THE HODGE LAPLACIAN 1. the Hodge Star Operator Let (M,G)

THE HODGE LAPLACIAN 1. the Hodge Star Operator Let (M,G)

LECTURE 25: THE HODGE LAPLACIAN 1. The Hodge star operator Let (M; g) be an oriented Riemannian manifold of dimension m. Then in lecture 3 we have seen that for any orientation-preserving chart, the Riemannian volume form (which is independent of the choice of coordinates) is given by p 1 m !g = Gdx ^ · · · ^ dx : Now let p 2 M. Then the Riemannian metric g induces a dual inner product ∗ structure on Tp M via i j ij h!idx ; ηjdx i = g !iηj: k ∗ More generally, one can define an inner product on Λ Tp M as follows: For any 1 m ∗ orthogonal basis θ ; ··· ; θ of Tp M, we require the set i1 ik fθ ^ · · · ^ θ j i1 < ··· < ikg k ∗ form an orthonormal basis of Λ Tp M. One can check that this definition is inde- pendent of the choice of θi's. Note that in particular we have h!g;!gi = 1 since in normal coordinates (gij) = I at p. As in the case of functions, the pointwise inner product induces an L2 inner k product structure on Ωc (M) via Z (!; η) := h!; ηi!g: M To define the Hodge-Laplacian of a differential form, one need to define the so-called Hodge star operator. We first use the pointwise inner product to get an k ∗ k ∗ ∗ k ∗ identification between Λ Tp M and (Λ Tp M) that sends β 2 Λ Tp M to k ∗ m ∗ Lβ :Λ Tp M ! R = Λ Tp M; α 7! hα; βi!g: On the other hand, the wedge product gives us a non-degenerate pairing k ∗ m−k ∗ m ∗ ^ :Λ Tp M × Λ Tp M ! R = Λ Tp M; (α; β) 7! α ^ β: k ∗ ∗ m−k ∗ which identifies any element in (Λ Tp M) as an element in Λ Tp M. In particular, k ∗ m−k ∗ for β 2 Λ Tp M one can get an element ?β 2 Λ Tp M that is identified with Lβ, i.e. α ^ ?β = hα; βi!g: 1 2 LECTURE 25: THE HODGE LAPLACIAN This construction gives us a linear isomorphism k ∗ m−k ∗ ? :Λ Tp M ! Λ Tp M at each point. Glue these constructions together, we are able to define Definition 1.1. The Hodge star operator ? :Ωk(M) ! Ωm−k(M) maps any k-form η 2 Ωk(M) to the (m − k)-form ?η 2 Ωm−k(M) so that for any ! 2 Ωk(M), ! ^ ?η = h!; ηi!g: Remark. Obviously ? is C1(M)-linear. 2 k Remark. The L inner product structure on Ωc (M) can be written as Z (!; η) = ! ^ ?η: M Note that by definition, ?1 = !g; ?!g = 1;! ^ ?η = η ^ ?!: More generally, if we let !1; ··· ;!m be a (local) basis of T ∗M so that 1 m ! ^ · · · ^ ! = f!g is a positive m-form, then for i1 < ··· < ik, if we let j1 < ··· < jm−k be the complement indeces of i's, i.e., such that fi1; ··· ; ik; j1; ··· ; jm−kg = f1; ··· ; mg; we have h!; !i ?(!i1 ^ · · · ^ !ik ) = ± !j1 ^ · · · ^ !jm−k ; f where we denoted ! = !i1 ^ · · · ^ !ik , and the sign ± is chosen so that ! ^ ?! = h!; !i!g; i.e. so that !i1 ^ · · · ^ !ik ^ !j1 ^ · · · ^ !jm−k = ±!1 ^ · · · ^ !m: One can check that in this case ± = (−1)i1+···+ik+1+···+k: Lemma 1.2. For ! 2 Ωk(M), ??! = (−1)k(m−k)!: Proof. By C1(M)-linearity, we may assume without loss of generality that ! = !i1 ^ · · · ^ !ik ; where !1; ··· ;!m is an orthonormal basis at one point p. Then the above compu- tations show ?! = (−1)i1+···+ik+1+···+k!j1 ^ · · · ^ !jm−k LECTURE 25: THE HODGE LAPLACIAN 3 and thus ??! = (−1)i1+···+ik+1+···+k(−1)j1+···+jm−k+1+···+m−k!: It remains to check the following elementary identity m(m + 1) k(k + 1) (m − k)(m − k + 1) + + ≡ k(m − k) (mod2): 2 2 2 As a consequence, we see ? is in fact a linear isometry: Corollary 1.3. For any !; η 2 Ωk(M), one has h?!; ?ηi = h!; ηi: Proof. We have k(m−k) h?!; ?ηi!g = (?!) ^ ?(?η) = (−1) (?!) ^ η = η ^ ?! = hη; !i!g: So h?!; ?ηi = hη; !i = h!; ηi. Remark. The Hodge star operator is of particular important in dimension 4. In fact, 2 ∗ 2 ∗ for m = 4 and k = 2, the linear map ? :Λ Tp M ! Λ Tp M satisfies ?2 = I: So one can decompose (according to eigenvalues of ?) 2 ∗ 2 ∗ 2 ∗ Λ Tp M = Λ+Tp M ⊕ Λ−Tp M: 2 ∗ 2 ∗ Sections of Λ+T M are called self-dual 2-forms, while sections of Λ−T M are called anti-self-dual 2-forms. 2. The Hodge-Laplace operator Using the Hodge star operator, one can define Definition 2.1. The co-differential of ! 2 Ωk(M) is δ! 2 Ωk−1(M) defined by δ! = (−1)km+m+1 ? d ? : k 2 The next lemma states that when we endow all Ωc (M)'s with this L structure, k k−1 the co-differential operator δ :Ωc (M) ! Ωc (M) is the adjoint of the differential k−1 k operator d :Ωc (M) ! Ωc (M). k k−1 Lemma 2.2. For any ! 2 Ωc (M) and η 2 Ωc (M), (!; dη) = (δ!; η): Proof. By Stokes' theorem, we have Z Z Z (!; dη) = (dη; !) = dη^?! = d(η^?!)−(−1)k−1η^d?! = (−1)k η^d?! M M M 4 LECTURE 25: THE HODGE LAPLACIAN On the other hand, by lemma 1.2, Z Z ((−1)km+m+1?d?!; η) = η^(−1)m(k+1)+1??d?! = (−1)km+m+1(−1)(m−k+1)(k−1) η^d?!; M M km+m+1 (m−k+1)(k−1) k so the conclusion follows from the fact (−1) (−1) = (−1) . The following formula will be useful. i Proposition 2.3. Let feig be an orthonormal frame and f! g the dual frame. Let r be the Levi-Civita connection. Then i (1) d = ! ^ re . P i (2) δ = − j ιej rej . Proof. (1) One can check that the right hand side is independent of choice of basis. So at each point p that is fixed, with out loss of generality one may take ei = @i to be the coordinate vector field for a normal coordinate system centered at p. The dual basis is then dxi. Recall that by definition, at the point p one has, for any i; j; k, j j j (r@i dx )(@k) = r@i (dx (@k)) − dx (r@i @j) = 0: j So at p one has r@i dx = 0 for any i; j. Now we denote ¯ i i d = ! ^ rei = dx ^ r@i : Consider η = fdxi1 ^ · · · ^ dxik . Then at p, @f dη¯ = dxi ^ dxi1 ^ · · · ^ dxik = dη: @xi This implies d = d¯. (2) The proof is similar. We denote ¯ X X δ = − ιej rej = − ι@j r@j : j j Then at p, X ¯ j−1 i1 ij ik δη = − (−1) (@ij f)dx ^ · · · ^ dxd ^ · · · ^ dx : j On the other hand, by the definition of δ one can calculate δη and prove that at p, X j i1 ij ik δη = (−1) (@ij f)dx ^ · · · ^ dxd ^ · · · ^ dx : j This completes the proof. Definition 2.4. The Hodge-Laplace operator on k-forms is ∆ = dδ + δd :Ωk(M) ! Ωk(M): LECTURE 25: THE HODGE LAPLACIAN 5 Remark. Since d2 = 0;?2 = ±1, we immediately get δ2 = 0: As a consequence, ∆ = (d + δ)2: Example. One can check that when k = 0, the operator ∆ = δd equals with the Laplace-Beltrami operator ∆ that we defined in lecture 3. To see this, again we do computation in normal coordinates. Then for any f 2 C1(M) we have j df = (@jf)dx and thus X j−1 1 j m ?df = (@jf)(−1) dx ^ · · · ^ dxc ^ · · · ^ dx ; which implies X 1 m d ? df = @j(@jf)dx ^ · · · ^ dx : It follows X ∆f = δdf = − ? d ? df = − @j(@jf); which is exactly the Laplace-Beltrami operator we defined in lecture 3 (but now calculated in normal coordinates). One can also see this by applying proposition 2.3: 2 ∆f = δdf = −ιej rej df = −tr(r f): Like the Beltrami-Laplacian, the Hodge-Laplacian also have very nice proposi- tions: Proposition 2.5. We have (1)( !; ∆η) = (∆!; η), i.e. ∆ is symmetric. (2) (∆!; !) = jδ!j2 + jd!j2 ≥ 0, i.e. ∆ is non-negative. (3) ?∆ = ∆?. k Proof. By lemma 2.2, for any !; η 2 Ωc (M), (!; ∆η) = (!; dδη) + (!; δdη) = (δ!; δη) + (d!; dη): Both (1) and (2) follows. To prove (3), we let ! be any k-form. Then ?δ! = (−1)km+m+1 ? ?d ? ! = (−1)km+m+1(−1)(m−k+1)(k−1)d ? ! = (−1)kd ? !: Similarly δ ? ! = (−1)(m−k)m+m+1 ? d ? ?! = (−1)(m−k)m+m+1(−1)k(m−k) ? d! = (−1)k+1 ? d!: So we get ?dδ! = (−1)kδ ? δ! = δd ? ! 6 LECTURE 25: THE HODGE LAPLACIAN and ?δd! = (−1)k+1d ? d! = dδ ? !: It follows ?∆ = ?dδ + ?δd = δd ? ! + dδ? = ∆ ?: In problem set 1 we have seen that if M is connected, then ∆f = 0 if and only if f is a constant function. Corollary 2.6. ∆(f!g) = 0 if and only if f is a constant function. Proof. This follows from ∆(f!g) = ∆ ? f = ?∆f = (∆f)!g: Definition 2.7. A k-form ! is called harmonic if ∆! = 0. We will denote the set of all harmonic k-forms on (M; g) by Hk(M). It is obviously a vector space. Obviously if M is connected, 0 m H (M) ' R; H (M) ' R: According to proposition 2.3, if ! 2 Ωk(M) is parallel, i.e.

View Full Text

Details

  • File Type
    pdf
  • Upload Time
    -
  • Content Languages
    English
  • Upload User
    Anonymous/Not logged-in
  • File Pages
    6 Page
  • File Size
    -

Download

Channel Download Status
Express Download Enable

Copyright

We respect the copyrights and intellectual property rights of all users. All uploaded documents are either original works of the uploader or authorized works of the rightful owners.

  • Not to be reproduced or distributed without explicit permission.
  • Not used for commercial purposes outside of approved use cases.
  • Not used to infringe on the rights of the original creators.
  • If you believe any content infringes your copyright, please contact us immediately.

Support

For help with questions, suggestions, or problems, please contact us