6 −1/6 2/6 1/6 (b) Calculate the matrix of the linear transformation T : R3 → R3 , if T (x) = projV ⊥ (x) with respect to B, if B = {v1 , v2 , v3 }. That is, calculate [projV ⊥ ]B . While we could use the change of basis formula to calculate [projV ⊥ ]B , it is actually less computationally intensive to compute [projV ⊥ ]B directly from its definition. Recalling that projV ⊥ (x) = x if x ∈ V ⊥ and projV ⊥ (x) = 0 if x ∈ V , we have that projV ⊥ (v1 ) = projV ⊥ (v2 ) = 0 and projV ⊥ (v3 ) = v3 . Writing out the coordinates of each of these vectors, we obtain: 1 Math 317: Linear Algebra Homework 10 Solutions Due: November 20, 2015 0 = 0v1 + 0v2 + 0v3 0 = 0v1 + 0v2 + 0v3 v3 = 0v1 + 0v2 + 1v3 , (1) (2) (3) 0 0 0 and so [projV ⊥ ]B = 0 0 0. 0 0 1 2. Section 4.3, Problem 5b 2 1 We consider the basis B = {v1 , v2 } for R2 where v1 = and v2 = . 3 2 Given the following linear transformation S : R2 → R2 defined by S(v1 ) = 2v1 + v2 S(v2 ) = −v1 + 3v2 , we calculate [S]stand , the standard matrix for S. To do this, we first compute [S]B and then use the change of basis formula. Now, from above we have S(v1 ) and S(v2 ), and thus all that is left to do is to compute the coordinates with respect to B. Note that we already have S(v1 ) and S(v2 ) written as a linear combination of the elements of 2 −1 B, thus [S]B = . Now to find [S]stand we use the change of basis 1 3 formula which says that [S]stand = P [S]B P −1 , where the columns of P are the vectors in the basis B. Thus, we have that [S]stand 2 1 = 3 2 −1 2 −1 2 1 1 3 3 2 3. Section 4.3, Problem 9 Let T : R3 → R3 be the linear transformation which reflects a given vector x across V where V = {(x1 , x2 , x3 ) : −x1 + x2 + x3 = 0}.We find a basis for V by solving the linear equation associated with V . Letting x2 and x3 be the free variables, we obtain x1 = x2 +x 3 , x2 = x2 and x3 = x3 . Thus a basis 1 1 for V is given by v1 = 1 , v2 = 0. To find a nonzero vector v3 that is 0 1 orthogonal to both v1 and v2 , we find a basis for V ⊥ (which will consist of one single vector v3 since V + V ⊥ = R3 ). Noting that the basis given above 2 Math 317: Linear Algebra Homework 10 Solutions Due: November 20, 2015 −1 1 1 is a basis for N (A) if A = , we know that V ⊥ = R(A) and so −1 v3 = 1 . Note that B = {v1 , v2 , v3 } forms a basis for R3 . To create an 1 orthogonal basis for R3 , we apply the Gram-Schmidt procedure to B. To begin we let 1 w1 = v1 = 1 . 0 w2 is given by 1 1/2 1 1 (w1 · v2 ) w1 = 0 − 1 = −1/2 . w 2 = v2 − 2 kw1 k 2 0 1 1 Finally, w3 is given by −1 (w1 · v3 ) (w2 · v3 ) w3 = v3 − w1 − w2 = 1 . 2 kw1 k kw2 k2 1 Next, we find [T ]B , where B = {w1 , w2 , w3 }. Recalling that T (x) = projV (x) − projV ⊥ (x), we have that T (w1 ) = projV (w1 ) − projV ⊥ (w1 ) = w1 = 1w1 + 0w2 + 0w3 T (w2 ) = projV (w2 ) − projV ⊥ (w2 ) = w2 = 0w1 + 1w2 + 0w3 T (w3 ) = projV (w3 ) − projV ⊥ (w3 ) = −w3 = 0w1 + 0w2 − 1w3 1 0 0 Thus, [T ]B = 0 1 0 . 0 0 −1 To find [T ]stand , we use the change of basis formula: [T ]stand = P [T ]B P −1 , where the columns of P are the vectors in the basis B. Thus, we have that [T ]stand −1 1 1/2 −1 1 0 0 1 1/2 −1 = 1 −1/2 1 0 1 0 1 −1/2 1 . 0 1 1 0 0 −1 0 1 1 4. Section 4.3, Problem 18 3 Math 317: Linear Algebra Homework 10 Solutions Due: November 20, 2015 (a) Proof : Suppose that c is any scalar. We show that cI is similar only to itself. Suppose that cI is similar to a matrix A. Then there is a nonsingular matrix P such that A = P (cI)P −1 =⇒ A = cP P −1 = cI. So A = cI and hence cI is similar only to itself. b 0 a 0 (b) To show that A = is similar to B = , we find a matrix 0 b 0a p p 1 2 P so that A = P BP −1 or AP = P B. Let P = . Then AP = p3 p 4 ap1 ap2 bp1 ap2 P B =⇒ = . This implies that p1 = p4 = 0 bp3 bp4 bp3 ap4 and p2 and p3 is any scalar. One particular P that works in this case 0 1 is P = . 1 0 2 1 2 0 (c) To show that A = is not similar to B = , we show that 0 2 0 2 p p there is not a P = 1 2 such that AP = P B. By way of contradicp3 p4 tion, suppose that we could find a nonsingular P so that AP = P B. 2p1 + p3 2p2 + p4 2p1 2p2 Then AP = P B =⇒ = =⇒ p3 = 2p3 2p4 2p3 2p4 p4 = 0 =⇒ P has a row of zeros and is hence singular, a contradiction. So A is not similar to B. (d) To show that A is not similar to any diagonal matrix, we repeat the a 0 p1 p2 same argument as above letting B = . Let P = . Then 0 b p3 p4 2p1 + p3 2p2 + p4 if A were similar to B, then AP = P B =⇒ = 2p3 2p4 ap1 bp2 =⇒ p3 = p4 = 0 thus making P a singular matrix, which ap3 bp4 is not possible. Thus, A is not similar to any diagonal matrix. 5. Section 4.3, Problem 19 2 2 (a) Using the fact that T : R → R isdefined by T (e1 ) = 8e1 − 4e2 = 8 9 and T (e2 ) = 9e1 − 4e2 = , we have that −4 −4 8 9 [T ]stand = −4 −4 3 −1 and v2 = −e1 + e2 = , we find (b) Given v1 = 3e1 − 2e2 = −2 1 [T ]B where B = {v1 , v2 } is a basis for R2 . This can be done via the change of basis formula. Recalling that [T ]stand = P [T ]B P −1 where the 4 Math 317: Linear Algebra Homework 10 Solutions Due: November 20, 2015 columns of P contain the basis B, we have that [T ]B = P −1 −1 3 −1 8 9 3 −1 2 1 [T ]stand P = = . −2 1 −4 −4 −2 1 0 2 (c) No, because under different bases the matrices for T are similar to each other. If one such matrix is not diagonalizable, then neither is the rest. (See Problem 18, part d) 6. Section 4.3, Problem 20 (a) True. Proof : Suppose that B is similar to A. Then there is a nonsingular matrix P such that B = P AP −1 . This implies that T B T = (P AP −1 ) = (P −1 )T AT P T = (P T )−1 AT P T . Thus B T is similar to AT with P T being the change of basis matrix. 0 1 0 0 (b) False. Let B = and let A = . Then B 2 = A = A2 and 0 0 0 0 so B 2 is similar to A2 since they are both the zero matrix. However, B is not similar to A because rank(A) = 0 and rank(B) = 1 and rank is a similarity invariant. (c) True. Proof: Suppose that B is similar to A and that A is nonsingular. We show that B is nonsingular. Now, since B is similar to A, we know that there is a nonsingular matrix P such that B = P AP −1 . Note that P AP −1 is the product of nonsingular matrices and so P AP −1 is a nonsingular matrix. Of course, this is precisely B and hence B is nonsingular. 2 1 2 0 (d) False. In class we showed that A = was similar to B = . 0 3 0 3 Notice that A is not symmetric whereas B is a symmetric matrix. 1 1 0 0 (e) False. Let A = and B = . A is similar to B through the 1 1 1 2 1 1 invertible matrix P = . (Verify this!) However, A and B have 0 1 different nullspaces. (f) True. Proof : Suppose that A is similar to B. Then there is a nonsingular P such that A = P BP −1 . We recall from a previous homework that rank(A) = rank(AB) if B is nonsingular, and rank(B) = rank(AB) if A is nonsingular. Equipped with these facts, we find that rank(A) = rank(P (BP −1 )) = rank(BP −1 ) = rank(B) since P and P −1 are both nonsingular. 7. Prove that the rank and trace of a matrix are similarity invariants. That is rank(An×n ) = rank(Bn×n ) and trace(An×n ) = trace(Bn×n ) when A is similar to 5 Math 317: Linear Algebra Homework 10 Solutions Due: November 20, 2015 B. Proof: In the previous problem, we have shown that rank is a similarity invariant. To prove that the trace is a similarity invariant, we recall that for any two n × n matrices, trace(AB) = trace(BA). Suppose that A is similar to B. Then there is a nonsingular matrix P such that A = P BP −1 . Thus trace(A) = trace((P B)P −1 ) = trace(P −1 P B) = trace(B). 8. Section 5.1, Problem 1a −1 3 5 We calculate det 6 4 2 as follows: −2 5 1 −1 3 5 −1 3 5 −1 3 5 det 6 4 2 = det 0 22 32 = − det 0 −1 −9 −2 5 1 0 −1 −9 0 22 32 −1 3 5 = − det 0 −1 −9 = 166 0 0 −166 9. Section 5.1, Problem 2 Proof : Suppose that one row of an n × n matrix A consists only of 0 entries. That is, A1 A2 .. . A = 0 , . .. An−1 An where Ai is the ith row of A. Write the 0 row as A1 − A1 . Then A1 A2 .. . A1 A2 .. . A1 A2 .. . A1 A2 .. . det A = det 0 = A1 − A1 = det A1 − det A1 = 0 . . . .. .. .. .. . An−1 An−1 An−1 An−1 An An An An 6