betaFeedback

WorksMathematics

Euler characteristic and flat cohomology of convex projective 44-manifolds

Abstract

Let M=Ω/ΓM=Ω/Γ be a closed properly convex real projective 44-manifold. We prove the Chern-Hopf-Thurston conjecture for this class of manifolds, i.e. χ(M)≥0χ(M)\geq0. The argument transfers Akagawa's double-form identity on the Cheng–Yau cone over ΩΩ to a Bochner identity on MM, from which we deduce that Hq(M;ΛpE∗)=0H^q(M;Λ^pE^*)=0 whenever q<p(n+1−p)/(n+1)q<p(n+1-p)/(n+1), in arbitrary dimension nn, for some flat bundle EE naturally attached to MM. In dimension four, this implies that the cohomology with coefficients in Λ2E∗Λ^2E^* concentrates in degree two.

Introduction

The Chern–Hopf–Thurston conjecture predicts that every closed aspherical manifold of dimension 2m2m satisfies

(−1)mχ(M)≥0.(1)(-1)^m\chi(M) \ge0. \tag*{(1)}

The conjecture has been established under some geometric, complex-analytic, and topological assumptions. For closed real hyperbolic manifolds, the predicted sign follows from the Gauss–Bonnet theorem, and more generally it holds for closed locally symmetric manifolds of noncompact type [15]. Gromov proved the strict inequality for closed Kähler hyperbolic manifolds [11], while Cao–Xavier and Jost–Zuo established the non-strict inequality for compact Kähler manifolds with nonpositive sectional curvature [3, 13]. In dimension four, the conjecture holds for closed Riemannian manifolds with nonpositive sectional curvature [6], closed aspherical complex surfaces [12], closed nonpositively curved piecewise Euclidean cubical manifolds [8], and closed generalized Haken manifolds [7]. For closed aspherical manifolds whose fundamental groups contain an infinite amenable normal subgroup, Cheeger–Gromov proved the stronger conclusion χ(M)=0\chi(M) = 0 [4]. More recently, Deng–Wang established the conjecture for closed aspherical complex projective manifolds whose fundamental groups admit a linear representation with finite kernel over a field of arbitrary characteristic [9].

In this paper we establish the conjecture for closed properly convex real projective 44-manifolds. A properly convex projective nn-manifold is a quotient M=Ω/ΓM = \Omega/\Gamma, where Ω⊂RPn\Omega\subset\mathbb{R}P^n is an open convex set bounded in an affine chart, and Γ\Gamma acts freely, properly discontinuously, and cocompactly by projective transformations. The domain Ω\Omega is contractible, so MM is aspherical. For a general domain, the Cheng–Yau construction supplies a canonical Hessian metric on its cone [5]. We use this metric to prove a vanishing result for flat cohomology with coefficients in exterior powers of the holonomy representation. For oriented four-manifolds, this yields the following:

Theorem A. Let MM be a closed properly convex real projective 44-manifold. Then χ(M)≥0\chi(M) \ge0. More precisely, if MM is oriented, let E→ME \to M be the flat rank-five bundle defined by the cone-preserving unimodular lift of its holonomy. Then

10χ(M)=dim⁡RH2(M;Λ2E∗).(2)10\chi(M) = \dim_{\mathbb{R}} H^{2}(M;\Lambda^{2}E^{*}). \tag*{(2)}

The holonomy lift and its flat bundle are constructed in § 2. No strict convexity, boundary regularity, or irreducibility hypothesis is imposed, and the proof makes no assumption on the sectional curvature of a metric on MM. The identity (2) explains the sign by realizing a positive multiple of the Euler characteristic as the dimension of a cohomology group.

The result also completes the determination of the possible Euler characteristics in the oriented four-dimensional case. Kobayashi, using an earlier Pontryagin-form identity of Avez, proved the rational vanishing of the Pontryagin classes of real projective manifolds [14 Theorem 4.16]. In dimension four, this forces the signature to vanish and the Euler characteristic to be even. These restrictions alone leave its sign undetermined. Riolo–Seppi–Slavich construct closed oriented properly convex projective four-manifolds with every positive even Euler characteristic [16 Theorem 1.3]. Together with the simplex quotients, for which χ=0\chi= 0, Theorem A shows that the possible values are precisely the even non-negative integers.

Cohomology with flat coefficients. The vanishing theorem underlying Theorem A holds in arbitrary dimension. Let MM be oriented of dimension nn, put N=n+1N = n + 1, and let EE be the standard flat rank-NN bundle defined by its unimodular holonomy lift. For 0≤p≤N0 \le p \le N, set

Fp=ΛpE∗.F_p = \Lambda^p E^{*}.

The flat connection defines an exterior differential dpd_p on Ω∙(M;Fp)\Omega^\bullet(M; F_p), and its cohomology is denoted by H∙(M;Fp)H^\bullet(M; F_p).

Theorem B. Let MM be a closed oriented properly convex real projective nn-manifold, n≥2n \ge2, and let EE be its standard unimodular flat bundle. For 0≤p≤n+10 \le p \le n+1 and 0≤q≤n0 \le q \le n,

Hq(M;ΛpE∗)=0ifq<p(n+1−p)n+1.(3)H^q(M;\Lambda^p E^{*}) = 0 \quad\text{if} \quad q < \frac{p(n+1-p)}{n+1}. \tag*{(3)}

The theorem follows from a Bochner identity on the compact manifold MM. The Cheng–Yau metric on the cone determines a Riemannian metric kk on MM and a metric on FpF_p. Let dp∗d_p^* be the resulting formal adjoint and let Δp,q=dpdp∗+dp∗dp\Delta_{p,q} = d_p d_p^* + d_p^*d_p. We prove the quantitative estimate

∥dpα∥2+∥dp∗α∥2≥(p(N−p)N−q)∥α∥2,α∈Ωq(M;Fp).(4)\lVert d_p \alpha\rVert^2 + \lVert d_p^*\alpha\rVert^2 \ge\left(\frac{p(N-p)}{N} - q\right)\lVert\alpha\rVert^2,\qquad\alpha\in\Omega^q(M;F_p). \tag*{(4)}

More precisely, we construct first-order operators Ap,qA_{p,q}, Bp,qB_{p,q} on bundles over MM such that

Δp,q=Ap,q∗Ap,q+Bp,q∗Bp,q+(p−q−p2N)Id⁡.(5)\Delta_{p,q} = A_{p,q}^{*}A_{p,q} + B_{p,q}^{*}B_{p,q} + \left(p-q-\frac{p^2}{N}\right)\operatorname{Id}. \tag*{(5)}

The bundles, metrics, and operators are defined in § 4.3. Taking the inner product with α\alpha gives (4). The flat connection need not preserve the fiber metric, but the corresponding elliptic Hodge theory still represents flat cohomology by harmonic forms; we include the standard argument in Lemma 3.6.

Theorem A is a consequence of (3). For N=5N = 5, and p=2,3p = 2, 3 we have

p(5−p)5=65>1.\frac{p(5-p)}{5} = \frac{6}{5} > 1.

Theorem B therefore eliminates degrees zero and one cohomology with coefficients in F2F_2 and F3F_3. A parallel volume form on EE identifies F2∗F_2^* with F3F_3. Poincaré duality then eliminates degrees three and four for F2F_2. Its cohomology is thus concentrated in degree two, and the Euler characteristic with rank-ten flat coefficients yields (2). In higher even dimensions, the vanishing range and its dual can leave other degrees besides the middle one, so the theorem does not by itself establish the sign conjecture there.

Organization of the paper. § 2 constructs the unimodular holonomy bundle and the radial product. § 3 defines the double-form operators and derives their identity in the cone over Ω\Omega following [1]. § 4 transfers it to the aforementioned Bochner identity on MM and proves Theorem B. § 5 gives the topological argument for Theorem A. Appendix § 6 gives a direct proof in coordinates of the identities in [1].

AI declaration. Chat GPT Sol 6.1 was used for literature review, proofreading and editing. All mathematical ideas are original of the author.

Acknowledgment. The author acknowledges support from the Italian Ministry of University and Research (MUR) through the FIS3 project “Geometry and Topology of higher Teichmüller spaces”, grant no. FIS-2024-05142.

The Cheng–Yau cone and the holonomy bundle

We work in dimension n≥2n \ge2 and write N=n+1N = n + 1. Unless explicitly stated otherwise, M=Ω/ΓM = \Omega/\Gamma is oriented. We begin by fixing the lift of the holonomy.

The unimodular lift

Choose an affine chart in which Ω\Omega is bounded and convex. After identifying this chart with {x0=1}⊂RN\{x^0 = 1\} \subset\mathbb{R}^N, let

C={t(1,z):t>0, z∈Ω}⊂V=RN.C = \{t(1,z) : t > 0,\ z \in\Omega\} \subset V = \mathbb{R}^N.

This is an open convex cone. Its radial projection identifies C/R>0C/\mathbb{R}_{>0} with Ω\Omega.

Lemma 1.1. The holonomy of an oriented properly convex projective manifold has a unique lift Γ→SL(V)\Gamma\to\mathrm{SL}(V) that preserves CC.

Proof. By [2 Section 2.1], the holonomy has a unique lift to

SL±(V)={A∈GL(V):∣det⁡A∣=1}.\mathrm{SL}^{\pm}(V) = \{A \in\mathrm{GL}(V) : \lvert\det A \rvert= 1\}.

whose image preserves the cone CC. We check that this lift takes values in SL(V)\mathrm{SL}(V). The radial projection C→ΩC \to\Omega has naturally oriented fibers: each fiber is a positive ray, oriented by increasing radial parameter. Every cone-preserving linear map preserves these fiber orientations. Moreover, the holonomy acts on Ω\Omega by orientation-preserving maps, because M=Ω/ΓM = \Omega/\Gamma is oriented. The lifted maps therefore preserve the induced orientation of CC, so their determinants are positive. Uniqueness follows from the uniqueness of the cone-preserving lift to SL±(V)\mathrm{SL}^{\pm}(V). □\square

We use the same symbol Γ\Gamma for the lifted subgroup of SL(V)\mathrm{SL}(V). Define

E=(Ω×V)/Γ,γ⋅(z,v)=(γz,γv).(6)E = (\Omega\times V)/\Gamma,\qquad\gamma\cdot(z,v) = (\gamma z,\gamma v). \tag*{(6)}

Local constant sections on Ω\Omega define its flat connection. The induced bundle Fp=ΛpE∗F_p = \Lambda^p E^* has monodromy Λp(γ−T)\Lambda^p(\gamma^{-T}). Since Γ⊂SL(V)\Gamma\subset\mathrm{SL}(V), any fixed volume form on VV defines a parallel volume form on EE.

The canonical potential

Let DD denote the standard flat affine connection on CC, and choose a constant volume form on VV. We use the following normalization of the Cheng–Yau canonical potential.

Theorem 1.2 (Cheng–Yau). There is a unique smooth strictly convex function ϕ\phi on CC satisfying

det⁡D2ϕ=e2ϕ(7)\det D^2\phi= e^{2\phi} \tag*{(7)}

such that the Hessian metric g=Ddϕg = Dd\phi is complete.

This is the canonical potential theorem of Cheng–Yau [5]; see also [10 Theorem 1.3 and Corollary 1.1] for a precise formulation and the uniqueness statement. Completeness is used here to characterize the potential; the compact Bochner estimate below uses only local identities. The determinant in (7) is computed in affine coordinates of unit volume for the chosen constant volume form.

Lemma 1.3. If A∈GL⁡(V)A \in\operatorname{GL}(V) preserves CC, then

ϕ(Ax)=ϕ(x)−log⁡∣det⁡A∣.(8)\phi(Ax) = \phi(x) - \log\lvert\det A\rvert. \tag*{(8)}

In particular,

ϕ(tx)=ϕ(x)−Nlog⁡t(t>0),ϕ(γx)=ϕ(x)(γ∈Γ).(9)\phi(tx) = \phi(x) - N\log t \quad(t > 0), \qquad\phi(\gamma x) = \phi(x) \quad(\gamma\in\Gamma). \tag*{(9)}

The transformations AA and the radial dilations preserve gg.

Proof. Set ψ(x)=ϕ(Ax)+log⁡∣det⁡A∣\psi(x) = \phi(Ax) + \log\lvert\det A\rvert. The chain rule gives

D2ψ(x)=AT(D2ϕ)(Ax)A.D^2\psi(x) = A^{\mathsf T}(D^2\phi)(Ax)A.

It follows that

det⁡D2ψ(x)=∣det⁡A∣2e2ϕ(Ax)=e2ψ(x).\det D^2\psi(x) = \lvert\det A\rvert^2 e^{2\phi(Ax)} = e^{2\psi(x)}.

The function ψ\psi is strictly convex, and its Hessian metric is A∗gA^*g, which is complete because AA is a diffeomorphism of CC. Uniqueness in Theorem 1.2 gives ψ=ϕ\psi= \phi, which proves (8). Taking A=tId⁡A = t\operatorname{Id} gives the first identity in (9), and taking A=γ∈ΓA = \gamma\in\Gamma gives the second. Applying DdDd to (8) proves A∗g=gA^*g = g. □

The radial product

Let HH be the vector field, defined by Hx=xH_x = x in the affine identification TxC=VT_xC = V. Set

Σ=ϕ−1(0),k=g∣TΣ.\Sigma= \phi^{-1}(0), \qquad k = g|_{T\Sigma}.

Every radial ray meets Σ\Sigma once. Since ϕ\phi is Γ\Gamma-invariant, the radial projection induces a diffeomorphism Σ/Γ≅M\Sigma/\Gamma\cong M. We use kk also for the quotient metric.

Proposition 1.4. The map

Φ:R×Σ⟶C,Φ(r,x)=erx,\Phi: \mathbb{R} \times\Sigma\longrightarrow C, \qquad\Phi(r,x) = e^r x,

is a Γ\Gamma-equivariant diffeomorphism, and

Φ∗g=N dr2+k.(10)\Phi^*g = N\,dr^2 + k. \tag*{(10)}

In particular, X=C/ΓX = C/\Gamma identifies with R×M\mathbb{R} \times M with this product metric. In these coordinates,

H=∂r,ϕ=−Nr,∣dr∣g2=N−1,dvol⁡g=N dr dvol⁡k.(11)H = \partial_r, \qquad\phi= -Nr, \qquad\lvert dr\rvert_g^2 = N^{-1}, \qquad d\operatorname{vol}_g = \sqrt{N}\,dr\,d\operatorname{vol}_k. \tag*{(11)}

Proof. In affine coordinates H=∑ixi∂iH = \sum_i x^i\partial_i. Differentiating ϕ(tx)=ϕ(x)−Nlog⁡t\phi(tx) = \phi(x) - N\log t in tt at t=1t = 1 gives

∑ixiϕi=−N.\sum_i x^i\phi_i = -N.

Differentiating this equation in xjx^j then gives

ϕj+∑ixiϕij=0.\phi_j+\sum_i x^i\phi_{ij}=0.

Since gij=ϕijg_{ij}=\phi_{ij}, this is the coordinate form of

g(H,Y)=−dϕ(Y),grad⁡gϕ=−H.(12)g(H,Y)=-\mathrm{d}\phi(Y),\qquad\operatorname{grad}_g\phi=-H. \tag*{(12)}

In particular, g(H,H)=−dϕ(H)=Ng(H,H)=-\mathrm{d}\phi(H)=N, and HH is orthogonal to every level set of ϕ\phi. The inverse of Φ\Phi is

y⟼(−ϕ(y)N,eϕ(y)/Ny).y\longmapsto\left(-\frac{\phi(y)}{N},\frac{e^{\phi(y)/N}}{}y\right).

This proves that Φ\Phi is a diffeomorphism. To compute its metric in coordinates, write a local parametrization of Σ\Sigma as σ(z)\sigma(z) and put σa=∂aσ\sigma_a=\partial_a\sigma. Then

dΦ(∂r)=erσ,dΦ(∂a)=erσa.\mathrm{d}\Phi(\partial_r)=e^r\sigma,\qquad\mathrm{d}\Phi(\partial_a)=e^r\sigma_a.

Radial dilation preserves gg by Lemma 1.3. Equivalently, for fixed ambient vectors v,wv,w, gerx(v,w)=e−2rgx(v,w)g_{e^r x}(v,w)=e^{-2r}g_x(v,w). Therefore the product metric coefficients are

gerσ(erσ,erσ)=N,g_{e^r\sigma}(e^r\sigma,e^r\sigma)=N,
gerσ(erσ,erσa)=0,g_{e^r\sigma}(e^r\sigma,e^r\sigma_a)=0,
gerσ(erσa,erσb)=kab.g_{e^r\sigma}(e^r\sigma_a,e^r\sigma_b)=k_{ab}.

These identities prove (10). Moreover, r=−ϕ/Nr=-\phi/N, and the displayed inverse shows that Γ\Gamma acts trivially on rr and by its original action on Σ\Sigma. Thus the product is equivariant and descends to X=R×MX=\mathbb{R}\times M. Finally, the inverse metric and determinant of the block matrix diag⁡(N,kab)\operatorname{diag}(N,k_{ab}) give ∣dr∣2=N−1\lvert\mathrm{d}r\rvert^2=N^{-1} and dvol⁡g=N dr dvol⁡k\mathrm{d}\operatorname{vol}_g=\sqrt{N}\,\mathrm{d}r\,\mathrm{d}\operatorname{vol}_k, while dΦ(∂r)=H\mathrm{d}\Phi(\partial_r)=H gives the remaining radial formulas. □

Double forms and Akagawa’s identity

The cone has two structures that will be used separately: its flat affine connection and its Hessian metric. The flat connection defines the two differentials below, while the metric defines their adjoints.

Double forms, exterior operators, and inner products

Let (Y,D)(Y,D) be a flat affine manifold of dimension NN. A double form of bidegree (p,q)(p,q) is a smooth section of

ΛpT∗Y⊗ΛqT∗Y.\Lambda^pT^*Y\otimes\Lambda^qT^*Y.

We denote the space of these sections by Ap,q(Y)\mathcal{A}^{p,q}(Y). In affine coordinates x1,…,xNx^1,\ldots,x^N, write

ω=1p!q!ωi1…ip ; j1…jq dxi1∧⋯∧dxip⊗dxj1∧⋯∧dxjq.\omega=\frac{1}{p!q!}\omega_{i_1\ldots i_p\,;\,j_1\ldots j_q}\,\mathrm{d}x^{i_1}\wedge\cdots\wedge\mathrm{d}x^{i_p}\otimes\mathrm{d}x^{j_1}\wedge\cdots\wedge\mathrm{d}x^{j_q}.

Here and below repeated indices are summed from 11 to NN unless otherwise indicated. The components are alternating separately in the indices on the two sides of the semicolon; no symmetry between the two groups is imposed.

Exterior multiplication takes place independently in the two factors. In particular,

(ξ⊗η)∧(ξ′⊗η′)=(ξ∧ξ′)⊗(η∧η′)(\xi\otimes\eta)\wedge(\xi'\otimes\eta')=(\xi\wedge\xi')\otimes(\eta\wedge\eta')

for decomposable double forms. There is no sign from passing a form in the first factor past a form in the second factor. This is the convention of [1 Definition 2.1.2].

For a covector ξ\xi, let ε(ξ)\varepsilon(\xi) and εˉ(ξ)\bar{\varepsilon}(\xi) denote wedge multiplication in the first and second factors. For a vector ZZ, let ι(Z)\iota(Z) and ιˉ(Z)\bar{\iota}(Z) denote contraction in the respective factors. Our contraction convention is

(ι(Z)ω)i2…ip ;J=Zaωai2…ip ;J,(ιˉ(Z)ω)I ;j2…jq=ZaωI ;aj2…jq.(\iota(Z)\omega)_{i_2\ldots i_p\,;J}=Z^a\omega_{ai_2\ldots i_p\,;J},\qquad (\bar{\iota}(Z)\omega)_{I\,;j_2\ldots j_q}=Z^a\omega_{I\,;aj_2\ldots j_q}.

Thus contraction inserts the vector in the first slot of the chosen factor.

Write ei=ε(dxi)e_i=\varepsilon(dx^i), eˉi=εˉ(dxi)\bar e_i=\bar\varepsilon(dx^i), ii=ι(∂i)i_i=\iota(\partial_i), and iˉi=ιˉ(∂i)\bar i_i=\bar\iota(\partial_i). In each factor the usual exterior-algebra relations hold:

iiej+ejii=δijId⁡,iˉieˉj+eˉjiˉi=δijId⁡.i_i e_j+e_j i_i=\delta_{ij}\operatorname{Id},\qquad\bar i_i\bar e_j+\bar e_j\bar i_i=\delta_{ij}\operatorname{Id}.

Two wedge operators, or two contraction operators, anticommute within the same factor. All operators acting on different factors commute. Put Di=D∂iD_i=D_{\partial_i}. In an affine coordinate frame, DiD_i differentiates the components of the entire double form ordinarily. Define

∂=∑ieiDi,∂ˉ=∑ieˉiDi.(13)\partial=\sum_i e_iD_i,\qquad\bar\partial=\sum_i\bar e_iD_i. \tag*{(13)}

For example,

(∂ω)i0…ip;J=∑r=0p(−1)r∂irωi0…ir^…ip;J.(\partial\omega)_{i_0\ldots i_p;J}=\sum_{r=0}^{p}(-1)^r\partial_{i_r}\omega_{i_0\ldots\widehat{i_r}\ldots i_p;J}.

and ∂ˉ\bar\partial has the identical formula in the second group of indices. These definitions are independent of the affine chart. Since affine derivatives commute and the two exterior factors commute, they give

∂2=∂ˉ2=0,∂∂ˉ=∂ˉ∂.(14)\partial^2=\bar\partial^2=0,\qquad\partial\bar\partial=\bar\partial\partial. \tag*{(14)}

The last equality has no minus sign. In particular, the two factors are not being combined into a single exterior algebra with the usual total-degree sign convention.

For fixed pp, the first exterior factor is a flat coefficient bundle. More precisely,

(Acp,∙(Y),∂ˉ)=(Ωc∙(Y;Λp(T∗Y,D)),dΛp(T∗Y,D)).(15)(A_{\mathrm{c}}^{p,\bullet}(Y),\bar\partial)=(\Omega_{\mathrm{c}}^\bullet(Y;\Lambda^p(T^*Y,D)),d_{\Lambda^p(T^*Y,D)}). \tag*{(15)}

To check this description, choose any local parallel frame sAs_A of the first exterior factor and write ω=∑AsA⊗ωA\omega=\sum_A s_A\otimes\omega^A. Then

∂ˉω=∑AsA⊗dωA.\bar\partial\omega=\sum_A s_A\otimes d\omega^A.

The differential on the right requires only the connection on the coefficient bundle; the ωA\omega^A are ordinary differential forms.

Suppose now that g=Ddϕg=Dd\phi is positive definite, and write gij=ϕijg_{ij}=\phi_{ij} and (gij)=(gij)−1(g^{ij})=(g_{ij})^{-1}. The metric on T∗YT^*Y is g−1g^{-1}. We use its determinant metric on exterior powers and the tensor-product metric on double forms. Explicitly,

⟨ω,η⟩g=1p! q!ωI;JηK;L(∏a=1pgiaka)(∏b=1qgjblb).\langle\omega,\eta\rangle_g=\frac{1}{p!\,q!}\omega_{I;J}\eta_{K;L}\left(\prod_{a=1}^{p}g^{i_a k_a}\right)\left(\prod_{b=1}^{q}g^{j_b l_b}\right).

For example, on a double (1,1)(1,1)-form this is gikgjℓωi;jηk;ℓg^{ik}g^{j\ell}\omega_{i;j}\eta_{k;\ell}. The L2L^2 inner product is defined with the Riemannian density:

(ω,η)Y=∫Y⟨ω,η⟩gdet⁡(gij) dx1⋯dxN.(\omega,\eta)_Y=\int_Y\langle\omega,\eta\rangle_g\sqrt{\det(g_{ij})}\,dx^1\cdots dx^N.

The notation Acp,q(Y)A_{\mathrm{c}}^{p,q}(Y) means compact support.

Let δ\delta and δˉ\bar\delta be the formal adjoints of ∂\partial and ∂ˉ\bar\partial for these inner products. They are determined by integration by parts on compactly supported smooth test forms. Define the local differential operators

□=∂δ+δ∂,□ˉ=∂ˉδˉ+δˉ∂ˉ.\Box=\partial\delta+\delta\partial,\qquad\bar\Box=\bar\partial\bar\delta+\bar\delta\bar\partial.

The Hessian identities

Regard gg as the double (1,1)(1,1)-form gij dxi⊗dxjg_{ij}\,dx^i\otimes dx^j, and let LL be wedge multiplication by this double form. Its pointwise adjoint is

Λ=gijiiiˉj.\Lambda=g^{ij}i_i\bar{i}_j.

Indeed, ei∗=giaiae_i^*=g^{ia}i_a and eˉj∗=gjbiˉb\bar{e}_j^*=g^{jb}\bar{i}_b, so L∗=gijgiagjbiaiˉb=gabiaiˉbL^*=g_{ij}g^{ia}g^{jb}i_a\bar{i}_b=g^{ab}i_a\bar{i}_b. In components,

(Λω)I;J=gabωaI;bJ.(\Lambda\omega)_{I;J}=g^{ab}\omega_{aI;bJ}.

Thus Λ\Lambda traces one covariant index from each factor. For a (1,1)(1,1)-form, Λω=gijωi;j\Lambda\omega=g^{ij}\omega_{i;j}.

The first Koszul form and its metric dual are

α=12dlog⁡det⁡(gij),Xα=α♯g.\alpha=\frac{1}{2}d\log\det(g_{ij}),\qquad X_\alpha=\alpha^{\sharp_g}.

An affine coordinate change multiplies det⁡(gij)\det(g_{ij}) by a positive constant, so its logarithmic differential is well defined. The local Hessian Kähler identities are

Λ∂+∂Λ=−δˉ+ıˉ(Xα),Λ∂ˉ+∂ˉΛ=−δ+ı(Xα).(16)\Lambda\partial+\partial\Lambda=-\bar{\delta}+\bar{\imath}(X_\alpha),\qquad\Lambda\bar{\partial}+\bar{\partial}\Lambda=-\delta+\imath(X_\alpha). \tag*{(16)}

These are [1 Theorem 2.3.6]. In Appendix § 6, we prove (16) directly in coordinates. For the Cheng–Yau potential, its Monge–Ampère equation gives 12dlog⁡det⁡(D2ϕ)=dϕ\frac{1}{2}d\log\det(D^2\phi)=d\phi. The radial identity (12) then gives

α=dϕ,Xα=−H.(17)\alpha=d\phi,\qquad X_\alpha=-H. \tag*{(17)}

The following calculation explains how the Euler field introduces the exterior degrees.

Lemma 2.1. On double forms of bidegree (p,q)(p,q) on CC,

∂ı(H)+ı(H)∂=DH+pId⁡,∂ˉıˉ(H)+ıˉ(H)∂ˉ=DH+qId⁡.(18)\partial\imath(H)+\imath(H)\partial=D_H+p\operatorname{Id},\qquad\bar{\partial}\bar{\imath}(H)+\bar{\imath}(H)\bar{\partial}=D_H+q\operatorname{Id}. \tag*{(18)}

Proof. Since DiH=∂iD_iH=\partial_i, differentiation of contraction gives

Di(ı(H)ω)=ıiω+ı(H)Diω.D_i(\imath(H)\omega)=\imath_i\omega+\imath(H)D_i\omega.

Therefore

(∂ı(H)+ı(H)∂)ω=∑ieiıiω+∑i(eiı(H)+ı(H)ei)Diω=∑ieiıiω+DHω.\begin{aligned} (\partial\imath(H)+\imath(H)\partial)\omega=\sum_i e_i\imath_i\omega+\sum_i(e_i\imath(H)+\imath(H)e_i)D_i\omega\\ &=\sum_i e_i\imath_i\omega+D_H\omega. \end{aligned}

The second equality uses eiı(H)+ı(H)ei=xiId⁡e_i\imath(H)+\imath(H)e_i=x^i\operatorname{Id}. On an elementary exterior monomial of first degree pp, the operator ∑ieiıi\sum_i e_i\imath_i counts its pp indices: a term vanishes unless dxidx^i occurs, and contraction followed by reinsertion gives that monomial for each such index. Thus ∑ieiıi=pId⁡\sum_i e_i\imath_i=p\operatorname{Id}. The identical calculation in the second factor gives ∑ieˉiıˉi=qId⁡\sum_i\bar{e}_i\bar{\imath}_i=q\operatorname{Id} and proves the second identity. These formulas also appear in [1 Corollary 3.3.3]. □

The Lie derivative in the HH-direction of the entire covariant tensor is

(LHω)I;J=Ha∂aωI;J+(p+q)ωI;J,(\mathcal{L}_H\omega)_{I;J}=H^a\partial_a\omega_{I;J}+(p+q)\omega_{I;J},

because ∂iHa=δia\partial_iH^a=\delta_i^a contributes once for every covariant slot. Hence LHω=DHω+(p+q)ω\mathcal{L}_H\omega=D_H\omega+(p+q)\omega. In particular, dilation invariance means DHω=−(p+q)ωD_H\omega=-(p+q)\omega.

Proposition 2.2 (Akagawa’s cone identity). On double forms of bidegree (p,q)(p,q) on the Cheng–Yau cone,

□ˉ=□+(p−q)Id⁡.(19)\bar{\Box}=\Box+(p-q)\operatorname{Id}. \tag*{(19)}

Proof. This is the local identity established in [1 proof of Main Theorem 3, p. 86]. Put I=ı(Xα)I=\imath(X_\alpha) and Iˉ=ıˉ(Xα)\bar{I}=\bar{\imath}(X_\alpha). Solving (16) for the adjoints gives

δ=−Λ∂ˉ−∂ˉΛ+I,δˉ=−Λ∂−∂Λ+Iˉ.\delta=-\Lambda\bar{\partial}-\bar{\partial}\Lambda+I,\qquad\bar{\delta}=-\Lambda\partial-\partial\Lambda+\bar{I}.

Substitution into the two Laplacians yields

□=−∂Λ∂ˉ−∂∂ˉΛ−Λ∂ˉ∂−∂ˉΛ∂+∂I+I∂,\Box= -\partial\Lambda\bar\partial- \partial\bar\partial\Lambda- \Lambda\bar\partial\partial- \bar\partial\Lambda\partial+ \partial I + I\partial,
□‾=−∂ˉΛ∂−∂ˉ∂Λ−Λ∂∂ˉ−∂Λ∂ˉ+∂ˉIˉ+Iˉ∂ˉ.\overline{\Box} = -\bar\partial\Lambda\partial- \bar\partial\partial\Lambda- \Lambda\partial\bar\partial- \partial\Lambda\bar\partial+ \bar\partial\bar I + \bar I\bar\partial.

The four terms involving Λ\Lambda agree between the two lines: the only interchanges needed are ∂ˉ∂ˉ=∂ˉ∂ˉ\bar\partial\bar\partial= \bar\partial\bar\partial, with Λ\Lambda left in its original position. Their difference is therefore

□−□‾=−(∂I+I∂)+(∂ˉIˉ+Iˉ∂ˉ).\Box-\overline{\Box}=-(\partial I+I\partial)+(\bar\partial\bar I+\bar I\bar\partial).

By (17), I=−ι(H)I=-\iota(H) and Iˉ=−ιˉ(H)\bar I=-\bar\iota(H). Applying Lemma 2.1 gives

□−□‾=(DH+p)−(DH+q)=(p−q)Id⁡.\Box-\overline{\Box}=(D_H+p)-(D_H+q)=(p-q)\operatorname{Id}.

This proves the asserted identity. □

Proposition 2.3. The four operators ∂,∂ˉ,δ,δˉ\partial,\bar\partial,\delta,\bar\delta descend to X=C/ΓX=C/\Gamma, and (19) holds on XX.

Proof. Each element of Γ\Gamma is affine for DD and preserves gg. Its pullback therefore commutes with both differentials and preserves the pointwise tensor metrics and the Riemannian density. It also commutes with their formal adjoints: the defining integration-by-parts identity on a coordinate neighborhood is unchanged by this pullback. □

The next section rewrites (19) as an identity of operators on MM.

A Bochner identity

The aim of this section is to express the two sides of Akagawa’s cone identity as operators on bundles over MM. In particular, we define the operators in the Bochner identity (5) and compute the action of □\Box and □‾\overline{\Box} on dilation invariant forms.

The coefficient bundle and its metric

The identification TC=C×VTC=C\times V identifies its flat cotangent bundle with C×V∗C\times V^*. The action on the fiber is γ−T\gamma^{-T}. Its ppth exterior power therefore descends to a flat bundle FpF_p on X=C/ΓX=C/\Gamma. Radial projection gives a flat bundle isomorphism

Fp≅πM∗Fp,πM:R×M⟶M.(20)F_p\cong\pi_M^*F_p,\qquad\pi_M:\mathbb{R}\times M\longrightarrow M. \tag*{(20)}

To describe it, write a point of the cone as erσ(y)e^r\sigma(y), where σ(y)∈Σ\sigma(y)\in\Sigma. A covector with affine components λi\lambda_i at σ(y)\sigma(y) is identified with the covector having the same components λi\lambda_i at erσ(y)e^r\sigma(y). This is flat parallel transport in the ambient space VV. It is Γ\Gamma-equivariant, so the description gives the stated isomorphism on the quotient. In particular, the connection in the radial direction is ordinary differentiation.

Under (20), ∂ˉ\bar\partial is the exterior covariant derivative of πM∗Fp\pi_M^*F_p, as explained in (15). In a local flat coefficient frame sAs_A, it is characterized by

∂ˉ(∑AsA⊗ωA)=∑AsA⊗dωA.\bar\partial\left(\sum_A s_A\otimes\omega^A\right)=\sum_A s_A\otimes\mathrm{d}\omega^A.

where ωA\omega^A are ordinary differential forms on R×M\mathbb{R}\times M.

The inverse Cheng–Yau metric g−1g^{-1} induces a fiber metric on Fp=ΛpT∗XF_p=\Lambda^pT^*X. Under the identification (20), the fiber of FpF_p at (0,y)(0,y) is identified with (Fp)y(F_p)_y. We denote by Hp,yH_{p,y} the inner product transported to (Fp)y(F_p)_y through this identification. This defines a fiber metric HpH_p on FpF_p.

Lemma 3.1. Under (20), the fiber metric on FpF_p over (r,y)(r,y) is

e2prHp(y).(21)e^{2pr}H_p(y). \tag*{(21)}

Proof. Fix t>0t > 0 and differentiate ϕ(tx)=ϕ(x)−Nlog⁡t\phi(tx) = \phi(x) - N \log t in the affine variables xx. The first and second derivatives give

tϕi(tx)=ϕi(x),t2ϕij(tx)=ϕij(x).t\phi_i(tx) = \phi_i(x), \qquad t^2\phi_{ij}(tx) = \phi_{ij}(x).

Consequently, for the matrices of the metric and its inverse,

gij(erx)=e−2rgij(x),gij(erx)=e2rgij(x).(22)g_{ij}(e^r x) = e^{-2r}g_{ij}(x), \qquad g^{ij}(e^r x) = e^{2r}g^{ij}(x). \tag*{(22)}

For covectors λ,μ\lambda,\mu we then have

⟨λ,μ⟩erx=∑i,jgij(erx)λiμj=e2r⟨λ,μ⟩x.\langle\lambda,\mu\rangle_{e^r x} = \sum_{i,j} g^{ij}(e^r x)\lambda_i\mu_j = e^{2r}\langle\lambda,\mu\rangle_x.

The exterior metric is the determinant metric:

⟨λ1∧⋯∧λp,μ1∧⋯∧μp⟩=det⁡(⟨λa,μb⟩)a,b=1p.\langle\lambda_1 \wedge\cdots\wedge\lambda_p,\mu_1 \wedge\cdots\wedge\mu_p\rangle= \det\left(\langle\lambda_a,\mu_b\rangle\right)_{a,b=1}^{p}.

Each entry of this p×pp \times p matrix acquires the factor e2re^{2r}, and its determinant acquires e2pre^{2pr}. Decomposable exterior tensors span the fiber, so bilinearity proves (21). □

We regard double (p,q)(p,q)-forms on XX as ordinary qq-forms on R×M\mathbb{R} \times M with coefficients in πM∗Fp\pi_M^*\mathcal{F}_p:

ΛpT∗X⊗ΛqT∗X≅πM∗Fp⊗ΛqT∗(R×M).\Lambda^p T^*X \otimes\Lambda^q T^*X \cong\pi_M^*\mathcal{F}_p \otimes\Lambda^q T^*(\mathbb{R} \times M).

Their pointwise metric is therefore the tensor product of the metric in Lemma 3.1 with the metric on ordinary qq-forms induced by the product Riemannian metric N dr2+kN\,dr^2+k. In particular, for s∈(Fp)ys \in(\mathcal{F}_p)_y and α∈ΛqT(r,y)∗(R×M)\alpha\in\Lambda^q T^*_{(r,y)}(\mathbb{R} \times M),

∣s⊗α∣g2=e2pr∣s∣Hp2∣α∣N dr2+k2.\lvert s \otimes\alpha\rvert_g^2 = e^{2pr}\lvert s\rvert_{H_p}^2\lvert\alpha\rvert_{N\,dr^2+k}^2.

On Ωq(M;Fp)\Omega^q(M;\mathcal{F}_p), we use the pointwise metric Hp⊗Λqk−1H_p \otimes\Lambda^q k^{-1} and the volume density dvol⁡k\operatorname{dvol}_k. Let dpd_p be the flat exterior derivative, let dp∗d_p^* be its formal adjoint, and put

Qp,q(a)=∥dpa∥M2+∥dp∗a∥M2,Δp,q=dpdp∗+dp∗dp.(23)Q_{p,q}(a) = \lVert d_p a\rVert_M^2 + \lVert d_p^*a\rVert_M^2, \qquad\Delta_{p,q} = d_p d_p^* + d_p^*d_p. \tag*{(23)}

Thus Qp,q(a)=(Δp,qa,a)MQ_{p,q}(a) = (\Delta_{p,q}a,a)_M.

The differential and its formal adjoint on the product

Every Fp\mathcal{F}_p-valued qq-form on XX has a unique decomposition

ω=a(r)+dr∧b(r),\omega= a(r) + dr \wedge b(r),

where a(r)∈Ωq(M;Fp)a(r) \in\Omega^q(M;\mathcal{F}_p) and b(r)∈Ωq−1(M;Fp)b(r) \in\Omega^{q-1}(M;\mathcal{F}_p). The form is horizontal if b=0b = 0. Wedge multiplication by drdr in this decomposition acts in the second exterior factor. By definition of the metric (see (11)), together with dvol⁡g=N dr dvol⁡kd\operatorname{vol}_g = \sqrt{N}\,dr\,d\operatorname{vol}_k, give

(a+dr∧b,c+dr∧d)X=N∫Re2pr((a,c)M+1N(b,d)M) dr.(24)(a + dr \wedge b,c + dr \wedge d)_X = \sqrt{N}\int_{\mathbb{R}} e^{2pr}\left((a,c)_M + \frac{1}{N}(b,d)_M\right)\,dr. \tag*{(24)}

for compactly supported families.

Lemma 3.2. In this decomposition,

∂ˉ(a+dr∧b)=dpa+dr∧(∂ra−dpb),(25)\bar{\partial}(a + dr \wedge b) = d_p a + dr \wedge(\partial_r a - d_p b), \tag*{(25)}
δˉ(a+dr∧b)=dp∗a−1N(∂r+2p)b−dr∧dp∗b.(26)\bar{\delta}(a + dr \wedge b) = d_p^*a - \frac{1}{N}(\partial_r + 2p)b - dr \wedge d_p^*b. \tag*{(26)}

Proof. In a flat coefficient frame, ∂ˉ\bar{\partial} is ordinary exterior differentiation of the second-factor forms. Thus,

∂ˉa=dpa+dr∧∂ra,∂ˉ(dr∧b)=−dr∧dpb,\bar{\partial}a = d_p a + dr \wedge\partial_r a, \qquad\bar{\partial}(dr \wedge b) = -dr \wedge d_p b,

which proves (25).

To compute the adjoint, take a compactly supported test form c(r)+dr∧d(r)c(r)+\mathrm{d}r\wedge d(r) with c(r)∈Ωq−1(M,Fp)c(r)\in\Omega^{q-1}(M,F_p) and d(r)∈Ωq−2(M,Fp)d(r)\in\Omega^{q-2}(M,F_p). Its derivative paired with ω=a+dr∧b\omega=a+\mathrm{d}r\wedge b is

N∫Re2pr((a,dpc)M+1N(b,∂rc−dpd)M) dr.\sqrt{N}\int_{\mathbb{R}}e^{2pr}\left((a,d_pc)_M+\frac{1}{N}(b,\partial_rc-d_pd)_M\right)\,\mathrm{d}r.

The terms involving dpd_p become (dp∗a,c)M(d_p^*a,c)_M and (−dp∗b,d)M(-d_p^*b,d)_M by integration on MM. Compact support in rr gives

∫Re2pr(b,∂rc)M dr=−∫Re2pr(∂rb+2pb,c)M dr.\int_{\mathbb{R}}e^{2pr}(b,\partial_rc)_M\,\mathrm{d}r=-\int_{\mathbb{R}}e^{2pr}(\partial_rb+2pb,c)_M\,\mathrm{d}r.

The pairing is therefore

N∫Re2pr((dp∗a−1N(∂r+2p)b,c)M+1N(−dp∗b,d)M) dr.\sqrt{N}\int_{\mathbb{R}}e^{2pr}\left(\left(d_p^*a-\frac{1}{N}(\partial_r+2p)b,c\right)_M+\frac{1}{N}(-d_p^*b,d)_M\right)\,\mathrm{d}r.

Comparison with (24) proves (26). □

For a horizontal form f(r)a(y)f(r)a(y), the formulas reduce to

∂ˉ(fa)=f dpa+f′dr∧a,δˉ(fa)=f dp∗a.(27)\bar{\partial}(fa)=f\,d_pa+f'\mathrm{d}r\wedge a,\qquad\bar{\delta}(fa)=f\,d_p^*a. \tag*{(27)}

We will need the Laplacian on such forms. Applying (26) to the two components of ∂ˉ(fa)\bar{\partial}(fa) gives

δˉ∂ˉ(fa)=f dp∗dpa−1N(f′′+2pf′)a−f′dr∧dp∗a,\bar{\delta}\bar{\partial}(fa)=f\,d_p^*d_pa-\frac{1}{N}(f''+2pf')a-f'\mathrm{d}r\wedge d_p^*a,
∂ˉδˉ(fa)=f dpdp∗a+f′dr∧dp∗a.\bar{\partial}\bar{\delta}(fa)=f\,d_pd_p^*a+f'\mathrm{d}r\wedge d_p^*a.

The radial exterior terms cancel, so

□‾(fa)=f Δp,qa−1N(f′′+2pf′)a.(28)\overline{\Box}(fa)=f\,\Delta_{p,q}a-\frac{1}{N}(f''+2pf')a. \tag*{(28)}

This holds for arbitrary smooth functions ff.

A Bochner identity on the compact quotient

We now define all the bundles and operators needed for the Bochner identity (5). The level hypersurface Σ\Sigma gives a section σ\sigma of EE: at the point represented by x∈Σx\in\Sigma, its value is x∈Vx\in V. The equivariance of Σ\Sigma makes this section well defined on MM. Let DED^E be the flat connection on EE. Since DYEσD^E_Y\sigma is the derivative of the immersion of Σ\Sigma, it identifies TMTM with a subbundle of EE. Moreover, dϕ(σ)=−N\mathrm{d}\phi(\sigma)=-N and dϕ(DYEσ)=0\mathrm{d}\phi(D^E_Y\sigma)=0, so

E=Rσ⊕DEσ(TM).E=\mathbb{R}\sigma\oplus D^E\sigma(TM).

In this subsection, a tangent vector YY regarded as an element of EE means DYEσD^E_Y\sigma. The cone metric on the slice defines a fiber metric GG on EE:

G(σ,σ)=N,G(σ,Y)=0,G(Y,Z)=k(Y,Z).(29)G(\sigma,\sigma)=N,\qquad G(\sigma,Y)=0,\qquad G(Y,Z)=k(Y,Z). \tag*{(29)}

The metric on ΛpE∗\Lambda^pE^* induced by GG is precisely the metric HpH_p defined in Section 3.1. Define ζ∈E∗\zeta\in E^* by ζ(σ)=1\zeta(\sigma)=1 and ζ(TM)=0\zeta(TM)=0. Then

ζ=G(σ,⋅)/N,ζ♯G=σ/N,∣ζ∣2=N−1.\zeta=G(\sigma,\cdot)/N,\qquad\zeta^{\sharp_G}=\sigma/N,\qquad|\zeta|^2=N^{-1}.

For any local frame eae_a of TMTM, let θa∈E∗\theta^a\in E^* annihilate σ\sigma and be dual to eae_a on TMTM. When working on the cone, we use the same symbols ζ,θa\zeta,\theta^a for their extensions by flat parallel transport along the rays, so their affine components are kept fixed. In a coordinate frame ea=∂yae_a=\partial_{y^a}, write σa=∂yaσ\sigma_a=\partial_{y^a}\sigma. The product parametrization Φ(r,y)=erσ(y)\Phi(r,y)=e^r\sigma(y) satisfies

dΦ=erσ dr+er∑bσb dyb.\mathrm{d}\Phi=e^r\sigma\,\mathrm{d}r+e^r\sum_b\sigma_b\,\mathrm{d}y^b.

Applying the defining duality relations gives

Φ∗ζ=er(ζ(σ) dr+∑bζ(σb) dyb)=er dr,\Phi^*\zeta=e^r\left(\zeta(\sigma)\,dr+\sum_b\zeta(\sigma_b)\,dy^b\right)=e^r\,dr,
Φ∗θa=er(θa(σ) dr+∑bθa(σb) dyb)=er dya.\Phi^*\theta^a=e^r\left(\theta^a(\sigma)\,dr+\sum_b\theta^a(\sigma_b)\,dy^b\right)=e^r\,dy^a.

Thus, suppressing pullback by Φ\Phi, we write

ζ=er dr,θa=er dya.\zeta=e^r\,dr,\qquad\theta^a=e^r\,dy^a.

In particular, θa1∧⋯∧θaq=eqrdya1∧⋯∧dyaq\theta^{a_1}\wedge\cdots\wedge\theta^{a_q}=e^{qr}dy^{a_1}\wedge\cdots\wedge dy^{a_q}.

For 0≤p,q≤N0\le p,q\le N, set

Bp,q=ΛpE∗⊗ΛqE∗,\mathcal{B}^{p,q}=\Lambda^pE^*\otimes\Lambda^qE^*,

with the tensor metric induced by GG and the base measure d volk\mathrm{d\,vol}_k. The inclusion of covectors annihilating σ\sigma gives an isometric bundle map

Jq:ΛpE∗⊗ΛqT∗M⟶Bp,q.(30)J_q:\Lambda^pE^*\otimes\Lambda^qT^*M\longrightarrow\mathcal{B}^{p,q}. \tag*{(30)}

It is the identity on the first factor. In a coordinate coframe,

Jq(λ⊗dya1∧⋯∧dyaq)=λ⊗θa1∧⋯∧θaq.J_q(\lambda\otimes dy^{a_1}\wedge\cdots\wedge dy^{a_q}) =\lambda\otimes\theta^{a_1}\wedge\cdots\wedge\theta^{a_q}.

Note that

ΛqE∗=Jq(ΛqT∗M)⊕ζ∧Jq−1(Λq−1T∗M).(31)\Lambda^qE^*=J_q(\Lambda^qT^*M)\oplus\zeta\wedge J_{q-1}(\Lambda^{q-1}T^*M). \tag*{(31)}

Here the notation JqJ_q in the single exterior factor denotes the same inclusion without the first-factor identity. Thus Jq∗J_q^* discards the terms containing a second-factor ζ\zeta and identifies the horizontal part with ordinary forms. In particular, Jq∗Jq=Id⁡J_q^*J_q=\operatorname{Id}.

Let DtotD^{\mathrm{tot}} be the connection on Bp,q\mathcal{B}^{p,q} induced by DED^E on both factors, in other words

DXtot(λ⊗η)=(DXλ)⊗η+λ⊗DXη,D_X^{\mathrm{tot}}(\lambda\otimes\eta)=(D_X\lambda)\otimes\eta+\lambda\otimes D_X\eta,

with the induced exterior-power connections on λ,η\lambda,\eta. In a local flat frame of EE, this is ordinary differentiation of all components.

Define

Pp,qu=∑a=1nε(θa)Deatotu−(p+q)ε(ζ)u,(32)P_{p,q}u=\sum_{a=1}^n\varepsilon(\theta^a)D_{e_a}^{\mathrm{tot}}u-(p+q)\varepsilon(\zeta)u, \tag*{(32)}
Pp,q:Γ(Bp,q)⟶Γ(Bp+1,q).P_{p,q}:\Gamma(\mathcal{B}^{p,q})\longrightarrow\Gamma(\mathcal{B}^{p+1,q}).

Let Pp,q∗:Γ(Bp+1,q)⟶Γ(Bp,q)P_{p,q}^*:\Gamma(\mathcal{B}^{p+1,q})\longrightarrow\Gamma(\mathcal{B}^{p,q}) be its formal adjoint for the metrics induced by GG and the volume d volk\mathrm{d\,vol}_k, i.e.

(Pp,qu,v)M=(u,Pp,q∗v)M(33)(P_{p,q}u,v)_M=(u,P_{p,q}^*v)_M \tag*{(33)}

for all smooth sections u,vu,v.

For the adjoint calculation below, we will also use the following identity

Lemma 3.3. The adjoint of exterior multiplication by ζ\zeta is the operator

ε(ζ)∗=ι(σ/N).\varepsilon(\zeta)^*=\iota(\sigma/N).

Proof. At each point, choose a GG-orthonormal basis v0=σ/N,v1,…,vnv_0=\sigma/\sqrt{N},v_1,\ldots,v_n, with dual basis ω0,…,ωn\omega^0,\ldots,\omega^n. Then ζ=ω0/N\zeta=\omega^0/\sqrt{N}. On the orthonormal basis of exterior monomials, exterior multiplication by ω0\omega^0 and contraction by v0v_0 have transposed matrices. Hence

ε(ζ)∗=1Nε(ω0)∗=1Nι(v0)=ι(σ/N).\varepsilon(\zeta)^*=\frac{1}{\sqrt{N}}\varepsilon(\omega^0)^* =\frac{1}{\sqrt{N}}\iota(v_0)=\iota(\sigma/N).

The identity extends to double forms since both operators act as the identity on the second exterior factor. □\square

We set exterior powers and corresponding operators to zero outside degrees 0,…,N0,\ldots,N. Define

Ap,q=Pp,qJq,Bp,q=Pp−1,q∗Jq.(34)A_{p,q}=P_{p,q}J_q,\qquad B_{p,q}=P^{*}_{p-1,q}J_q. \tag*{(34)}

Their domains and targets are

Ap,q:Ωq(M;Fp)⟶Γ(Bp+1,q),A_{p,q}:\Omega^q(M;F_p)\longrightarrow\Gamma(\mathcal{B}^{p+1,q}),
Bp,q:Ωq(M;Fp)⟶Γ(Bp−1,q).B_{p,q}:\Omega^q(M;F_p)\longrightarrow\Gamma(\mathcal{B}^{p-1,q}).

Proposition 3.4. On Ωq(M;Fp)\Omega^q(M;F_p) one has the differential-operator identity

Δp,q=Ap,q∗Ap,q+Bp,q∗Bp,q+(p−q−p2N)Id⁡.(35)\Delta_{p,q}=A^{*}_{p,q}A_{p,q}+B^{*}_{p,q}B_{p,q}+\left(p-q-\frac{p^2}{N}\right)\operatorname{Id}. \tag*{(35)}

Consequently,

Qp,q(a)=∥Ap,qa∥M2+∥Bp,qa∥M2+(p(N−p)N−q)∥a∥M2.(36)Q_{p,q}(a)=\lVert A_{p,q}a\rVert_M^2+\lVert B_{p,q}a\rVert_M^2+\left(\frac{p(N-p)}{N}-q\right)\lVert a\rVert_M^2. \tag*{(36)}

Proof. Put X=C/Γ≃R×MX=C/\Gamma\simeq\mathbb{R}\times M, and let ms:X→Xm_s:X\to X be the map induced by the dilation x↦esxx\mapsto e^s x on VV. We first define the invariant extension map. Recall that

Bp,q=ΛpE∗⊗ΛqE∗⟶M.\mathcal{B}^{p,q}=\Lambda^pE^*\otimes\Lambda^qE^*\longrightarrow M.

Define the set of dilation invariant sections as

Ap,q(X)dil={ω∈Γ∞(X,ΛpT∗X⊗ΛqT∗X)∣ms∗ω=ω for every s∈R}.\mathcal{A}^{p,q}(X)^{\mathrm{dil}}=\{\omega\in\Gamma^\infty(X,\Lambda^pT^*X\otimes\Lambda^qT^*X)\mid m_s^*\omega=\omega\text{ for every }s\in\mathbb{R}\}.

Then, the map

Up,q:Γ∞(M,Bp,q)⟶Ap,q(X)dilU_{p,q}:\Gamma^\infty(M,\mathcal{B}^{p,q})\longrightarrow\mathcal{A}^{p,q}(X)^{\mathrm{dil}}

defined by the componentwise formula

(Up,qu)I;J(erσ(y))=e−(p+q)ruI;J(y)(37)(U_{p,q}u)_{I;J}(e^r\sigma(y))=e^{-(p+q)r}u_{I;J}(y) \tag*{(37)}

is an isomorphism. Indeed, pullback by msm_s multiplies by e(p+q)se^{(p+q)s}, so

(ms∗Up,qu)I;J(erσ(y))=e(p+q)se−(p+q)(r+s)uI;J(y)=(Up,qu)I;J(erσ(y)).(m_s^*U_{p,q}u)_{I;J}(e^r\sigma(y))=e^{(p+q)s}e^{-(p+q)(r+s)}u_{I;J}(y)=(U_{p,q}u)_{I;J}(e^r\sigma(y)).

Conversely, dilation invariance forces precisely this scaling of the affine components. Its inverse, denoted by R0R_0, evaluates the full tensor at r=0r=0.

We next compute the operator induced by ∂\partial on dilation invariant double forms. Put m=p+qm=p+q. In local coordinates yay^a on MM, we have

D∂r(Up,qu)=−me−mru,D∂a(Up,qu)=e−mrDatotu.D_{\partial_r}(U_{p,q}u)=-me^{-mr}u,\qquad D_{\partial_a}(U_{p,q}u)=e^{-mr}D_a^{\mathrm{tot}}u.

Using

∂=ε(dr)D∂r+∑aε(dya)D∂a\partial=\varepsilon(dr)D_{\partial_r}+\sum_a\varepsilon(dy^a)D_{\partial_a}

we compute

∂(Up,qu)=e−(m+1)r(∑aε(ϑa)Datotu−mε(ζ)u).\partial(U_{p,q}u)=e^{-(m+1)r}\left(\sum_a\varepsilon(\vartheta^a)D_a^{\mathrm{tot}}u-m\varepsilon(\zeta)u\right).

The expression in parentheses is Pp,quP_{p,q}u by definition. Therefore,

∂(Up,qu)=Up+1,q(Pp,qu).(38)\partial(U_{p,q}u)=U_{p+1,q}(P_{p,q}u). \tag*{(38)}

To identify its adjoint operator, choose local frames of Bp−1,q\mathcal{B}^{p-1,q} and Bp,q\mathcal{B}^{p,q} over an open subset of MM, and extend their elements invariantly to XX using Up−1,qU_{p-1,q} and Up,qU_{p,q}, respectively. These extensions form local frames of the corresponding double-form bundles on XX. In these invariant frames, the operator ∂\partial acts like

∂=Pp−1,q+ε(ζ)∂r.\partial= P_{p-1,q} + \varepsilon(\zeta)\partial_r.

Indeed, its action on the frame elements is given by ∂U=UP\partial U = UP, while its action on their scalar coefficients follows from the Leibniz rule. The formal adjoint can now be computed in these same frames. We first make a remark about the metric used to compute the adjoint in this frame. Under flat affine transport, the metric on covectors scales by e2re^{2r}. Hence the induced metric on double forms of bidegree (p,q)(p,q) scales by e2(p+q)re^{2(p+q)r}. On the other hand, each invariant extension has the coefficient factor e−(p+q)re^{-(p+q)r}. Consequently, for u,v∈Γ∞(M,Bp,q)u,v \in\Gamma^\infty(M,\mathcal{B}^{p,q}),

⟨Up,qu,Up,qv⟩(r,y)=e2(p+q)re−2(p+q)r⟨u,v⟩G(y)=⟨u,v⟩(y).\langle U_{p,q}u,U_{p,q}v\rangle(r,y)=e^{2(p+q)r}e^{-2(p+q)r}\langle u,v\rangle_{G(y)}=\langle u,v\rangle(y).

Thus the invariant extension is a pointwise isometry, and the Gram matrices of the invariant frames are independent of rr. Moreover,

dvol⁡g=N dr dvol⁡k.d\operatorname{vol}_g=\sqrt{N}\,dr\,d\operatorname{vol}_k.

Integration by parts on compactly supported test forms therefore shows that the radial derivative of the scalar coefficients in these frames has formal adjoint

∂r∗=−∂r.\partial_r^*=-\partial_r.

Furthermore,

ε(ζ)∗=ι(σ/N).\varepsilon(\zeta)^*=\iota(\sigma/N).

Consequently,

δ=Pp−1,q∗−ι(σ/N)∂r.\delta=P^*_{p-1,q}-\iota(\sigma/N)\partial_r.

Here ∂r\partial_r differentiates the scalar coefficients relative to the invariant frames. The derivatives of the frame elements themselves are already included in Pp−1,qP_{p-1,q}. For u∈Γ∞(M,Bp,q)u \in\Gamma^\infty(M,\mathcal{B}^{p,q}), the coefficients of Up,quU_{p,q}u relative to these frames are independent of rr. It follows that

δ(Up,qu)=Up−1,q(Pp−1,q∗u).\delta(U_{p,q}u)=U_{p-1,q}(P^*_{p-1,q}u).

Together with the corresponding formula for ∂\partial, this gives

□(Up,qu)=Up,q(Lp,qu),(39)\Box(U_{p,q}u)=U_{p,q}(L_{p,q}u), \tag*{(39)}

where

Lp,q=Pp,q∗Pp,q+Pp−1,qPp−1,q∗.L_{p,q}=P^*_{p,q}P_{p,q}+P_{p-1,q}P^*_{p-1,q}.

We next compute □‾\overline{\Box} on the extension of JqηJ_q\eta, where η∈Ωq(M;Fp)\eta\in\Omega^q(M;F_p). Recall that in a flat coefficient frame, if

η=∑I,AηI;A(y) dxI⊗dyA\eta=\sum_{I,A}\eta_{I;A}(y)\,dx^I\otimes dy^A

then the inclusion JqJ_q replaces dyAdy^A by θA\theta^A at the slice. Since θA=eqrdyA\theta^A=e^{qr}dy^A, in flat coordinates we have

Up,q(Jqη)=e−prη.(40)U_{p,q}(J_q\eta)=e^{-pr}\eta. \tag*{(40)}

We already established in Section § 4.2 that

□‾(f(r)η)=f(r)Δp,qη−1N(f′′+2pf′)η.\overline{\Box}(f(r)\eta)=f(r)\Delta_{p,q}\eta-\frac{1}{N}(f''+2pf')\eta.

Taking f=e−prf=e^{-pr}, we have

□‾(f(r)η)=e−prΔp,qη+p2Ne−prη.\overline{\Box}(f(r)\eta)=e^{-pr}\Delta_{p,q}\eta+\frac{p^2}{N}e^{-pr}\eta.

Together with (40), this yields

□ˉ(Up,qJqη)=Up,qJq(Δp,qη+p2Nη).(41)\bar{\square}(U_{p,q}J_q\eta)=U_{p,q}J_q\left(\Delta_{p,q}\eta+\frac{p^2}{N}\eta\right). \tag*{(41)}

Substituting (39) and (41) into the cone identity

□ˉ=□+(p−q)Id⁡,\bar{\square}=\square+(p-q)\operatorname{Id},

we obtain

Up,q[Jq(Δp,qη+p2Nη)]=Up,q[Lp,qJqη+(p−q)Jqη].U_{p,q}\left[J_q\left(\Delta_{p,q}\eta+\frac{p^2}{N}\eta\right)\right]=U_{p,q}\left[L_{p,q}J_q\eta+(p-q)J_q\eta\right].

Applying R0R_0, we obtain

Jq(Δp,qη+p2Nη)=Lp,qJqη+(p−q)Jqη.J_q\left(\Delta_{p,q}\eta+\frac{p^2}{N}\eta\right)=L_{p,q}J_q\eta+(p-q)J_q\eta.

Finally, we apply Jq∗J_q^* on both sides and, since Jq∗Jq=Id⁡J_q^*J_q=\operatorname{Id} and the adjoints of Ap,qA_{p,q}, Bp,qB_{p,q} satisfy

Ap,q∗=Jq∗Pp,q∗andBp,q∗=Jq∗Pp−1,q,A_{p,q}^*=J_q^*P_{p,q}^* \qquad\text{and}\qquad B_{p,q}^*=J_q^*P_{p-1,q},

we have the desired

Δp,q=Ap,q∗Ap,q+Bp,q∗Bp,q+(p−q−p2N)Id⁡.\Delta_{p,q}=A_{p,q}^*A_{p,q}+B_{p,q}^*B_{p,q}+\left(p-q-\frac{p^2}{N}\right)\operatorname{Id}.

The formula for Qp,q(a)Q_{p,q}(a) follows by taking the inner product (Δp,qa,a)(\Delta_{p,q}a,a).

Computation of cohomology groups

The Bochner identity implies the following:

Proposition 3.5. For every a∈Ωq(M;Fp)a\in\Omega^q(M;F_p),

Qp,q(a)≥(p(N−p)N−q)∥a∥M2.(42)Q_{p,q}(a)\geq\left(\frac{p(N-p)}{N}-q\right)\lVert a\rVert_M^2. \tag*{(42)}

Proof. Both squared norms on the right-hand side of (36) are nonnegative. Removing them gives (42).

The flat connections need not preserve the positive fiber metrics used in this estimate. We recall the passage from the corresponding Hodge Laplacian to flat cohomology.

Lemma 3.6. Let FF be a flat real vector bundle over a closed Riemannian manifold, equipped with any positive fiber metric. Let dFd_F be its flat exterior derivative and dF∗d_F^* its formal adjoint. Every class in Hq(M;F)H^q(M;F) has a unique representative in

ker⁡ΔF,q,ΔF,q=dFdF∗+dF∗dF.\ker\Delta_{F,q},\qquad\Delta_{F,q}=d_Fd_F^*+d_F^*d_F.

Proof. Flatness gives dF2=0d_F^2=0. The principal symbol of dFd_F is exterior multiplication by a covector, tensored with Id⁡F\operatorname{Id}_F. Its adjoint symbol is contraction by the metric dual covector. The relation ε(ξ)ι(ξ♯)+ι(ξ♯)ε(ξ)=∣ξ∣2Id⁡\varepsilon(\xi)\iota(\xi^\sharp)+\iota(\xi^\sharp)\varepsilon(\xi)=|\xi|^2\operatorname{Id} shows that the principal symbol for ΔF,q\Delta_{F,q} is ∣ξ∣2Id⁡|\xi|^2\operatorname{Id}. Thus this operator is elliptic. It is formally self-adjoint and nonnegative, since

(ΔF,qa,a)=∥dFa∥2+∥dF∗a∥2.(\Delta_{F,q}a,a)=\lVert d_Fa\rVert^2+\lVert d_F^*a\rVert^2.

On a closed manifold, self-adjoint elliptic spectral theory gives its finite-dimensional smooth kernel, its orthogonal projection PP onto ker⁡(ΔF,q)\ker(\Delta_{F,q}), and a Green operator GG, satisfying

Id⁡=P+ΔFG=P+GΔF.\operatorname{Id}=P+\Delta_FG=P+G\Delta_F.

These operators preserve smooth forms. From dF2=(dF∗)2=0d_F^2=(d_F^*)^2=0 one obtains, by expanding the products, dFΔF=ΔFdFd_F\Delta_F=\Delta_Fd_F and the corresponding identity for dF∗d_F^*. The same commutation holds with GG. Therefore, if dFa=0d_Fa=0, then dFGa=GdFa=0d_FGa=Gd_Fa=0, so

a=Pa+ΔFGa=Pa+dFdF∗Ga.a=Pa+\Delta_FGa=Pa+d_Fd_F^*Ga.

The harmonic form PaPa represents the same cohomology class. If a harmonic form is exact, say a=dFba=d_{F}b, then ∥a∥2=(dF∗a,b)=0\lVert a\rVert^{2}=(d_{F}^{*}a,b)=0. This proves uniqueness as well. □

Proof of Theorem B. Apply Lemma 3.6 to FpF_p with the metrics in Proposition 3.5. A harmonic representative aa has Qp,q(a)=0Q_{p,q}(a)=0. If q<p(N−p)/Nq<p(N-p)/N, the coefficient in (42) is positive, and hence a=0a=0. Therefore Hq(M;Fp)=0H^{q}(M;F_p)=0. □

The Euler characteristic in dimension four

We now turn the analytic estimate into the topological result in Theorem A. We recall explicitly a well-known fact about cohomology with flat coefficients.

Lemma 4.1. Let MM be a closed oriented nn-manifold and let FF be a flat real vector bundle. Integration of the fiberwise evaluation pairing induces a nondegenerate pairing

Hq(M;F)×Hn−q(M;F∗)⟶R.(43)H^{q}(M;F)\times H^{n-q}(M;F^{*})\longrightarrow\mathbb{R}. \tag*{(43)}

Moreover,

χ(M;F)=rank⁡(F)χ(M).(44)\chi(M;F)=\operatorname{rank}(F)\chi(M). \tag*{(44)}

Proof of Theorem A. Assume first that MM is oriented. For n=4n=4, the standard bundle EE has rank N=5N=5. Taking p=2p=2 and p=3p=3 in Theorem B gives

H0(M;F2)=H1(M;F2)=0,H0(M;F3)=H1(M;F3)=0,(45)H^{0}(M;F_{2})=H^{1}(M;F_{2})=0,\qquad H^{0}(M;F_{3})=H^{1}(M;F_{3})=0, \tag*{(45)}

because p(5−p)/5=6/5p(5-p)/5=6/5 in both cases.

Let ν\nu be a parallel nonzero volume form on EE. Contraction with ν\nu is a flat bundle isomorphism

Λ2E⟶Λ3E∗,v∧w⟼ι(w)ι(v)ν.\Lambda^{2}E\longrightarrow\Lambda^{3}E^{*},\qquad v\wedge w\longmapsto\iota(w)\iota(v)\nu.

Consequently F2∗≅F3F_{2}^{*}\cong F_{3}. Applying Lemma 4.1 and (45), we obtain

H4(M;F2)≅H0(M;F3)∗=0,H3(M;F2)≅H1(M;F3)∗=0.H^{4}(M;F_{2})\cong H^{0}(M;F_{3})^{*}=0,\qquad H^{3}(M;F_{2})\cong H^{1}(M;F_{3})^{*}=0.

The only possibly nonzero cohomology group of F2F_{2} is therefore H2(M;F2)H^{2}(M;F_{2}). Since rank⁡(F2)=(52)=10\operatorname{rank}(F_{2})=\binom{5}{2}=10, (44) gives

10χ(M)=χ(M;F2)=dim⁡RH2(M;F2)≥0.10\chi(M)=\chi(M;F_{2})=\dim_{\mathbb{R}}H^{2}(M;F_{2})\geq0.

This proves both conclusions for oriented MM.

If MM is nonorientable, its orientation double cover M^\widehat{M} is again a closed properly convex projective four-manifold. The oriented case applies to M^\widehat{M}. Euler characteristic multiplies by the degree of a finite cover, so 2χ(M)=χ(M^)≥02\chi(M)=\chi(\widehat{M})\geq0. □

The spectrum of Euler characteristics.

Example 4.2. The inequality is attained by convex projective four-tori. Take the positive cone C=R>05C=\mathbb{R}^{5}_{>0}. Its projectivization is an open simplex. The determinant-one positive diagonal group is

A={diag⁡(eu1,…,eu5):u1+⋯+u5=0}≅R4.A=\{\operatorname{diag}(e^{u_{1}},\ldots,e^{u_{5}):u_{1}+\cdots+u_{5}=0\}\cong\mathbb{R}^{4}.}

In logarithmic coordinates, AA acts by translations on R4\mathbb{R}^{4}. A lattice Γ<A\Gamma<A therefore acts freely and cocompactly on the simplex, with quotient T4T^{4}. Its Euler characteristic is zero.

Corollary 4.3. The Euler characteristics of closed oriented properly convex real projective four-manifolds are exactly the non-negative even integers.

Proof. The splitting E=Rσ⊕DσE(TM)E = \mathbb{R}\sigma\oplus D^{E}_{\sigma}(TM) in § 4.3 gives a smooth bundle isomorphism TM⊕R≅ETM \oplus\mathbb{R} \cong E. Since EE is flat, Chern–Weil theory and stability give p1(TM)=0p_{1}(TM) = 0. This is the present case of Kobayashi’s theorem [14 Theorem 4.16], based on Avez’s identity; see [16 Section 2.2]. The Hirzebruch signature theorem gives σ(M)=0\sigma(M) = 0. Writing b2=b2++b2−b_{2} = b_{2}^{+} + b_{2}^{-}, this implies b2+=b2−b_{2}^{+} = b_{2}^{-}, and Poincaré duality gives

χ(M)=2−2b1+b2=2−2b1+2b2+∈2Z.\chi(M) = 2 - 2b_{1} + b_{2} = 2 - 2b_{1} + 2b_{2}^{+} \in2\mathbb{Z}.

It is nonnegative by Theorem A. Conversely, zero is realized in Example 4.2, and every positive even integer is realized by [16 Theorem 1.3]. □

Appendix A. Coordinate proof of Akagawa’s identities

We prove (16) for a positive Hessian metric g=Ddϕg = Dd\phi. The argument is local and uses only the first derivatives of gg. In particular, neither the canonical Monge–Ampère equation, nor completeness is assumed.

Metric derivatives and the Levi–Civita connection

Choose affine coordinates x1,…,xNx^{1}, \ldots, x^{N} and put Di=D∂iD_{i} = D_{\partial_{i}}. Thus Didxj=0D_{i}dx^{j} = 0, and the covariant components of gg and its cubic derivative are

gij=ϕij,Cijk=12ϕijk.g_{ij} = \phi_{ij}, \qquad C_{ijk} = \frac{1}{2}\phi_{ijk}.

The tensor CC is totally symmetric. We reserve ∇g\nabla^{g} for the Levi–Civita connection, to distinguish it from DD. Since every first derivative of gg is a third derivative of ϕ\phi, its Christoffel symbols simplify to

Γijk=12gkℓ(∂igjℓ+∂jgiℓ−∂ℓgij)=12gkℓϕijℓ,αi=∑jΓijj=12gabϕiab.(46)\Gamma^{k}_{ij} = \frac{1}{2}g^{k\ell}\left(\partial_{i}g_{j\ell} + \partial_{j}g_{i\ell} - \partial_{\ell}g_{ij}\right) = \frac{1}{2}g^{k\ell}\phi_{ij\ell}, \qquad\alpha_{i} = \sum_{j}\Gamma^{j}_{ij} = \frac{1}{2}g^{ab}\phi_{iab}. \tag*{(46)}

Thus ∇∂ig∂j=Γijk∂k\nabla^{g}_{\partial_{i}}\partial_{j} = \Gamma^{k}_{ij}\partial_{k}, whereas D∂i∂j=0D_{\partial_{i}}\partial_{j} = 0.

Differentiating giagaj=δjig^{ia}g_{aj} = \delta^{i}_{j} gives

∂kgij=−gia(∂kgab)gbj=−2giagjbCkab.\partial_{k}g^{ij} = -g^{ia}(\partial_{k}g_{ab})g^{bj} = -2g^{ia}g^{jb}C_{kab}.

For the density μ=det⁡(gij)\mu= \sqrt{\det(g_{ij})}, Jacobi’s determinant formula gives

∂iμ=12μ gab∂igab=αiμ.\partial_{i}\mu= \frac{1}{2}\mu\,g^{ab}\partial_{i}g_{ab} = \alpha_{i}\mu.

Fix a point x0x_{0}. A constant linear change of affine coordinates makes gij(x0)=δijg_{ij}(x_{0}) = \delta_{ij}. These remain affine coordinates; they are not Riemannian normal coordinates, and the derivatives of gg must still be retained. At this point,

Γijk=Cijk,∂kgij=−2Ckij,αi=∑jCijj,∑i∂igij=−2αj.(47)\Gamma^{k}_{ij} = C_{ijk}, \qquad\partial_{k}g^{ij} = -2C_{kij}, \qquad\alpha_{i} = \sum_{j}C_{ijj}, \qquad\sum_{i}\partial_{i}g^{ij} = -2\alpha_{j}. \tag*{(47)}

The last equality uses the total symmetry of CC.

Exterior operators and differentiation of the tensor metric

As in § 3, write

ei=ε(dxi),eˉi=εˉ(dxi),ιi=ι(∂i),ιˉi=ιˉ(∂i).e_{i} = \varepsilon(dx^{i}), \qquad\bar{e}_{i} = \bar{\varepsilon}(dx^{i}), \qquad\iota_{i} = \iota(\partial_{i}), \qquad\bar{\iota}_{i} = \bar{\iota}(\partial_{i}).

Their relations are

ιiej+ejιi=δijId⁡,ιˉieˉj+eˉjιˉi=δijId⁡.(48)\iota_{i}e_{j} + e_{j}\iota_{i} = \delta_{ij}\operatorname{Id}, \qquad\bar{\iota}_{i}\bar{e}_{j} + \bar{e}_{j}\bar{\iota}_{i} = \delta_{ij}\operatorname{Id}. \tag*{(48)}

Two contractions in the same factor anticommute, and all operators in different factors commute. In particular,

∂=∑ieiDi,∂ˉ=∑ieˉiDi,Λ=∑i,jgijιiιˉj.(49)\partial= \sum_{i}e_{i}D_{i}, \qquad\bar{\partial} = \sum_{i}\bar{e}_{i}D_{i}, \qquad\Lambda= \sum_{i,j}g^{ij}\iota_{i}\bar{\iota}_{j}. \tag*{(49)}

On double forms of a fixed bidegree, let

Si=∑j,kΓijk(ejik+eˉjiˉk).S_i=\sum_{j,k}\Gamma^k_{ij}(e_j i_k+\bar e_j\bar i_k).

This is the connection action on every covariant slot:

(Siω)I;J=∑r=1pΓiiraωi1…ir−1air+1…ip;J+∑s=1qΓijsbωI;j1…js−1bjs+1…jq.(S_i\omega)_{I;J}=\sum_{r=1}^{p}\Gamma^a_{ii_r}\omega_{i_1\ldots i_{r-1}ai_{r+1}\ldots i_p;J} +\sum_{s=1}^{q}\Gamma^b_{ij_s}\omega_{I;j_1\ldots j_{s-1}bj_{s+1}\ldots j_q}.

Consequently ∇igω=Diω−Siω\nabla^g_i\omega=D_i\omega-S_i\omega. At x0x_0, its expression is

Si=∑j,kCijk(ejik+eˉjiˉk).S_i=\sum_{j,k}C_{ijk}(e_j i_k+\bar e_j\bar i_k).

The matrix Γi\Gamma_i is self-adjoint for the relevant metric: lowering its upper index gives gkℓΓijℓ=Cijkg_{k\ell}\Gamma^\ell_{ij}=C_{ijk}, which is symmetric in j,kj,k. Its induced action SiS_i on exterior tensors is therefore self-adjoint. Since ∇g\nabla^g preserves their metric,

∂i⟨ω,η⟩g=⟨∇igω,η⟩g+⟨ω,∇igη⟩g=⟨Diω,η⟩g+⟨ω,Diη⟩g−2⟨ω,Siη⟩g.(50)\partial_i\langle\omega,\eta\rangle_g=\langle\nabla^g_i\omega,\eta\rangle_g+\langle\omega,\nabla^g_i\eta\rangle_g =\langle D_i\omega,\eta\rangle_g+\langle\omega,D_i\eta\rangle_g-2\langle\omega,S_i\eta\rangle_g. \tag*{(50)}

The formal adjoints

Take smooth test double forms supported in the affine chart. The integral of the ordinary derivative ∂i(μ⟨ω,η⟩g)\partial_i(\mu\langle\omega,\eta\rangle_g) is zero. Using ∂iμ=αiμ\partial_i\mu=\alpha_i\mu and (A.5), we get

0=∫(⟨Diω,η⟩g+⟨ω,Diη⟩g−2⟨ω,Siη⟩g+αi⟨ω,η⟩g) dvol⁡g.0=\int\left(\langle D_i\omega,\eta\rangle_g+\langle\omega,D_i\eta\rangle_g-2\langle\omega,S_i\eta\rangle_g+\alpha_i\langle\omega,\eta\rangle_g\right)\,d\operatorname{vol}_g.

Solving for the first term gives the exact local formal adjoint

Di∗=−Di−αi+2Si.(51)D_i^*=-D_i-\alpha_i+2S_i. \tag*{(51)}

Pointwise wedge multiplication has adjoints

ei∗=gijij,eˉi∗=gijˉiˉj.e_i^*=g^{ij}i_j,\qquad\bar e_i^*=g^{i\bar j}\bar i_j.

It follows that, on input of bidegree (p,q)(p,q),

δˉ=∑i,jDi∗(gijˉiˉj).\bar\delta=\sum_{i,j}D_i^*(g^{i\bar j}\bar i_j).

Here gijˉiˉjg^{i\bar j}\bar i_j first maps to bidegree (p,q−1)(p,q-1); the Di∗D_i^* that follows it acts on that lower bidegree. Expanding the derivative and then setting gij(x0)=δijg_{ij}(x_0)=\delta_{ij} gives

δˉ=−∑i,jgijˉiˉjDi−∑i,j(∂igijˉ)iˉj−∑i,jαigijˉiˉj+2∑i,jgijˉSiiˉj=−∑iiˉiDi+2∑iαiiˉi−∑iαiiˉi+2∑iSiiˉiat x0.\begin{aligned} \bar\delta=-\sum_{i,j}g^{i\bar j}\bar i_jD_i-\sum_{i,j}(\partial_i g^{i\bar j})\bar i_j-\sum_{i,j}\alpha_i g^{i\bar j}\bar i_j+2\sum_{i,j}g^{i\bar j}S_i\bar i_j \\ &=-\sum_i\bar i_iD_i+2\sum_i\alpha_i\bar i_i-\sum_i\alpha_i\bar i_i+2\sum_iS_i\bar i_i\qquad\text{at }x_0. \end{aligned}

There are two cubic contributions in ∑iSiiˉi\sum_iS_i\bar i_i. The one acting only in the second factor vanishes:

T:=∑i,j,kCijkeˉjiˉkiˉi=0.T:=\sum_{i,j,k}C_{ijk}\bar e_j\bar i_k\bar i_i=0.

Indeed, exchanging the dummy indices i,ki,k leaves CijkC_{ijk} unchanged and replaces iˉkiˉi\bar i_k\bar i_i by iˉiiˉk=−iˉkiˉi\bar i_i\bar i_k=-\bar i_k\bar i_i, so T=−TT=-T. The mixed contribution remains. We obtain

δˉ=−∑iiˉiDi+∑iαiiˉi+2∑i,j,kCijkejikiˉiat x0.(52)\bar\delta=-\sum_i\bar i_iD_i+\sum_i\alpha_i\bar i_i+2\sum_{i,j,k}C_{ijk}e_j i_k\bar i_i\qquad\text{at }x_0. \tag*{(52)}

Interchanging the two exterior factors gives

δ=−∑iιiDi+∑iαiιi+2∑i,j,kCijkeˉjιˉkιiat x0.(53)\delta=-\sum_i \iota_iD_i+\sum_i\alpha_i\iota_i+2\sum_{i,j,k}C_{ijk}\bar e_j\bar\iota_k\iota_i \quad\text{at }x_0. \tag*{(53)}

For this second formula, SiS_i acts after contraction on bidegree (p−1,q)(p-1,q), and its pure first-factor contribution vanishes by the same symmetry argument.

Commuting contraction past the exterior derivatives

We now compute Λ∂+∂Λ\Lambda\partial+\partial\Lambda at x0x_0. The coefficients gijg^{ij} in Λ\Lambda must again be differentiated before evaluating there. Using (A.4),

Λ∂+∂Λ=∑i,j,kqij(ιiek+ekιi)ιˉjDk+∑i,j,k(∂kgij)ekιˉiιˉj=∑iιˉiDi−2∑i,j,kCkjiekιˉiιˉj.\begin{aligned} \Lambda\partial+\partial\Lambda=\sum_{i,j,k}q^{ij}(\iota_i e_k+e_k\iota_i)\bar\iota_jD_k+\sum_{i,j,k}(\partial_k g^{ij})e_k\bar\iota_i\bar\iota_j \\ &=\sum_i\bar\iota_iD_i-2\sum_{i,j,k}C_{kji}e_k\bar\iota_i\bar\iota_j. \end{aligned}

In the first line, operators in different factors commute, and DkD_k commutes with contractions in the affine frame. The exterior relation ιiek+ekιi=δikId⁡\iota_i e_k+e_k\iota_i=\delta_{ik}\operatorname{Id} gives the first term in the second line. Total symmetry of CC and a renaming of indices rewrite its cubic term as

−2∑i,j,kCijkejιkιˉi.-2\sum_{i,j,k}C_{ijk}e_j\iota_k\bar\iota_i.

Comparison with (A.7) therefore gives

Λ∂+∂Λ=−δˉ+∑iαiιˉi=−δˉ+ιˉ(Xα).\Lambda\partial+\partial\Lambda=-\bar\delta+\sum_i\alpha_i\bar\iota_i=-\bar\delta+\bar\iota(X_\alpha).

At x0x_0, Xα=∑iαi∂iX_\alpha=\sum_i\alpha_i\partial_i, which explains the last expression. The mixed cubic terms agree exactly and cancel; only the trace α\alpha remains.

For the other factor, the same expansion gives

Λ∂ˉ+∂ˉΛ=∑i,j,kgijιi(ιˉjeˉk+eˉkιˉj)Dk+∑i,j,k(∂kgij)eˉkιiιˉj=∑iιiDi−2∑i,j,kCkjieˉkιiιˉj=∑iιiDi−2∑i,j,kCijkeˉjιˉkιi=−δ+ι(Xα),\begin{aligned} \Lambda\bar\partial+\bar\partial\Lambda=\sum_{i,j,k}g^{ij}\iota_i(\bar\iota_j\bar e_k+\bar e_k\bar\iota_j)D_k+\sum_{i,j,k}(\partial_k g^{ij})\bar e_k\iota_i\bar\iota_j \\ &=\sum_i\iota_iD_i-2\sum_{i,j,k}C_{kji}\bar e_k\iota_i\bar\iota_j \\ &=\sum_i\iota_iD_i-2\sum_{i,j,k}C_{ijk}\bar e_j\bar\iota_k\iota_i \\ &=-\delta+\iota(X_\alpha), \end{aligned}

where cross-factor contractions commute and the final comparison uses (A.8). This proves both identities in (16). Since x0x_0 was arbitrary and the operators are intrinsic, the identities hold on the entire Hessian manifold.

References

References

  1. [1][Aka19] S. Akagawa, Vanishing theorems of $L^2$-cohomology groups on Hessian manifolds, J. Math. Soc. Japan 71 (2019), no. 1, 65–89. doi:10.2969/jmsj/73797739.arxiv.org/abs/1702.06678
  2. [2][BDL18] S. A. Ballas, J. Danciger, and G.-S. Lee, Convex projective structures on nonhyperbolic three-manifolds, Geom. Topol. 22 (2018), no. 3, 1593–1646. doi:10.2140/gt.2018.22.1593.DOI
  3. [3][CX01] J. Cao and F. Xavier, Kähler parabolicity and the Euler number of compact manifolds of non-positive sectional curvature, Math. Ann. 319 (2001), no. 3, 483–491. doi:10.1007/PL00004444.DOI
  4. [4][CG86] J. Cheeger and M. Gromov, $L^2$-cohomology and group cohomology, Topology 25 (1986), no. 2, 189–215.
  5. [5][CY82] S.-Y. Cheng and S.-T. Yau, The real Monge–Ampère equation and affine flat structures, in Proceedings of the 1980 Beijing Symposium on Differential Geometry and Differential Equations, Science Press, Beijing, 1982, 339–370.
  6. [6]S.-S. Chern, On curvature and characteristic classes of a Riemann manifold, Abh. Math. Sem. Univ. Hamburg 20 (1955), 117–126. doi:10.1007/BF02960745.DOI
  7. [7]M. W. Davis and A. L. Edmonds, Euler characteristics of generalized Haken manifolds, Algebr. Geom. Topol. 14 (2014), no. 6, 3701–3716. doi:10.2140/agt.2014.14.3701.DOI
  8. [8]M. W. Davis and B. Okun, Vanishing theorems and conjectures for the ℓ²-homology of right-angled Coxeter groups, Geom. Topol. 5 (2001), 7–74. doi:10.2140/gt.2001.5.7.DOI
  9. [9]Y. Deng and B. Wang, Linear Chern–Hopf–Thurston conjecture, arXiv:2405.12012 (2024).arxiv.org/abs/2405.12012
  10. [10]D. J. F. Fox, A Schwarz lemma for Kähler affine metrics and the canonical potential of a proper convex cone, Ann. Mat. Pura Appl. (4) 194 (2015), no. 1, 1–42. doi:10.1007/s10231-013-0362-6.
  11. [11]M. Gromov, Kähler hyperbolicity and L²-Hodge theory, J. Differential Geom. 33 (1991), no. 1, 263–292. doi:10.4310/jdg/1214446039.DOI
  12. [12]F. E. A. Johnson and D. Kotschick, On the signature and Euler characteristic of certain four-manifolds, Math. Proc. Cambridge Philos. Soc. 114 (1993), no. 3, 431–437. doi:10.1017/S0305004100071711.DOI
  13. [13]J. Jost and K. Zuo, Vanishing theorems for L²-cohomology on infinite coverings of compact Kähler manifolds and applications in algebraic geometry, Comm. Anal. Geom. 8 (2000), no. 1, 1–30. doi:10.4310/CAG.2000.v8.n1.a1.DOI
  14. [14]S. Kobayashi, Projectively invariant distances for affine and projective structures, in Differential Geometry (Warsaw, 1979), Banach Center Publications 12, PWN, Warsaw, 1984, 127–152.
  15. [15]M. Olbrich, L²-invariants of locally symmetric spaces, Doc. Math. 7 (2002), 219–237.
  16. [16]S. Riolo, A. Seppi, and L. Slavich, Convex projective manifolds, symmetric spaces and geometric decompositions, arXiv:2607.05662.arxiv.org/abs/2607.05662

Paper details

vlnc:W_wyhykc9t
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Mathematics

Not recorded for this paper

View statement
Looking for related papers…
Contents