Notes on Homework ? Some calculations omitted; ask if you don’t know how to do a part. 1. Let A = 4 −1 . 6 −1 (a) Its characteristic polynomial is p A ( z) = det( A − zI2 ) 4 −1 z 0 = det( − ) 6 −1 0 z 4−z −1 = det( ) 6 −1 − z = z2 − 3z + 2. (b) The eigenvalues of A are the roots of p A ( z), namely, 1 and 2. (c) We calculate the associated eigenspaces: • We have v ∈ ε A (1) if and only if Av = v; so, let v = x y , and solve the matrix equation 4 −1 6 −1 x x = , y y which is equivalent to the system of equations ( 4 − 1) x − y = 0 6 + (−1 − 1) y = 0 This system has one free variable, and thus a one-parameter family of solutions, 1 spanned by, for example, . 3 1 ). Explicitly, ε A (1) = span ( 3 • Similarly, ε A (2) consists of those v such that Av = 2v; solve this, and find that 1 ε A (2) = span( ). 2 (d) Since p A ( z) factors as z2 − 3z + 2 = ( z − 1)1 ( z − 2)1 , we have α A (1) = α A (2) = 1. (For any other number λ, α A (λ) = 0.) Professor Dan Bates Colorado State University M369 Linear Algebra Fall 2008 (e) We have γ A (1) = dim ε A (1) = 1, and similarly γ A (2) = dim ε A (2) = 1; and for any other λ, γ A (λ) = 0. (f) Since for all λ, γ A (λ) = α A (λ), the matrix is diagonalizable. (g) Since 0 is not an eigenvalue of A, there is no nontrivial solution to Av = 0 · v, and A is injective; and an injective map from a vector space to itself is invertible. 2. With even less annotation: (a) pC ( z) = det(C − zI2 ) 1−z 1 = det( ) −1 −1 − z = z2 (b) The eigenvalues are the roots of pC ( z) = z2 ; 0 is the only eigenvalue. (c) We need to solve Cv = 0v; the solution space is (calculation omitted) εC (0) = span ( (d) (e) (f) (g) 1 −1 αC (0) = 2, since pC ( z) = ( z − 0)2 . γC (0) = 1, since dim εC (0) = 1. The matrix C is not diagonalizable, since there’s an eigenvalue, 0, with γC (0) < al phaC (0). Since 0 is an eigenvalue, the nullspace of C has positive dimension (in fact, dimension one), and thus C is not invertible. 3. Let A = [T ]E ←E . (a) As above, compute the characteristic polynomial p A ( z) = det( A − z id ) = z2 − z − 6 = ( z − 3)( z + 2). So the eigenvalues of A are 3 and −2. 1 (b) For λ ∈ {−2, 3}, solve the equation Av = λv. One finds that v1 = is an eigen1 2 is an eigenvalue for −2. value for 3, and v2 = 1 (c) Let B be the basis Then T (v1 ) = 3v1 = 3v1 + 0v2 , so {v1, v2 } with v1 and v2 as above. 0 3 . Similarly, [T (v2 )]B = . Therefore, that [T (v1 )]B = 0 −2 [T ]B←B = [T (v1 )]B [T (v2 )]B 3 0 = . 0 −2 Professor Dan Bates Colorado State University M369 Linear Algebra Fall 2008 ). (d) Just compute: −1 −7 10 1 −5 8 1 −1 2 −7 10 1 = 1 1 −1 −5 8 3 0 = 0 −2 [id]B←E [T ]E id E ←B = 1 2 1 1 2 1 2 1 = [T ]B←B . 4. Since λ is an eigenvalue of A, there is a nonzero element v ∈ Fn with Av = λv. (a) Compute: (α A)v = α · Av = α · λv = (αλ)v. Therefore, v is an eigenvector of α A with eigenvalue αλ. (b) If k = 0, then we define A0 = id, and the statement follows immediately. Otherwise, one could prove this by induction on k, but: Let v be as above. Then Ak v = Ak−1 Av = A k− 1 λ v = λ A k− 1 v = λ Ak−2 Av = λ 2 A k− 2 v · · · = λk v. To prove this by induction on k: The claim is clearly true for k = 1. Otherwise, suppose it is known that Ak−1 v = λk−1 v. Then Ak v = AAk−1 v = Aλk−1 v = λk−1 Av = λk−1 λv = λk v. (c) Let p( z) = ∑kN=0 α k zk be a polynomial with coefficients in F. Let’s figure out the effect of p( A) (a sum of multiples of powers of A on v. Using the first two parts, we have N p( A ) v = ( ∑ α k A k ) v k= 0 N = ∑ αk Ak v (distribute) k= 0 N = ∑ αk λk v k= 0 = p(λ)v. Professor Dan Bates Colorado State University M369 Linear Algebra Fall 2008 5. Let p A ( z) be the characteristic polynomial of A. We evaluate p A (0) in two different ways. On one hand, since p A ( z) = det( A − zI ) if we substitute z = 0 we have p A (0) = det( A − 0I ) = det A. On the other hand, if p A ( z) = an zn + · · · + a1 z + a0 , then substituting z = 0 yields p A (0) = a0 , the constant term of p A ( z). Professor Dan Bates Colorado State University M369 Linear Algebra Fall 2008