Solutions to Afternoon Exam 5 Page

Solutions to Afternoon Exam Part I. Solve four of the following five problems. Problem 1. Find the inverse of the matrix   1 0 −1 A = 0 2 0  . 5 0 3 Solution: Any sequence of row operations which puts the matrix (A|I) into rowreduced upper-echelon form terminates in the matrix (I|B), where   3/8 0 1/8 1/2 0 . B = A−1 =  0 −5/8 0 1/8 Alternatively, one can use the formula A−1 = (bij ), where bij = (−1)i+j (det A)−1 (det Aji ), where Aji is the matrix obtained from j by removing the jth row and ith column. Problem 2. The 2 × 2 matrix A has trace 1 and determinant −2. Find the trace of A100 , indicating your reasoning. Solution: The characteristic polynomial of A is x2 − x − 2 = (x + 1)(x − 2). Thus the eigenvalues of A are −1 and 2, and consequently the eigenvalues of A100 are 1 and 2100 . So tr (A) = 1 + 2100 . Problem 3. Find a basis for the space of solutions to the simultaneous equations ( x1 + 2x3 + 3x4 + 5x5 = 0 x2 + 5x3 + 4x5 = 0. Solution: The associated matrix is already in row-reduced upper-echelon form, so one can solve for the pivotal variables x1 and x2 in terms of the nonpivotal variables x3 , x4 , and x5 . Thus the general solution to the system is (−2x3 − 3x4 − 5x5 , −5x3 − 4x5 , x3 , x4 , x5 ) = x3 v1 + x4 v2 + x5 v3 , where v1 = (−2, −5, 1, 0, 0), v2 = (−3, 0, 0, 1, 0), v3 = (−5, −4, 0, 0, 1), and x3 , x4 , and x5 are arbitrary scalars. Hence a basis for the solution space is {v1 , v2 , v3 }. Problem 4. Let A be a 3 × 3 matrix with coefficients in R. Show that if A4 = 0 then A3 = 0, and give an example where A3 = 0 but A2 = 6 0. Solution: If λ ∈ C is an eigenvalue of A then for some nonzero column vector X ∈ C3 we have AX = λX. By iteration, A4 X = λ4 X and consequently λ4 X = 0 and hence λ = 0. Thus 0 is the only eigenvalue of A. It follows that the characteristic polynomial of A is X 3 , whence A3 = 0 by the Cayley-Hamilton theorem. If   0 1 0 A = 0 0 1 0 0 0 1 2 then A3 = 0 but  0 A2 = 0 0 0 0 0  1 0 . 0 Problem 5. Let W be the subspace of R4 spanned by the vectors √ √ √ w1 = (1/ 3, −1/ 3, 0, 1/ 3) and √ √ √ w2 = (1/ 3, 1/ 3, 1/ 3, 0). √ √ √ √ Let v = ( 3, 3, 3, 3). Write v = w + w⊥ , where w ∈ W and w⊥ is orthogonal to W relative to the dot product. Solution: Since {w1 , w2 } is already an orthonormal basis for W , we have w = (v · w1 )w1 + (v · w2 )w2 = 1w1 + 3w2 ⊥ and w = v − w. So and √ √ √ √ w = (4/ 3, 2/ 3, 3, 1/ 3) √ √ √ w⊥ = (−1/ 3, 1/ 3, 0, 2/ 3). Part II. Solve three of the following six problems. Problem 6. In a certain group there are elements g and h satisfying ghg −1 = h . Show that (gh)2 = g 2 . Solution: Multiplying both sides of the equation ghg −1 = h−1 by g −1 on the left and g on the right, we obtain h = gh−1 g −1 . Returning to ghg −1 = h−1 and conjugating by g a second time, we deduce that g 2 hg −2 = h. Thus −1 (gh)2 = (ghg −1 )(g 2 h) = h−1 (g 2 hg −2 )g 2 = g 2 . Problem 7. The linear transformation T : R3 → R3 with matrix (relative to the standard basis) √   5/8 3/4 √3/8 A = √ −3/4 1/2 3/4 √ 3/8 − 3/4 7/8 is a rotation about an axis through the origin. Find the axis. Solution: The axis is the line spanned by an eigenvector of eigenvalue 1. So we need to find a basis for the null space of √   −3/8 3/4 √3/8 A − I = √ −3/4 −1/2 3/4 √ 3/8 − 3/4 −1/8 √ Subtract 2 times the first row from the second and add 1/ 3 times the first row to the third. Then divide the first row by −3/8 and subtract the second row from the first row. Finally, divide the second row by −2. The matrix is now in row-reduced 3 √ upper-echelon form, and we see that a basis for the kernel in (1, 0, 3). So the axis is the line spanned by this vector. Problem 8. In this problem you may quote the Rank-Nullity Theorem and general facts from set theory without proof. (a) Let V be a finite-dimensional vector space, let idV be the identity map on V (so idV (v) = v for all v ∈ V ), and let f, g : V → V be linear maps satisfying f ◦ g = idV . Prove that g ◦ f = idV . (b) Is (a) still true if we remove the assumption that V has finite dimension? Either prove or give a counterexample. Solution: (a) It is given that g is a right inverse of f , and we must show that g is actually a genuine two-sided inverse of f . In other words, we are given that g is injective, and we must show that g is also surjective. But by the Rank-Nullity Theorem, dim(V ) is the sum of the rank of g and the nullity of g. Since the nullity is 0 (g is injective), the rank is dim(V ); in other words, the image of g is all of V . So g is surjective. (b) The statement no longer holds if the dimension of V is infinite. For example, let V be the space of infinite sequences (a1 , a2 , a3 , . . . ) of real numbers, and let f (a1 , a2 , a3 , . . . ) = (a2 , a3 , . . . ). Let g(a1 , a2 , . . . ) = (0, a1 , a2 , . . . ). Then f ◦ g = idV , but g ◦ f 6= idV : Indeed if v = (1, 0, 0, . . . ) then g(f (v)) = 0, not v. Problem 9. Let V be the real vector space consisting of polynomials with real coefficients and degree at most 2, and consider the linear transformation T : V → V given by T (f (x)) = f (x) + f 0 (x) + f 00 (x). What is the Jordan normal form of T ? Your answer should be a matrix together with justification. Solution: As ordered basis for V we take 1, x, x2 . The matrix of T relative to this basis is   1 1 2 A = 0 1 2 . 0 0 1 A computation shows that (A − I)3 = normal form of T is  1 J = 0 0 0 but (A − I)2 6= 0, whence the Jordan 1 1 0  0 1 . 1 Problem 10. Let L be the lattice in Z3 spanned by the vectors (1, 1, 60), (2, 0, 60), and (1, 1, 0). Write Z3 /L as a direct sum of cyclic factors, Z3 /L ∼ = (Z/a1 Z) ⊕ (Z/a2 Z) ⊕ · · · ⊕ (Z/ak Z), where aj > 2 for 1 6 j 6 k and k, the number of cyclic direct summands, is (i) minimal, (ii) maximal. Solution: Using row and column operations over Z, we can put the relevant matrix in elementary divisor form:       0 1 0 1 2 1 0 −1 1 1 0 0 1 −1 0  1 0 1 0 1 0  = 0 2 0  . 0 0 1 60 60 0 1 1 −1 0 0 60 4 Thus Z3 /L ∼ = (Z/2Z) ⊕ (Z/60Z) and hence by the Chinese Remainder Theorem also Z3 /L ∼ = (Z/2Z) ⊕ (Z/4Z) ⊕ (Z/3Z) ⊕ (Z/5Z). The first decomposition, where k = 2, is the decomposition with minimal k, and the second expression, where k = 4, is the decomposition with maximal k. Indeed k = 1 is not possible because Z3 /L is not cyclic (indeed the subgroup (Z/2Z) ⊕ (Z/4Z) is not cyclic) and k = 5 is also not possible because (Z/2Z)⊕(Z/4Z) is not isomorphic to (Z/2Z) ⊕ (Z/2Z) ⊕ (Z/2Z). Problem 11. Let Sn denote the group of permutations of {1, 2, 3, . . . , n}. In each case, give an example of the indicated type or explain why none exists: (a) an element of order 40 in S13 . (b) an element of order 34 in S16 Solution: The order of an element σ of Sn is the least common multiple of the lengths of the cycles in a disjoint cycle decomposition of σ. In case (a) we can take σ = (1, 2, 3, 4, 5)(6, 7, 8, 9, 10, 11, 12, 13). However in case (b) we would need either a cycle of length 34 (impossible in S16 ) or else the product of disjoint cycles of length 2 and length 17. This is again impossible because a cycle of length 17 does not exist in S16 . Part III. Solve one of the following three problems. Problem 12. Give an example, with supporting justification, of two commutative rings which are isomorphic as abelian groups under addition but not isomorphic as rings. Solution: There are many possiblities. For example, let p be a prime, let Fp be the field with p elements, and take R = Fp × Fp , and let R0 = Fp [x]/(x2 ). Then both R and R0 are two-dimensional vector spaces over Fp (the cosets of 1 and x are a basis for R0 ). But R0 has nilpotent elements, namely the multiples of the coset of x, whereas R does not have nilpotents. So R and R0 are not isomorphic as rings. Of course there are many other solutions. For example, we could let R0 = Fp2 and leave R unchanged. Then R and R0 are not isomorphic as rings because R0 is a field and R is not. Or take R = Q × Q and R0 = Q(i): Again, R0 is a field and R is not. Problem 13. Let Fp be the field with p elements. What are the degrees and multiplicities of the monic irreducible factors of the polynomial f (x) = x3 + x2 + 3x + 2 viewed over (i) F2 , (ii) F3 , (iii) R, and (iv) Q? Give reasons for your answers. Solution: (i) Over F2 we have f (x) = x3 + x2 + x = x(x2 + x + 1). Since x2 + x + 1 is irreducible over F2 (for f (0) = f (1) = 1 6= 0) there are two distinct irreducible factors, one of degree 1 and one of degree 2. 5 (ii) Over F3 we have f (x) = x3 + x2 + 2. Since f (0) = 2, f (1) = 1, and f (2) = 2, we see that f does not have any degree-one factors and is therefore irreducible of degree 3. (iii) Since f is a cubic, it has at least one real zero. But for all x ∈ R, f 0 (x) = 3x2 + 2x + 3 > x2 + 2x + 1 = (x + 1)2 > 0, so f is strictly increasing on R and therefore has exactly one real zero. So f has one irreducible factor of degree 1 and one of degree 2. (iv) Since f is irreducible over F3 it is irreducible over Q. Problem 14. Consider the matrix   1 1 0 A = 0 3 1 . 0 0 7 State the Jordan normal form, with justification, for A as a matrix with coefficients in (i) Q, (ii) F2 , (iii) F3 , and (iv) F5 . Solution: (i) Over Q, the matrix A has 3 distinct eigenvalues, namely 1, 3, and 7. Since the eigenvalues are distinct the matrix is diagonalizable, so the Jordan normal form is   1 0 0 J = 0 3 0 . 0 0 7 (ii) Over F2 , the matrix A is in Jordan  1 A = 0 0 normal form as it stands:  1 0 1 1 . 0 1 (iii) Over F3 , there are two eigenvalues, 0 with multiplicity one and 1 with multiplicity two. But A is not diagonalizable, for A(A−I) 6= 0 by direct calculation. Hence the Jordan normal form of A is   1 1 0 J = 0 1 0 . 0 0 0 (iv) Finally, over F5 there are again 3 distinct eigenvalues, namely 1,2, and 3. So A is diagonalizable and the Jordan normal form is the matrix J in (a), although now 7 = 2.