Ray-Singer Torsion and the Rumin Laplacian on Lens Spaces

We express explicitly the analytic torsion functions associated with the Rumin complex on lens spaces in terms of the Hurwitz zeta function. In particular, we find that the functions vanish at the origin and determine the analytic torsions. Moreover, we have a formula between this torsion and the Ra...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:Symmetry, Integrability and Geometry: Methods and Applications
Datum:2022
1. Verfasser: Kitaoka, Akira
Format: Artikel
Sprache:Englisch
Veröffentlicht: Інститут математики НАН України 2022
Online Zugang:https://nasplib.isofts.kiev.ua/handle/123456789/211813
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Назва журналу:Digital Library of Periodicals of National Academy of Sciences of Ukraine
Zitieren:Ray-Singer Torsion and the Rumin Laplacian on Lens Spaces. Akira Kitaoka. SIGMA 18 (2022), 091, 16 pages

Institution

Digital Library of Periodicals of National Academy of Sciences of Ukraine
_version_ 1859821195203969024
author Kitaoka, Akira
author_facet Kitaoka, Akira
citation_txt Ray-Singer Torsion and the Rumin Laplacian on Lens Spaces. Akira Kitaoka. SIGMA 18 (2022), 091, 16 pages
collection DSpace DC
container_title Symmetry, Integrability and Geometry: Methods and Applications
description We express explicitly the analytic torsion functions associated with the Rumin complex on lens spaces in terms of the Hurwitz zeta function. In particular, we find that the functions vanish at the origin and determine the analytic torsions. Moreover, we have a formula between this torsion and the Ray-Singer torsion.
first_indexed 2026-03-16T12:21:46Z
format Article
fulltext Symmetry, Integrability and Geometry: Methods and Applications SIGMA 18 (2022), 091, 16 pages Ray–Singer Torsion and the Rumin Laplacian on Lens Spaces Akira KITAOKA Graduate School of Mathematical Sciences, The University of Tokyo, Japan E-mail: akira5kitaoka@gmail.com URL: https://akira5kitaoka.github.io/Akira5Kitaoka-en.github.io/ Received May 19, 2022, in final form November 14, 2022; Published online November 28, 2022 https://doi.org/10.3842/SIGMA.2022.091 Abstract. We express explicitly the analytic torsion functions associated with the Rumin complex on lens spaces in terms of the Hurwitz zeta function. In particular, we find that the functions vanish at the origin and determine the analytic torsions. Moreover, we have a formula between this torsion and the Ray–Singer torsion. Key words: analytic torsion; Rumin complex; CR geometry; contact geometry 2020 Mathematics Subject Classification: 58J52; 32V20; 53D10; 43A85 1 Introduction Let (M,H) be a compact contact manifold of dimension 2n + 1 and E be the flat vector bun- dle with unitary holonomy on M . Rumin [12] introduced a complex (E•(M,E), d•R), which is a subquotient of the de Rham complex of E. A specific feature of the complex is that the operator D = dnR : En(M,E) → En+1(M,E) in ‘middle degree’ is a second-order, while dkR : Ek(M,E) → Ek+1(M,E) for k ̸= n are first order which are induced by the exterior deriva- tives. Let ak = 1/ √ |n− k| for k ̸= n and an = 1. Then, (E•(M,E), d•E), where dkE = akd k R, is also a complex. We call (E•(M,E), d•E) the Rumin complex. In virtue of the rescaling, d•E satis- fies Kähler-type identities on Sasakian manifolds [13, equation (34)], which include the case of lens spaces. Let θ be a contact form ofH and J be an almost complex structure onH. Then we may define a Riemann metric gθ,J on TM by extending the Levi metric dθ(−, J−) on H (see Section 2.1). Following [12], we define the Rumin Laplacians ∆E associated with (E•(M,E), d•E) and the metric gθ,J by ∆k E :=  (dEdE †)2 + (dE †dE) 2, k ̸= n, n+ 1, (dEdE †)2 +D†D, k = n, DD† + (dE †dE) 2, k = n+ 1. Rumin showed that ∆E has discrete eigenvalues with finite multiplicities. We next introduce the analytic torsion and metric of the Rumin complex (E•(M,E),d•E) by following [2, 8, 14]. We define the contact analytic torsion function associated with (E•(M,E), d•E) by κE(M,E, gθ,J)(s) := n∑ k=0 (−1)k+1(n+ 1− k)ζ ( ∆k E ) (s), (1.1) where ζ ( ∆k E ) (s) is the spectral zeta function of ∆k E . Since the Rumin Laplacians ∆E satisfies the Rockland condition by [12, p. 300], the spectral zeta function ζ ( ∆k E ) (s) extends to a meromorphic akira5kitaoka@gmail.com https://akira5kitaoka.github.io/Akira5Kitaoka-en.github.io/ https://doi.org/10.3842/SIGMA.2022.091 2 A. Kitaoka function on C which is holomorphic at zero by [9, Section 4]. Here, we use 0s = 1. We define the contact analytic torsion TE by 2 log TE(M,E, gθ,J) = κE(M,E, gθ,J) ′(0). Let H•(E•,d•E) be the cohomology of the Rumin complex. We define the contact metric on detH•(E•,d•E) by ∥ ∥E(M,E, gθ,J) = T−1 E (M,E, gθ,J)| |L2(E•), where the metric | |L2(E•) is induced by L2 metric on E•(M,E) via identification of the coho- mology classes by the harmonic forms on E•(M,E). Rumin and Seshadri [14] defined another analytic torsion function κR from (E•(M,E),d•R), which is different from κE except in dimension 3. In dimension 3, they showed that κR(M,E, gθ,J)(0) is a contact invariant, that is, independent of the metric gθ,J . Moreover, on 3-dimensional Sasakian manifolds with S1-action, κR(M,E, gθ,J)(0) = 0. Here, Sasakian manifolds with S1- action means Sasakian manifolds whose Reeb vector filed generates the circle action S1. 3-dimen- sional Sasakian manifolds with S1-action are CR Seifert manifolds. Furthermore, they showed that this analytic torsion and the Ray–Singer torsion TdR(M,E, gθ,J) equal for flat bundles with unitary holonomy on 3-dimensional Sasakian manifolds with S1 action. To extend the coincidence, with dE instead of dR, the author [8] showed that TE ( S2n+1,C, gθ,J ) = n!TdR ( S2n+1,C, gθ,J ) on the standard CR spheres S2n+1 ( ⊂Cn+1 ) , where C is the trivial line bundle. Here the standard CR sphere is triple ( S2n+1, θ, J ) , where θ is given the contact form by θ = √ −1 ( ∂̄ − ∂ ) |z|2 and J is an almost complex structure J induced from the complex structure of Cn+1. It is simply denoted by S2n+1. Moreover, Albin and Quan [1, Corollary 3 and equation (4)] showed the difference between the Ray–Singer torsion and the contact analytic torsion is given by some integrals of universal polynomials in the local invariants of the metric on contact manifolds. In this paper, we extend this coincidence on lens spaces and determine explicitly the analytic torsion functions associated with the Rumin complex in terms of the Hurwitz zeta function. Let gstd be the standard metric on S2n+1 and we note that gθ,J = 4gstd. Let µ, ν1, . . . , νn+1 be integers such that the νj are coprime to µ. Let Γ be the subgroup of ( S1 )n+1 generated by γ = (γ1, . . . , γn+1) := ( exp ( 2π √ −1ν1/µ ) , . . . , exp ( 2π √ −1νn+1/µ )) . We denote the lens space by K := S2n+1/Γ. Let C be the trivial line bundle on K. Fix u ∈ Z and consider the unitary representation αu : π1(K) = Γ → U(1), defined by αu ( γℓ ) := exp ( 2π √ −1uℓ/µ ) for ℓ ∈ Z. Let Eα be the flat vector bundle associated with the unitary representation α : π1(K) = Γ→U(r), and Eαu = Eu. The sections of this bundle correspond to αu-equivariant functions on S 2n+1. Our main result is Theorem 1.1. Let K be the lens space with the contact form and the almost complex structure which are induced by the action Γ on the standard CR sphere S2n+1. (1) The contact analytic torsion function of (K,C) is given by κE(K,C, gθ,J)(s) = −(n+ 1) ( 1 + 22s+1µ−2sζ(2s) ) , (1.2) Ray–Singer Torsion and the Rumin Laplacian on Lens Spaces 3 where ζ is the Riemann zeta function. In particular, we have κE(K,C, gθ,J)(0) = 0, (1.3) TE(K,C, gθ,J) = ( 4π µ )n+1 . (1.4) (2) The contact analytic torsion function of (K,Eu) for u ∈ {1, . . . , µ− 1} is given by κE(K,Eu, gθ,J)(s) = −22sµ−2s n+1∑ j=1 ( ζ(2s,Aµ(uτj)/µ) + ζ(2s,Aµ(−uτj)/µ) ) , (1.5) where ζ(s, a) := ∑∞ q=0(q + a)−s is the Hurwitz zeta function for 0 < a ≤ 1, Aµ(w) is the integer between 1 and µ such that Aµ(w) ≡ w mod µ and τjνj ≡ 1 mod µ. In particular, we have κE(K,Eu, gθ,J)(0) = 0, (1.6) TE(K,Eu, gθ,J) = n+1∏ j=1 ∣∣e2π√−1uτj/µ − 1 ∣∣. (1.7) The equations (1.2) and (1.5) extend the following results of κE the spheres to on lens spaces. Rumin and Seshadri [14, Theorem 5.4] showed (1.2) in the case of 3-dimensional lens spaces. The author [8] showed (1.2) in the case of ( S2n+1,C ) for arbitrary n. From (1.3) and (1.6), we see that the metric ∥ ∥E on (K,Eu, gθ,J) is invariant under the constant rescaling θ 7→ Cθ. The argument is exactly the same as the one in [14]. In the same way as [8], the fact that the representations determine the eigenvalues of ∆E cause several cancellations in the linear combination (1.1), which significantly simplifies the computation of κE(s). We cannot get such a simple formula for the contact analytic torsion function κR of (E•(M,E),d•R) for dimensions higher than 3. Let us compare the contact analytic torsion with the Ray–Singer torsion on lens spaces. Ray [10] showed that for u(= 1, . . . , µ− 1) TdR(K,Eu, 4gstd) = n+1∏ j=1 ∣∣e2π√−1uτj/µ − 1 ∣∣. Weng and You [15] calculate the Ray–Singer torsion on spheres. We extend their results for the trivial bundle on lens spaces: Proposition 1.2. In the setting of Theorem 1.1, we have TdR(K,C, 4gstd) = (4π)n+1 n!µn+1 . The metric 4gstd agrees with the metric gθ,J defined from the contact form θ = √ −1 ( ∂̄−∂ ) |z|2. Since the cohomology of (E•(M,E),d•E) coincides with that of (Ω•(M,E), d) (e.g., [12, p. 286]), there is a natural isomorphism detH•(E•(M,E), d•E) ∼= detH•(Ω•(M,E),d), which turns out to be isometric for the L2 metrics. Therefore (1.4) and (1.7) give Corollary 1.3. In the setting of Theorem 1.1, for all unitary holonomy α : π1(K) → U(r), we have TE(K,Eα, gθ,J) = n!dimH0(K,Eα)TdR(K,Eα, gθ,J), ∥ ∥E(K,Eα, gθ,J) = n!−dimH0(K,Eα)∥ ∥dR(K,Eα, gθ,J), via the isomorphism detH•(E•(M,Eα), d • E) ∼= detH•(Ω•(M,Eα), d). 4 A. Kitaoka The paper is organized as follows. In Section 2, we recall the definition and properties of the Rumin complex on S2n+1. In Section 3, we calculate the contact analytic torsion function κE of flat vector bundles on lens spaces. In Section 4, we compute the Ray–Singer torsion TdR of the trivial vector bundle. In Section 5, we compare the Ray–Singer torsion and the contact analytic torsion. 2 The Rumin complex 2.1 The Rumin complex on contact manifolds We call (M,H) an orientable contact manifold of dimension 2n+1 if H is a subbundle of TM of codimension 1 and there exists a 1-form θ, called a contact form, such that Ker(θ : TM→R) = H and θ ∧ (dθ)n ̸= 0. The Reeb vector field of θ is the unique vector field T satisfying θ(T ) = 1 and IntT dθ = 0, where IntT is the interior product with respect to T . ForH and θ, we call J ∈ End(TM) an almost complex structure associated with θ if J2 = − Id on H, JT = 0, and the Levi form dθ(−, J−) is positive definite on H. Given θ and J , we define a Riemannian metric gθ,J on TM by gθ,J(X,Y ) := dθ(X, JY ) + θ(X)θ(Y ) for X,Y ∈ TM. Let ∗ be the Hodge star operator on ∧•T ∗M with respect to gθ,J . Let M be a manifold, M̃ be the universal covering of M , π1(M) be the fundamental group of M . For each unitary representation α : π1(M) → U(r), we denote the flat vector bundle associated with α by Eα := M̃ ×α Cr →M. Let ∇α be the flat connection on Eα induced from the trivial connection on M̃ × Cr, and d∇α be the exterior covariant derivative of ∇α. The Rumin complex [12] is defined on contact manifolds as follows. We set L := dθ∧ and Λ := ∗−1L∗, which is the adjoint operator of L with respect to the metric gθ,J at each point. We set∧k primH ∗ := { v ∈ ∧kH∗ ∣∣Λv = 0 } ,∧k LH ∗ := { v ∈ ∧kH∗ ∣∣Lv = 0 } , Ek(M,Eα) := { C∞(M, ∧k primH ∗ ⊗ Eα ) , k ≤ n, C∞(M, θ ∧ ∧k−1 L H∗ ⊗ Eα ) , k ≥ n+ 1. We embed H∗ into T ∗M as the subbundle {ϕ ∈ T ∗M |ϕ(T ) = 0} so that we can regard Ωk H(M,Eα) := C∞(M, ∧kH∗ ⊗ Eα ) as a subspace of Ωk(M), the space of k-forms. We define db : Ω k H(M,Eα) → Ωk+1 H (M,Eα) by dbϕ := d∇αϕ− θ ∧ ( IntT d∇αϕ ) , and then D : En(M,Eα) → En+1(M,Eα) by D = θ ∧ ( LT + dbL −1db ) , where LT is the Lie derivative with respect to T , and we use the fact that L : ∧n−1H∗ →∧n+1H∗ is an isomorphism. Ray–Singer Torsion and the Rumin Laplacian on Lens Spaces 5 Let P : ∧kH∗ → ∧k primH ∗ be the fiberwise orthogonal projection with respect to gθ,J , which also defines a projection P : Ωk(M,Eα) → Ek(M,Eα). We set dkR :=  P ◦ d∇α on Ek(M,Eα), k ≤ n− 1, D on En(M,Eα), d∇α on Ek(M,Eα), k ≥ n+ 1. Then (E•(M,Eα), d • R) is a complex. Let dkE = akd k R, where ak = 1/ √ |n− k| for k ̸= n and an = 1. We call (E•(M,Eα),d • E) the Rumin complex. We define the L2-inner product on Ωk(M,Eα) by (ϕ, ψ) := ∫ M gθ,J(ϕ, ψ) d volgθ,J and the L2-norm on Ωk(M,Eα) by ∥ϕ∥ := √ (ϕ, ϕ). Since the Hodge star operator ∗ induces a bundle isomorphism from ∧k primH ∗ to θ ∧ ∧2n−k L H∗, it also induces a map Ek(M,Eα) → E2n+1−k(M,Eα). We note that Ek(M,Eα) = { ϕ ∈ Ek ( M̃,Cr ) | t∗ϕ = α(t)−1ϕ for t ∈ π1(M) } . Let d†E and D† denote the formal adjoint of dE and D, respectively for the L2-inner product. We define the fourth-order Laplacians ∆E on Ek(M,Eα) by ∆k E :=  ( dk−1 E dk−1 E †)2 + (dkE†dkE)2, k ̸= n, n+ 1,( dn−1 E dn−1 E †)2 +D†D, k = n, DD† + ( dn+1 E †dn+1 E )2 , k = n+ 1. We call it the Rumin Laplacian [12]. Since ∗ and ∆E commute, to determine the eigenvalue on E•(M,Eα), it is enough to compute them on Ek(M,Eα) for k ≤ n. 2.2 The Rumin complex on the CR spheres Let S := {z ∈ Cn+1 | |z|2 = 1} and θ := √ −1 ( ∂̄−∂ ) |z|2. (We will omit the dimension from S2n+1 for the simplicity of the notation.) Let gstd be the standard metric on S. Then, gθ,J coincides with 4gstd. With respect to the standard almost complex structure J , we decompose the bundles defined in the previous subsection as follows: H∗1,0 := { v ∈ CH∗ | Jv = √ −1v } , H∗0,1 := { v ∈ CH∗ | Jv = − √ −1v } ,∧i,j H∗ := ∧iH∗1,0 ⊗ ∧j H∗0,1,∧i,j primH ∗ := { ϕ ∈ ∧i,j H∗ ∣∣Λϕ = 0 } , E i,j := C∞(S,∧i,j primH ∗). We decompose E i,j into a direct sum of irreducible representations of the unitary group U(n+1). Recall that irreducible representations of U(m) are parametrized by the highest weight λ = (λ1, . . . , λm) ∈ Zm with λ1 ≥ λ2 ≥ · · · ≥ λm; the representation corresponding to λ will be denoted by V (λ). To simplify the notation, we introduce the following notation: for a1, . . ., al ∈ Z and k1, . . . , kl ∈ Z, (a1k1 , . . . , alkl) denotes the k1 + · · · + kl-tuple whose first k1 entries are a1, whose next k2 entries are a2, etc. For example, (13, 02,−12) = (1, 1, 1, 0, 0,−1,−1). We note that a1 is a, and a0 is the zero tuple. 6 A. Kitaoka In [7], it is shown that the multiplicity of V (q, 1j , 0n−1−i−j ,−1i,−p) in Es,t is at most one. Thus we may set Ψ (s,t) (q,j,i,p) := Es,t ∩ V (q, 1j , 0n−1−i−j ,−1i,−p). Proposition 2.1 ([7, Section 4(b)]). Given (q, j, i, p), we list up all (s, t) such that s + t ≤ n and Ψ (s,t) (q,j,i,p) ̸= {0} as the following: Case I: For i = j = 0 and p = q = 0, the space is Ψ (0,0) (0,0,0,0). Case II: For i+ j ≤ n− 2, p ≥ 1 and q ≥ 1, the spaces are Ψ (i,j) (q,j,i,p), Ψ (i+1,j) (q,j,i,p), Ψ (i,j+1) (q,j,i,p), Ψ (i+1,j+1) (q,j,i,p) . Case III: For 0 ≤ i ≤ n− 1, j = 0, p ≥ 1 and q = 0, the spaces are Ψ (i,0) (0,0,i,p), Ψ (i+1,0) (0,0,i,p). Case IV: For i = 0, 0 ≤ j ≤ n− 1, p = 0 and q ≥ 1, the spaces are Ψ (0,j) (q,j,0,0), Ψ (0,j+1) (q,j,0,0). Case V: For i+ j = n− 1, p ≥ 1 and q ≥ 1, the spaces are Ψ (i,j) (q,j,i,p), Ψ (i+1,j) (q,j,i,p), Ψ (i,j+1) (q,j,i,p). Case VI: i = n− 1, j = 0, p ≥ 1 and q = −1, the space is Ψ (n,0) (−1,0,n−1,p). Case VII: i = 0, j = n− 1, p = −1 and q ≥ 1, the space is Ψ (0,n) (q,n−1,0,−1). Remark 2.2. About Case VI, we substitute s = n, t = 0, q = −1, j = 0, i = n−1, for Ψ (s,t) (q,j,i,p), Ψ (n,0) (−1,0,n−1,p) = En,0 ∩ V (−1, 10, 00,−1n−1,−p) = En,0 ∩ V (−1n,−p). As the same way, about VII we obtain Ψ (0,n) (q,n−1,0,−1) = E0,n ∩ V (q, 1n−1, 00,−10,−(−1)) = E0,n ∩ V (q, 1n). In [8], ∆E acts as a scalar on these spaces, and eigenvalues of ∆E on the standard CR spheres are explicitly determined. Proposition 2.3 ([8, Theorem 0.1]). Let S be the standard CR sphere with the contact form θ = √ −1 ( ∂̄ − ∂ ) |z|2. Then, on the subspaces of the complexification of E•(S) corresponding to the representations Ψ (•,•) (q,j,i,p), the eigenvalue of ∆E is( (p+ i)(q + n− i) + (q + j)(p+ n− j) )2 4(n− i− j)2 . Ray–Singer Torsion and the Rumin Laplacian on Lens Spaces 7 3 Contact analytic torsion of flat vector bundles Let µ, ν1, . . . , νn+1 be integers such that the νj are coprime to µ. Let Γ be the subgroup of ( S1 )n+1 generated by γ = (γ1, . . . , γn+1) := ( exp ( 2π √ −1ν1/µ ) , . . . , exp ( 2π √ −1νn+1/µ )) . We denote the lens space by K := S2n+1/Γ. Fix u ∈ {1, . . . , µ} and consider the unitary representation αu : π1(K) = Γ → U(1), defined by αu ( γℓ ) := exp ( 2π √ −1uℓ/µ ) for ℓ ∈ Z, where γℓ := ( γℓ1, . . . , γ ℓ n+1 ) = ( exp ( 2π √ −1ν1ℓ/µ ) , . . . , exp ( 2π √ −1νn+1ℓ/µ )) . Let Eα be the flat vector bundle associated with the unitary representation α : π1(K) = Γ→U(r), and Eu := Eαu , which can be considered as αu-equivariant functions on S 2n+1. For each unitary representation (V, ρ) of U(n + 1), we define the vector subspace V αu of V by V αu := { ϕ ∈ V ∣∣ ρ(γ)ϕ = α(γ)−1ϕ } . Proposition 3.1. We have κE(K,Eu, gθ,J)(s) = κ1(K,Eu, gθ,J)(s) + κ2(K,Eu, gθ,J)(s) + κ3(K,Eu, gθ,J)(s), (3.1) where κ1(K,Eu, gθ,J)(s) := −(n+ 1) dimV αu(0n+1) = { −(n+ 1), u = 0, 0, u ̸= 0, (3.2) κ2(K,Eu, gθ,J)(s) := (−1)1(n+ 1) ∑ q≥1 dimV αu(q, 0n)( q/2 )2s + n∑ j=1 (−1)j+1(n+ 1− j) ∑ q≥1 ( dimV αu(q, 1j , 0n−j)( (q + j)/2 )2s + dimV αu(q, 1j−1, 0n−j+1)( (q + j − 1)/2 )2s ) , κ3(K,Eu, gθ,J)(s) := κ2(K,E−u, gθ,J)(s), (3.3) Proof. From Proposition 2.3, we check that the terms of κE(K,Eu, g)(s) in Cases II and V in Proposition 2.1 cancel each other. “The sum of the terms of κE(K,Eu, gθ,J)(s) in Case II” is n−2∑ a=0 ∑ i+j=a ∑ p≥1, q≥1 ( (−1)i+j+1(n+ 1− i− j) + 2(−1)i+j+2(n− i− j) + (−1)i+j+3(n− 1− i− j) ) × dimV αu(q, 1j , 0n−1−i−j ,−1i,−p)( ((p+ i)(q + n− i) + (q + j)(p+ n− j))/2(n− i− j) )2s 8 A. Kitaoka = n−2∑ a=0 ∑ i+j=a ∑ p≥1, q≥1 ( 0 ) dimV αu(q, 1j , 0n−1−i−j ,−1i,−p)( ((p+ i)(q + n− i) + (q + j)(p+ n− j))/2(n− i− j) )2s = 0. Similarly, “the sum of the terms of κE(K,Eu, gθ,J)(s) in Case V” is∑ i+j=n−1 ∑ p≥1, q≥1 ( (−1)n(n+ 1− (n− 1))) + 2(−1)n+1(n+ 1− n) ) × dimV αu(q, 1j ,−1i,−p)( ((p+ i)(q + n− i) + (q + j)(p+ n− j))/2(n+ 1− i− j) )2s = ∑ i+j=n−1 ∑ p≥1, q≥1 ( 0 ) dimV αu(q, 1j ,−1i,−p)( ((p+ i)(q + n− i) + (q + j)(p+ n− j))/2(n− i− j) )2s = 0. The function κ1(K,Eu, gθ,J)(s) is the sum of the terms of κE(K,Eu, gθ,J)(s) in Case I. Next we consider the sum of the terms of κE in Cases III and VI. For j, “the sum of the terms of κE(K,Eu, gθ,J) in E0,j , Cases III and VI” is (−1)1(n+ 1) ∑ q≥1 dimV αu(q, 0n)( q/2 )2s , j = 0, (−1)j+1(n+ 1− j) ∑ q≥1 ( dimV αu(q, 1j , 0n−j)( (q + j)/2 )2s + dimV αu(q, 1j−1, 0n−j+1)( (q + j − 1)/2 )2s ) , 1 ≤ j ≤ n. “The sum of the terms of κE in Cases III and VI” is (−1)1(n+ 1) ∑ q≥1 dimV αu(q, 0n)( q/2 )2s + n∑ j=1 (−1)j+1(n+ 1− j) ∑ q≥1 ( dimV αu(q, 1j , 0n−j)( (q + j)/2 )2s + dimV αu(q, 1j−1, 0n−j+1)( (q + j − 1)/2 )2s ) = κ2(K,Eu, gθ,J)(s). (3.4) Finally, we consider the sum of the terms of κE in Cases IV and VII. As the same way (3.4) in Cases III and VI, “the sum of the terms of κE in Cases IV and VII” is given by (−1)1(n+ 1) ∑ q≥1 dimV αu(0n,−q)( q/2 )2s (3.5) + n∑ j=1 (−1)j+1(n+ 1− j) ∑ q≥1 ( dimV αu(0n−j ,−1j ,−q)( (q + j)/2 )2s + dimV αu(0n−j+1,−1j−1,−q)( (q + j − 1)/2 )2s ) . Let (V, ρ) be the unitary representation of U(n+ 1). We define the representation (V , ρ) by V := V, ρ(U) := ρ(U) for U ∈ U(n+ 1). Since (V (q, 1j , 0n−j), ρ) is the unitary representation, its conjugate representation is isomor- phic to its dual representation as U(n + 1)-module. From [6, Theorem 3.2.13], the conjugate representation of (V (q, 1j , 0n−j), ρ) is isomorphic to (V (0n−j ,−1j ,−q), ρ) as U(n + 1)-module. Therefore, we have V α−u(q, 1j , 0n−j) = { ϕ ∈ V (q, 1j , 0n−j) ∣∣α−u(−γ)ϕ = ρ(γ)ϕ } Ray–Singer Torsion and the Rumin Laplacian on Lens Spaces 9 = { ϕ ∈ V (q, 1j , 0n−j) ∣∣αu(−γ)−1ϕ = ρ(γ)ϕ } = { ϕ ∈ V (q, 1j , 0n−j) ∣∣αu(γ)ϕ = ρ(γ)ϕ } ∼= { ϕ ∈ V (0n−j ,−1j ,−q) ∣∣αu(γ)ϕ = ρ(γ)ϕ } = { ϕ ∈ V (0n−j ,−1j ,−q) ∣∣αu(γ) −1ϕ = ρ(γ)ϕ } = V (0n−j ,−1j ,−q) αu , where ∼= means isomorphic as real vector spaces via the complex conjugate. Then, from (3.4) and (3.5), “the sum of the terms of κE in Cases IV and VII” is given by (−1)1(n+ 1) ∑ q≥1 dimV αu(0n,−q)( q/2 )2s + n∑ j=1 (−1)j+1(n+ 1− j) ∑ q≥1 ( dimV αu(0n−j ,−1j ,−q)( (q + j)/2 )2s + dimV αu(0n−j+1,−1j−1,−q)( (q + j − 1)/2 )2s ) = (−1)1(n+ 1) ∑ q≥1 dimV α−u(q, 0n)( q/2 )2s + n∑ j=1 (−1)j+1(n+ 1− j) ∑ q≥1 ( dimV α−u(q, 1j , 0n−j)( (q + j)/2 )2s + dimV α−u(q, 1j−1, 0n−j+1)( (q + j − 1)/2 )2s ) = κ2(K,E−u, gθ,J)(s). ■ We set for q ≥ 1, V (q, 1−1, 0n+1) := {0}. We have κ2(K,Eu, gθ,J)(s) = (−1)1(n+ 1) ∑ q≥1 dimV αu(q, 0n)( q/2 )2s + n∑ j=1 (−1)j+1(n+ 1− j) ∑ q≥1 ( dimV αu(q, 1j , 0n−j)( (q + j)/2 )2s + dimV αu(q, 1j−1, 0n−j+1)( (q + j − 1)/2 )2s ) = n∑ j=0 (−1)j+1(n+ 1− j) ∑ q≥1 ( dimV αu(q, 1j , 0n−j)( (q + j)/2 )2s + dimV αu(q, 1j−1, 0n−j+1)( (q + j − 1)/2 )2s ) = n∑ j=0 (−1)j+1(n+ 1− j) ∑ q≥1 dimV αu(q, 1j , 0n−j) + dimV αu(q + 1, 1j−1, 0n−j+1)( (q + j)/2 )2s + n∑ j=1 (−1)j+1(n+ 1− j) dimV αu(1j , 0n−j+1)( j/2 )2s . Let χV be the character of the representation (V, ρ) of U(n + 1). We note that for each representation (V, ρ), dimV αu = ∑ t∈Γ χV (t)αu(t)/#Γ, (cf. [5, equation (2.9)]). By Littlewood–Richardson’s rule (cf. [4, Corollary 3]), we have χV (1j ,0n−j+1) χV (q,0n) = χV (q,1j ,0n−j) + χV (q+1,1j−1,0n−j+1) . (3.6) 10 A. Kitaoka From (3.6), κ2(K,Eu, gθ,J)(s) = 1 µ n∑ j=0 (−1)j+1(n+ 1− j) ∑ q≥1 µ−1∑ ℓ=0 χV (1j ,0n−j+1) ( γℓ ) χV (q,0n) ( γℓ ) αu ( γℓ )( (q + j)/2 )2s + 1 µ n∑ j=1 (−1)j+1(n+ 1− j) µ−1∑ ℓ=0 χV (1j ,0n−j+1) ( γℓ ) αu ( γℓ )( j/2 )2s = 22s µΓ(2s) n∑ j=0 (−1)j+1(n+ 1− j) × ∑ q≥1 µ−1∑ ℓ=0 ∫ ∞ 0 χV (1j ,0n−j+1) ( γℓ ) χV (q,0n) ( γℓ ) αu ( γℓ ) e−(j+q)xx2s−1 dx + 22s µΓ(2s) n∑ j=1 (−1)j+1(n+ 1− j) µ−1∑ ℓ=0 ∫ ∞ 0 χV (1j ,0n−j+1) ( γℓ ) αu ( γℓ ) e−jxx2s−1 dx = 22s µΓ(2s) µ−1∑ ℓ=0 ∫ ∞ 0 ( n∑ j=0 (−1)j+1(n+ 1− j)χV (1j ,0n−j+1) ( γℓ ) e−jx ∑ q≥1 χV (q,0n) ( γℓ ) e−qx + n∑ j=1 (−1)j+1(n+ 1− j)χV (1j ,0n−j+1) ( γℓ ) e−jx ) αu ( γℓ ) x2s−1 dx. (3.7) We consider the contents of the integral for the last equation of (3.7). It is known that for t = (t1, . . . , tn+1) ∈ ( S1 )n+1 , χV (1j ,0n−j+1) (t) = ∑ β1+···+βn+1=j, 0≤β1,...,βn+1≤1 tβ1 1 · · · tβn+1 n+1 , (3.8) χV (q,0n) (t) = ∑ α1+···+αn+1=q α1,...,αn+1≥0 tα1 1 · · · tαn+1 n+1 , (3.9) (cf. [5, equations (6.1) and (6.2)]). We set X := e−x and F1(t,X) := n+1∑ j=0 (−1)jχV (1j ,0n−j+1) (t)Xj . Then (3.8) gives F1(t,X) = n+1∏ j=1 (1− tjX), (3.10) X ∂F1 ∂X (t,X) = n+1∑ j=0 (−1)jjχV (1j ,0n−j+1) (t)Xj = − n+1∑ i=1 tiX n+1∏ j=1,j ̸=i (1− tjX). (3.11) From the definition of F1 and (3.11), we have n∑ j=0 (−1)j+1(n+ 1− j)χV (1j ,0n−j+1) (t)Xj = −(n+ 1)F1(t,X) +X ∂F1 ∂X (t,X). (3.12) Ray–Singer Torsion and the Rumin Laplacian on Lens Spaces 11 We set F2(t,X) := ∑ q≥1 χV (q,0n) (t)Xq. From (3.9) and (3.10), we can rewrite F2 as F2(t,X) = n+1∏ j=1 1 1− tjX − 1 = 1 F1(t,X) − 1. (3.13) From (3.10)–(3.13), we can deduce that n∑ j=0 (−1)j+1(n+ 1− j)χV (1j ,0n−j+1) (t)Xj ∑ q≥1 χV (q,0n) (t)Xq + n∑ j=1 (−1)j+1(n+ 1− j)χV (1j ,0n−j+1) (t)Xj = − ( (n+ 1)F1(t,X)−X ∂F1 ∂X (t,X) )( 1 F1(t,X) − 1 ) − ( (n+ 1)(F1(t,X)− 1)−X ∂F1 ∂X (t,X) ) = X ∂F1 ∂X (t,X) F1(t,X) = − n+1∑ j=1 tjX 1− tjX . (3.14) From (3.7) and (3.14), we see κ2(K,Eu, gθ,J)(s) = − 22s µΓ(2s) µ−1∑ ℓ=0 n+1∑ j=1 ∫ ∞ 0 γℓje −x 1− γℓje −x e2π √ −1uℓ/µx2s−1 dx = − n+1∑ j=1 22s µΓ(2s) µ−1∑ ℓ=0 ∫ ∞ 0 ∞∑ q=1 exp ( 2π √ −1(qνj + u)ℓ/µ ) e−qxx2s−1 dx = −22s n+1∑ j=1 µ−1∑ ℓ=0 ∞∑ q=1 exp ( 2π √ −1(qνj + u)ℓ/µ ) µ q−2s. (3.15) Let τj be the integers in {1, . . . , µ} such that τjνj ≡ 1 mod µ. Since the multiplication of νj ∈ (Z/µZ)× induced the bijective map from Z/µZ to Z/µZ, We have µ−1∑ ℓ=0 exp ( 2π √ −1(qνj + u)ℓ/µ ) = µ−1∑ ℓ=0 exp ( 2π √ −1(q + uτj)νjℓ/µ ) = µ−1∑ ℓ=0 exp ( 2π √ −1(q + uτj)ℓ/µ ) = { 0, q ̸≡ −uτj mod µ, µ, q ≡ −uτj mod µ. For w ∈ Z let Aµ(w) be the integer between 1 and µ which is congruent to w modulo µ, then from (3.15), we can rewrite κ2 as κ2(K,Eu, gθ,J)(s) = −22s n+1∑ j=1 ∞∑ q>0, q≡−uτj mod µ q−2s = −22s n+1∑ j=1 ∞∑ q=0 ( qµ+Aµ(−uτj) )−2s 12 A. Kitaoka = −22sµ−2s n+1∑ j=1 ζ ( 2s,Aµ(−uτj)/µ ) , (3.16) where for 0 < a ≤ 1, ζ(s, a) := ∑∞ q=0(q + a)−s is the Hurwitz zeta function. Next, we calculate κ3. As the same way in calculating κ2, from (3.3), we can rewrite κ3 as κ3(K,Eu, gθ,J)(s) = −22sµ−2s n+1∑ j=1 ζ ( 2s,Aµ(uτj)/µ ) . (3.17) From (3.1), (3.2), (3.16) and (3.17), we have κE(K,Eu, gθ,J)(s) =  −(n+ 1) ( 1 + 22s+1µ−2sζ(2s) ) , u = 0. −22sµ−2s n+1∑ j=1 ( ζ ( 2s,Aµ(uτj)/µ ) + ζ ( 2s, 1−Aµ(uτj)/µ )) , u ̸= 0. It is known that ζ(0) = −1/2 and ζ ′(0) = − log(2π)/2 and for 0 < a < 1, ζ(0, a) + ζ(0, 1− a) = 0, ζ ′(0, a) + ζ ′(0, 1− a) = − log ∣∣e2π√−1a − 1 ∣∣. Using the above equations, we conclude for u = 0 κE(K,C, gθ,J)(0) = −(n+ 1) ( 1 + 2ζ(0) ) = 0, κE(K,C, gθ,J)′(0) = 2(n+ 1) log 4π µ , and for u ̸= 0 κE(K,Eu, gθ,J)(0) = 0, κE(K,Eu, gθ,J) ′(0) = 2 n+1∑ j=1 log ∣∣e2π√−1uτj/µ − 1 ∣∣ as claimed. 4 Ray–Singer torsion of the trivial bundle We compute the analytic torsion of the trivial bundle on lens spaces. We define the Ray–Singer torsion function associated with ( Ω•(M,E),d∇ ) by κdR(M,E, g)(s) := 2n+1∑ k=0 (−1)kkζ ( ∆k dR,g ) (s), where ζ ( ∆k dR,g ) (s) is the spectral zeta function of the k-th Hodge–de Rham Laplacian ∆k dR,g with respect to g. We define the Ray–Singer torsion TdR by 2 log TdR(M,E, g) := κdR(M,E, g)′(0). Following the derivation of [10, equation (3)], we have 2 log TdR(K,C, gstd) = 1 µ µ−1∑ ℓ=0 2n∑ j=0 (−1)j+1ζ ′ ( 0; j, γℓ ) , Ray–Singer Torsion and the Rumin Laplacian on Lens Spaces 13 where λm is the m-th eigenvalue of d†d and ζ ( s; j, γℓ ) = ∞∑ m=0 λ−s m Tr ( γℓ ∣∣ Xj,m ) , Xj,m := { ϕ ∈ Ωj(K) | d†dϕ = λmϕ } . We recall from [10, p. 123] 2n∑ j=0 (−1)j+1ζ ( s; j, γℓ ) = ( Γ(s) )−2 ∫ ∞ 0 t2s−1 ∫ 1 0 ( u(1− u) )s−1( f0(t, u) + f1(t, ℓν/µ) ) dudt+O ( s2 ) , where ν = (ν1, . . . , νn+1) and for σ = (σ1, . . . , σn+1) ∈ Rn+1, f0(t, u) = n+1∑ j=0 (−1)j ( n j )( 2 sinh tu sinh t(1− u) sinh t )j − e−(2n+1)tu 2 sinh tu − e−(2n+1)t(1−u) 2 sinh t(1− u) , f1(t, σ) = n+1∑ k ( 1− sinh t cosh t− cos 2πσk ) . We set h0(s) := ( Γ(s) )−2 ∫ ∞ 0 t2s−1 ∫ 1 0 ( u(1− u) )s−1 f0(t, u) dudt, h1 ( s, γℓ ) := ( Γ(s) )−2 ∫ ∞ 0 t2s−1 ∫ 1 0 ( u(1− u) )s−1 f1(t, ℓν/µ) dudt. From [10, p. 125], it is seen that h1 ( s, γℓ ) = −1 2 µ−1∑ j=0 Tr ( γjℓ )( ζ(2s, j/µ) + ζ(2s, 1− j/µ) ) − 2(n+ 1)µ−2sζ(2s), where Tr ( γℓ ) = n+1∑ j=1 ( e2π √ −1ℓνj/µ + e−2π √ −1ℓνj/µ ) . Taking the average of h1(s, γ ℓ), we have 1 µ µ−1∑ ℓ=0 h1 ( s, γℓ ) = −2(n+ 1)µ−2sζ(2s) Using ζ ′(0) = − log(2π)/2, we get 1 µ µ−1∑ ℓ=0 h′1 ( 0, γℓ ) = 2(n+ 1) log ( 2π µ ) . (4.1) We recall the Ray–Singer torsion on spheres, 14 A. Kitaoka Proposition 4.1 ([15]). TdR(S,C, gstd) = 2πn+1 n! Remark 4.2. The Ray–Singer torsion of spheres can easily be determined using the Cheeger– Müller theorem, a result which predates [15]. Details can be found in [3]. To put µ = 1, ν = (1, . . . , 1), from (4.1), it follows that h′0(0) = 2 log TdR(S,C, gstd)− 1 µ µ−1∑ ℓ=0 h′1 ( 0, γℓ ) = 2 log ( 2πn+1 n! ) − 2(n+ 1) log(2π) = 2 log ( 2−n n! ) . (4.2) By (4.1) and (4.2), we conclude 2 log TdR(K,C, gstd) = 1 µ µ−1∑ ℓ=0 h′0(0) + 1 µ µ−1∑ ℓ=0 h′1 ( 0, γℓ ) = 2 log ( 2−n n! ) + 2(n+ 1) log ( 2π µ ) = 2 log ( 2πn+1 n!µn+1 ) . (4.3) For the metric g on E over M , we set gρ := e2ρg for ρ ∈ R. Then, the Hodge–de Rham Laplacian ∆dR,gρ with respect to gρ is given by ∆dR,gρ = e−2ρ∆dR,g (4.4) (e.g., see [11, equation (5.4)]). From (4.4), we see κdR(M,E, gρ)(s) = 2n+1∑ k=0 (−1)kk dimHk dR(M,E) + e2ρs 2n+1∑ k=0 (−1)kk ( ζ ( ∆k dR,g ) (s)− dimHk dR(M,E) ) . To derivate the above equation, κdR(M,E, gρ) ′(s) = 2ρe2ρs 2n+1∑ k=0 (−1)kk ( ζ ( ∆k dR,g ) (s)− dimHk dR(M,E) ) + e2ρs 2n+1∑ k=0 (−1)kk ( ζ ( ∆k dR,g )′ (s) ) . To substitute s = 0, since ζ ( ∆k dR,g ) (0) = 0 on manifolds with dimension 2n+ 1, the Ray–Singer torsion is given by log TdR(M,E, gρ) = log TdR(M,E, g)− ρ 2n+1∑ k=0 (−1)kk dimHdR(M,E). (4.5) To substitute M = K, E is the trivial bundle, ρ = log 2 and g = gstd, we obtain TdR(K,C, 4gstd) = 22n+1TdR(K,C, gstd). (4.6) By (4.3) and (4.6), we conclude Proposition 1.2. Ray–Singer Torsion and the Rumin Laplacian on Lens Spaces 15 5 Proof of Corollary 1.3 Since α(γ) ∈ U(r) is diagonalizable by a unitary matrix, we have Eα = Eu1 ⊕ · · · ⊕ Eur . We recall the Ray–Singer torsion on lens spaces, Proposition 5.1 ([10]). For u (= 1, . . . , µ− 1), TdR(K,Eu, 4gstd) = n+1∏ j=1 ∣∣e2π√−1uτj/µ − 1 ∣∣. From Proposition 5.1, Theorem 1.1, and Proposition 1.2, we conclude TdR(K,Eα, gθ,J) = r∏ j=1 TdR(K,Euj , gθ,J) = r∏ j=1 n!− dimH0(K,Euj )TE(K,Euj , gθ,J) = n!− dimH0(K,Eα)TE(K,Eα, gθ,J). A Alternative shorter derivation of Proposition 1.2 The decomposition Ω•(S2n+1,C ) = µ−1⊕ u=0 Ω•(K,Eu) gives κdR ( S2n+1,C, 4gstd ) = µ−1∑ u=0 κdR(K,Eu, 4gstd). In particular, TdR ( S2n+1,C, 4gstd ) = TdR(K,C, 4gstd) µ−1∏ u=1 TdR(K,Eu, 4gstd). Combining this with (4.5), (4.6), Propositions 4.1 and 5.1 and [3], we obtain (4π)n+1 n! = TdR(K,C, 4gstd) µ−1∏ u=1 n+1∏ j=1 ∣∣e2π√−1uτj/µ − 1 ∣∣ = TdR(K,C, 4gstd)µn+1, whence Proposition 1.2. Acknowledgement The author is grateful to his supervisor Professor Kengo Hirachi for introducing this subject and for helpful comments. This work was supported by the program for Leading Graduate Schools, MEXT, Japan. The author also thanks the referees for their valuable comments. 16 A. Kitaoka References [1] Albin P., Quan H., Sub-Riemannian limit of the differential form heat kernels of contact manifolds, Int. Math. Res. Not. 2022 (2022), 5818–5881, arXiv:1912.02326. [2] Bismut J.M., Zhang W., An extension of a theorem by Cheeger and Müller (with an appendix by François Laudenbach), Astérisque 205 (1992), 235 pages. [3] de Melo T., Spreafico M., Reidemeister torsion and analytic torsion of spheres, J. Homotopy Relat. Struct. 4 (2009), 181–185, arXiv:0906.2570. [4] Fulton W., Young tableaux: with applications to representation theory and geometry, London Math. Soc. Stud. Texts, Vol. 35, Cambridge University Press, Cambridge, 1996. [5] Fulton W., Harris J., Representation theory. A first course, Grad. Texts in Math., Vol. 129, Springer-Verlag, New York, 1991. [6] Goodman R., Wallach N.R., Symmetry, representations, and invariants, Grad. Texts in Math., Vol. 255, Springer, Dordrecht, 2009. [7] Julg P., Kasparov G., Operator K-theory for the group SU(n, 1), J. Reine Angew. Math. 463 (1995), 99–152. [8] Kitaoka A., Analytic torsions associated with the Rumin complex on contact spheres, Internat. J. Math. 31 (2020), 2050112, 16 pages, arXiv:1911.03092. [9] Ponge R., Noncommutative residue for Heisenberg manifolds. Applications in CR and contact geometry, J. Funct. Anal. 252 (2007), 399–463, arXiv:math.DG/0607296. [10] Ray D.B., Reidemeister torsion and the Laplacian on lens spaces, Adv. Math. 4 (1970), 109–126. [11] Rosenberg S., The Laplacian on a Riemannian manifold: an introduction to analysis on manifolds, London Math. Soc. Stud. Texts, Vol. 31, Cambridge University Press, 1997. [12] Rumin M., Formes différentielles sur les variétés de contact, J. Differential Geom. 39 (1994), 281–330. [13] Rumin M., Sub-Riemannian limit of the differential form spectrum of contact manifolds, Geom. Funct. Anal. 10 (2000), 407–452. [14] Rumin M., Seshadri N., Analytic torsions on contact manifolds, Ann. Inst. Fourier (Grenoble) 62 (2012), 727–782, arXiv:0802.0123. [15] Weng L., You Y., Analytic torsions of spheres, Internat. J. Math. 7 (1996), 109–125. https://doi.org/10.1093/imrn/rnaa270 https://doi.org/10.1093/imrn/rnaa270 http://arxiv.org/abs/1912.02326 http://arxiv.org/abs/0906.2570 https://doi.org/10.1017/CBO9780511626241 https://doi.org/10.1007/978-1-4612-0979-9 https://doi.org/10.1007/978-0-387-79852-3 https://doi.org/10.1515/crll.1995.463.99 https://doi.org/10.1142/S0129167X20501128 http://arxiv.org/abs/1911.03092 https://doi.org/10.1016/j.jfa.2007.07.001 http://arxiv.org/abs/math.DG/0607296 https://doi.org/10.1016/0001-8708(70)90018-6 https://doi.org/10.1017/CBO9780511623783 https://doi.org/10.4310/jdg/1214454873 https://doi.org/10.1007/s000390050013 https://doi.org/10.5802/aif.2693 http://arxiv.org/abs/0802.0123 https://doi.org/10.1142/S0129167X96000074 1 Introduction 2 The Rumin complex 2.1 The Rumin complex on contact manifolds 2.2 The Rumin complex on the CR spheres 3 Contact analytic torsion of flat vector bundles 4 Ray–Singer torsion of the trivial bundle 5 Proof of Corollary 1.3 A Alternative shorter derivation of Proposition 1.2 References
id nasplib_isofts_kiev_ua-123456789-211813
institution Digital Library of Periodicals of National Academy of Sciences of Ukraine
issn 1815-0659
language English
last_indexed 2026-03-16T12:21:46Z
publishDate 2022
publisher Інститут математики НАН України
record_format dspace
spelling Kitaoka, Akira
2026-01-12T10:16:09Z
2022
Ray-Singer Torsion and the Rumin Laplacian on Lens Spaces. Akira Kitaoka. SIGMA 18 (2022), 091, 16 pages
1815-0659
2020 Mathematics Subject Classification: 58J52; 32V20; 53D10; 43A85
arXiv:2009.03276
https://nasplib.isofts.kiev.ua/handle/123456789/211813
https://doi.org/10.3842/SIGMA.2022.091
We express explicitly the analytic torsion functions associated with the Rumin complex on lens spaces in terms of the Hurwitz zeta function. In particular, we find that the functions vanish at the origin and determine the analytic torsions. Moreover, we have a formula between this torsion and the Ray-Singer torsion.
The author is grateful to his supervis or, Professor Kengo Hirachi, for introducing this subject and for helpful comments. This work was supported by the program for Leading Graduate Schools, MEXT, Japan. The author also thanks the referees for their valuable comments.
en
Інститут математики НАН України
Symmetry, Integrability and Geometry: Methods and Applications
Ray-Singer Torsion and the Rumin Laplacian on Lens Spaces
Article
published earlier
spellingShingle Ray-Singer Torsion and the Rumin Laplacian on Lens Spaces
Kitaoka, Akira
title Ray-Singer Torsion and the Rumin Laplacian on Lens Spaces
title_full Ray-Singer Torsion and the Rumin Laplacian on Lens Spaces
title_fullStr Ray-Singer Torsion and the Rumin Laplacian on Lens Spaces
title_full_unstemmed Ray-Singer Torsion and the Rumin Laplacian on Lens Spaces
title_short Ray-Singer Torsion and the Rumin Laplacian on Lens Spaces
title_sort ray-singer torsion and the rumin laplacian on lens spaces
url https://nasplib.isofts.kiev.ua/handle/123456789/211813
work_keys_str_mv AT kitaokaakira raysingertorsionandtheruminlaplacianonlensspaces