MATH 534. SOLUTIONS TO PROBLEM SET 4 Problem 1. (5pts) (a) Define the following linear map: Φ:V →V∗ v 7→ B(v, −) It is injective since Ker(Φ) = V ⊥ = {0} (non-degenerate). It is also surjective, since V and V ∗ has the same dimension. Now let r : V ∗ → W ∗ be the map obtained by restricting a linear function l : V → C to W . Note that l is surjective (check!). The composition l ◦ Φ: V ∗ W∗ gives a surjective map V → W ∗ . Explicitly, v 7→ B(v, −)|W . The kernel of this map is precisely W ⊥ . Thus dim V = dim W ∗ + dim W ⊥ = dim W + dim W ⊥ . (b) By definition, W ⊂ (W ⊥ )⊥ . Note that by part (a), dim(W ⊥ )⊥ = dim(V ) − dim(W ⊥ ) = dim(W ). Thus W ⊂ (W ⊥ )⊥ . Problem 2. (5pts) (“Only if”) Suppose y lies in the image of ad(x). That is, y = [x, v] for some v ∈ L. Then ∀z ∈ Ker(ad(x)), B(y, z) = −B([v, x], z) = −B(v, [x, z]) = −B(v, 0) = 0, as desired. (“If”) Conversely, assume B(y, z) = 0 ∀z ∈ Ker(ad(x)). Then y ∈ Ker(ad(x))⊥ . The “only if” part shows that Im(ad(x)) ⊂ (Ker(ad(x)))⊥ . Moreover, dim Im(ad(x)) = dim L − dim Ker(ad(x)) = dim(Ker(ad(x)))⊥ . Hence Im(ad(x)) = Ker(ad(x))⊥ and y ∈ Im(ad(x)). Problem 3. (5pts) Denote the image of v ⊗ w ∈ V ⊗ W in Hom(W ∗ , V ) under the isomorphism φ : V ⊗ W and Hom(W ∗ , V ) described in the statement of this problem by φv⊗w : V ⊗ W ' Hom(W ∗ , V ) v ⊗ w 7→ φv⊗w . We need to check that a · φv⊗w = φa·v⊗w + φv⊗a·w , in Hom(W ∗ , V ). 1 2 MATH 534. SOLUTIONS TO PROBLEM SET 4 Indeed, ∀l ∈ W ∗ , we have [a · φv⊗w ](l) = a · (φv⊗w (l)) − φv⊗w (a · l) = a · [l(w)v] − (a · l)(w)v = a · [l(w)v] − [−l(a · w)v] = l(w)a · v + l(a · w)v = φa·v⊗w (l) + φv⊗a·w (l). Problem 4 (Exercises 5a,d on page 30-31). (15pts) (a) (6pts) L is a representation of the Lie algebra ad(L) ' L/Z(L) = L/Rad(L) which is semisimple. Z(L) is an ad(L)-submodule in L. By Weyl’s Theorem, it has a complement submodule L0 (ideal), i.e. L = Z(L) ⊕ L0 . Then L0 ' L/Z(L) is semisimple. Hence L0 = [L0 , L0 ] ⊂ [L, L]. On the other hand [L, L] = [Z(L) ⊕ L0 , Z(L) ⊕ L0 ] ⊂ [L0 , L0 ] ⊂ L0 . Thus L0 = [L, L]. (d) (9pts) By (a) L = Z(L) ⊕ [L, L]. Let (ρ, V ) be an L-module such that Z(L) acts by semisimple elements of gl(V ). Then elements in ρ(Z(L)) can be diagonalized simultaneously, and V decomposes as a direct sum of weight spaces, as a Z(L)-module: M V = Vλ λ where λ ranges over Z(L)∗ and Vλ = {v ∈ V : ρ(z)v = λ(z)v, ∀z ∈ Z(L)}. Now each Vλ is an L-submodule. Indeed, ∀v ∈ Vλ , ∀x ∈ [L, L] we have ρ(x)v ∈ Vλ ∈ Vλ since ρ(z)ρ(x)v = ρ(x)ρ(z)v = λ(z)ρ(x)v, ∀z ∈ Z(L). Since [L, L] is a semisimple Lie algebra, each Vλ decomposes as a direct sum of irreducible [L, L] submodules: Vλ = mλ M Vλi . i=1 Here each Vλi is irreducible as an [L, L]-modules. Since Z(L) acts on each of these irreducible submodules by scalar multiplication, each Vλi is also an L-module. Moreover, each Vλi is irreducible as an L-module, since it is irreducible as an [L, L]-module. This shows that V is completely reducible, V = mλ MM λ i=1 Vλi . MATH 534. SOLUTIONS TO PROBLEM SET 4 3 Problem 5 (Exercise 6 on page 31). (5pts) β, γ give us two linear maps L → L∗ : L → L∗ x 7→ β(x, −) x 7→ γ(x, −) Both maps are homomorphisms of L-modules, since β, γ are associative. Both are isomorphisms of vector spaces, since β, γ are non-degenerate. As L-modules, L ' L∗ . Hence the above two maps are in HomL (L, L∗ ) = HomL (L, L). Since L is a simple Lie algebra (i.e., an irreducible L-module) Schur’s lemma tells us that HomL (L, L) = C. Thus β and γ are scalar multiples of each other. Problem 6 (Exercise 7 on page 31). (5pts) It is easy to check that the bilinear form (x, y) 7→ Tr(xy) is non-zero, symmetric and associative. We shall refer to it as ”the trace form”. As we saw in class, the radical of an associative symmetric bilinear form on a Lie algebra L is an ideal of L. Since sln is simple, we conclude that the radical of trace form is zero. In other words, the trace form is non-degenerate. By the previous problem, the Killing form is a scalar multiple of the trace form. In other words, κ(x, y) = cTr(xy), ∀x, y ∈ sln for some c ∈ C. To compute the scalar c, set x = y = E11 −E22 . xy = diag(1, −1, 0, . . . , 0)2 = diag(1, 1, 0, . . . , 0) and Tr(xy) = 2. On the other hand, let us examine ad(x) in the natural basis of sln consisting of the n2 − n elementary matrices Eij , i 6= j, and the n − 1 diagonal matrices Eii − Ei+1 i+1 , i = 1, . . . , n − 1. We have ad(x)(Eii − Ei+1,i+1 ) = 0 ∀i = 1, 2, . . . , n − 1 ad(x)(E12 ) = 2E12 ad(x)(E1k ) = E1k k = 3, . . . , , n ad(x)(E2k ) = −E2k k = 3, . . . , n ad(x)(E21 ) = −2E21 ad(x)(Ek1 ) = −Ek1 ad(x)(Ek2 ) = Ek2 ad(x)(Eij ) = 0 k = 3, . . . , , n k = 3, . . . , n for other i, j Hence ad(x) is diagonal in our basis of sln , with diagonal entries −2, 2, −1 and 1, appearing with multiplicities 1, 1, 2(n − 2) and 2(n − 2), respectively. Thus κ(x, y) = Tr(ad(x)ad(y)) = Tr(ad(x)2 ) = 22 +(−2)2 +12 ·2(n−2)+(−1)2 ·2(n−2) = 4n and c = 4n κ(x, y) = = 2n. Tr(xy) 2