Solution to Problem Set 1 October 1, 2015 Problem 1. By explicit computation with 2 × 2 matrices, [x, y] = h, (Check!) Thus tively, 0 0 0 [h, x] = 2x, [h, y] = −2x. the matrices of ad(x), ad(h), ad(y) in this basis are, respec −2 0 0 1 , 0 0 2 0 0 0 0 0 , 0 0 −2 0 0 0 −1 0 0 . 0 2 0 and Problem 2. Since x has n distinct eigenvalues a1 , . . . , an , it is diagonalizable. After choosing a suitable basis in Cn , we may assume that x is diagonal, x = diag(a1 , . . . , an ). We will now work in this basis. The gl(n, C) has a natural basis {Eij : i, j = 1, . . . , n}, where Eij is the elementary matrix with 1 in the (i, j)-position, and all other entries equal to 0. Then x = a1 E11 + · · · + an Enn . [x, Eij ] = aii [Eii , Eij ] + ajj [Ejj , Eij ] = (ai − aj )Eij . Thus Eij ’s form a basis of eigenvectors for gl(n, C), and the eigenvalue associated to Eij is ai − aj . Note: The word “precisely” in the statement of this problem means that the eigenvalues are counted with multiplicity. It is not enough to show that each ai − aj is an eigenvalue for adx, but you also need to know that each occurs with the right multiplicity. Note that assertion of this problem remains valid even if a1 , . . . , an are not distinct, as long as x is semisimple (i.e., diagonalizable). Problem 3. Since tr(AB) = tr(BA) for any two n × n-matrices A and B, we have [gl(n, C), gl(n, C)] ⊂ sl(n, C). 1 To prove the reverse inclusion, it suffices to show that the natural spanning set of sl(n, C) consisting of matrices Eij and Eii − Ejj where i 6= j, and Eii − Ejj lies inside [gl(n, C), gl(n, C)]. Indeed, for i 6= j, Eij = [Eik , Ekj ], ∀k 6= i, j, and Eii − Ejj = [Eij , Eji ]. for any pair of distinct integers 1 ≤ i, j ≤ n. Problem 4. (1) Center of gl(n, C): For every matrix X, use X(i, j) to denote its (i, j)-entry. If X is in the center of gl(n, C), first show it is diagonal. If not, assume ∃i 6= j such that X(i, j) 6= 0. Then [X, Ejj ](i, j) = (XEjj )(i, j) − (Ejj X)(i, j) = X(i, j) − 0 6= 0 contradiction. Now assume X is diagonal, X = diag(a1 , . . . , an ). If ∃i 6= j such that ai 6= aj , then [X, Eij ](i, j) = (XEij )(i, j) − (Eij X)(i, j) = (ai − aj )Eij 6= 0 contradiction. (2) Center of sl(n, C): If X commutes with all elements in sl(n, C), then it commutes with all elements in gl(n, C) since gl(n, C) = sl(n, C)+CIn . Hence X ∈ CIn and X ∈ sl(n, C) ∩ CIn = {0}. Problem 5. We argue by induction on n. P j i base case: n = 0. Here 0!1 d0 (xy) = xy = i+j=0 di!x dj!y , and the formula is true. Assume the the following equivalent formula holds for n: n X n i n−i d (xy) = d xd y, i n i=0 2 we show the formula for n + 1. dn+1 (xy) = d(dn (xy)) n X n i n−i = d( d xd y) i i=0 n n X X n i+1 n−i n i n+1−i = d xd y + d xd y i i i=0 i=0 n+1 n X n X n i n+1−i k n+1−k = d xd y+ d xd y k−1 i i=0 k=1 n n X X n n+1 0 n n i n+1−i n 0 n+1 k n+1−k = d xd y + d xd y+ d xd y+ d xd y n k−1 i 0 i=1 k=1 n X n n = dn+1 xd0 y + + di xdn+1−i y + d0 xdn+1 y i−1 i i=1 n X n + 1 n+1 0 n + 1 i n+1−i n + 1 0 n+1 = d xd y + d xd y+ d xd y n+1 i 0 i=1 n+1 X n + 1 = di xdn+1−i y. i i=0 Problem 6. Let I be a non-zero ideal of sl(n, C). Our goal is to show that I = sl(n, C). Note that the case, where n = 2 is settled in Section 2.1 see the Example at the bottom of p. 6. The folowing lemma is a consequence of this example. Lemma: Let a and b be distinct integers, between 1 and n and L = SpanC (Eab , Eba , Eaa − Ebb ) ⊂ sl(n, C). (a) If I ∩ L 6= (0), then L ⊂ I. In particular, (b) If Eab ∈ I for some a 6= b then Eaa − Eab ∈ I. (c) Conversely, if Eaa − Eab ∈ I, for some a 6= b, then Eab ∈ I and Eba ∈ I. Proof of the Lemma: L is easily seen to be a Lie subalgebra of sl(n, C) isomorphic to sl(V ), where V is the 2-dimensional subspace of Cn spanned by the standard basis vectors ei and ej . (Check!). Thus L is simple, L ∩ I is a non-trivial ideal of L, and part (a) follows. Parts (b) and (c) are special cases of (a). 3 Now choose t1 , . . . , tn ∈ C such that t1 + · · · + tn = 0 but ti − tj are all distinct for i 6= j. This can be done because the finitely many hyperplanes xi − xj = xk − xj in Cn do not cover the hyperplane x1 + · · · + xn = 0. Let h be the subalgebra of diagonal matrices in sl(n, C) and set H = diag(t1 , . . . , tn ) ∈ h. As we showed in Problem 2 above, the linear trasformation ad(H) : gl(n, C) → gl(n, C) is diagonalizable, with eigenvalues 0 of multiplicity n and ti − tj of multiplicity 1 for each i 6= j. The eigenspaces corresponding to these eigenvalues are h and SpanC (Eij ), respectively. Since I is invariant under H, it has a basis of eigenvectors of ad(H). Denote this basis by h1 , . . . , hr , Ei1 j1 , . . . , Eis js , where h1 , . . . , hr ∈ h. Note that a priori r may be 0 (i.e., I ∩ h = (0)) or s may be 0 (i.e., I ⊂ h). Case 1: s 6= 0, i.e., Eij ∈ I for some i 6= j. Then Eik = [Eij , Ejk ] ∈ I for any k 6= i, j. That is, Eik ∈ I for every k 6= i. By Lemma (b), Eii − Ekk ∈ I for every k = 1, . . . , n. (Here i is fixed.) Since these elements span h, we conclude that h ⊂ I. Thus Eaa − Ebb ∈ I for every pair of distinct integers a, b between 1 and n. By Lemma (c), Eab ∈ I. We have shown that h ⊂ I and Eab ∈ I for every a 6= b, Since these elements span sl(n, C), we conclude that I = sl(n, C). Case 2: s = 0, i.e., I ⊂ h. Then r ≥ 1. Let h1 = diag(a1 , . . . , an ), where a1 + · · · + an = 0. Since h1 6= 0, ai 6= aj for some i 6= j. Now [h1 , Eij ] = (ai − aj )Eij ∈ I. Thus Eij ∈ I, contradicting I ⊂ h. 4