ScalingStacks

5. Proof of Theorem 1.1 [05E4]

Original official author HTML, exact retained edition. Historical TeX conversion verdicts remain unchanged. Cited-edition alignment and mathematical self-containment are not assessed.

Complete original source context · Original author HTML

5. Proof of Theorem 1.1

Now we are ready to prove Theorem 1.1.

Proof of Theorem 1.1.

Assume that the conclusion is not true. Then, for any fixed σ>1\sigma>1, there is a family of closed Ricci-flat Calabi-Yau nn-manifolds {(Mk,ωk,Jk,gk,Ωk)}\{(M_{k},\omega_{k},J_{k},g_{k},\Omega_{k})\} with [ωk]∈H2​(Mk,ℤ)[\omega_{k}]\in H^{2}(M_{k},\mathbb{Z}), and pk∈Mkp_{k}\in M_{k} such that

  • i)

    the injectivity radius and the sectional curvature

    igk​(pk)<1k,supBgk​(pk,1)|Kgk|≤1,i_{g_{k}}(p_{k})<\frac{1}{k},\ \ \ \sup_{B_{g_{k}}(p_{k},1)}|K_{g_{k}}|\leq 1,
  • ii)

    [Ωk|Bgk​(pk,σ​igk​(pk))]≠0[\Omega_{k}|_{B_{g_{k}}(p_{k},\sigma i_{g_{k}}(p_{k}))}]\neq 0 in Hn​(Bgk​(pk,σ​igk​(pk)),ℂ)H^{n}(B_{g_{k}}(p_{k},\sigma i_{g_{k}}(p_{k})),\mathbb{C}).

  • iii)

    for any open subset Wk′⊃Bgk​(pk,σ​igk​(pk))W_{k}^{\prime}\supset B_{g_{k}}(p_{k},\sigma i_{g_{k}}(p_{k})), (Wk′,ωk,Ωk)(W_{k}^{\prime},\omega_{k},\Omega_{k}) wouldn’t admit special lagrangian fibrations.

If we denote ω~k=igk−2​(pk)​ωk\tilde{\omega}_{k}=i^{-2}_{g_{k}}(p_{k})\omega_{k}, g~k=igk−2​(pk)​gk\tilde{g}_{k}=i^{-2}_{g_{k}}(p_{k})g_{k}, and Ω~k=igk−n​(pk)​Ωk\tilde{\Omega}_{k}=i^{-n}_{g_{k}}(p_{k})\Omega_{k}, then

ig~k​(pk)=1,supBg~k​(pk,k)|Kg~k|≤1k2,i_{\tilde{g}_{k}}(p_{k})=1,\ \ \ \sup_{B_{\tilde{g}_{k}}(p_{k},k)}|K_{\tilde{g}_{k}}|\leq\frac{1}{k^{2}},

and [Ω~k|Bg~k​(pk,σ)]≠0[\tilde{\Omega}_{k}|_{B_{\tilde{g}_{k}}(p_{k},\sigma)}]\neq 0 in Hn​(Bg~k​(pk,σ),ℂ)H^{n}(B_{\tilde{g}_{k}}(p_{k},\sigma),\mathbb{C}). By the Cheeger-Gromov’s convergence theorem (c.f. Theorem 2.1), a subsequence of (Mk,ω~k,g~k,Jk,Ω~k,pk)(M_{k},\tilde{\omega}_{k},\tilde{g}_{k},J_{k},\tilde{\Omega}_{k},p_{k}) converges to a complete flat Calabi-Yau nn-manifold (X,ω0,g0,J0,Ω0,p0)(X,\omega_{0},g_{0},J_{0},\Omega_{0},p_{0}) in the C∞C^{\infty}-sense, i.e. for any r>σr>\sigma, there are embeddings Fr,k:Bg0​(p0,r)⟶MkF_{r,k}:B_{g_{0}}(p_{0},r)\longrightarrow M_{k} such that Fr,k​(p0)=pkF_{r,k}(p_{0})=p_{k}, and Fr,k∗​g~kF_{r,k}^{*}\tilde{g}_{k} (resp. Fr,k∗​ω~kF_{r,k}^{*}\tilde{\omega}_{k} and Fr,k∗​Ω~kF_{r,k}^{*}\tilde{\Omega}_{k}) converges to g0g_{0} (resp. ω0\omega_{0} and Ω0\Omega_{0}) in the C∞C^{\infty}-sense. Furthermore, ig0​(p0)=1i_{g_{0}}(p_{0})=1. The soul theorem (c.f. [6], [29]) implies that there is a compact flat totally geodesic submanifold S⊂XS\subset X, the soul, such that (X,g0)(X,g_{0}) is isometric to the total space of the normal bundle ν⁡(S)\nu(S) with a metric induced by g0|Sg_{0}|_{S} and a natural flat connection.

By Proposition (3.4), there is a finite normal covering π:X~⟶X\pi:\tilde{X}\longrightarrow X with covering group Γ\Gamma such that

  • i)

    (X~,π∗​g0)(\tilde{X},\pi^{*}g_{0}) is isometric to (Tn×ℝn,h+hE)(T^{n}\times\mathbb{R}^{n},h+h_{E}), where Tn=ℝn/ΛT^{n}=\mathbb{R}^{n}/\Lambda, Λ\Lambda is a lattice in ℝn\mathbb{R}^{n}, hEh_{E} is the standard Euclidean metric on ℝn\mathbb{R}^{n}, and hh is the standard flat metric on TnT^{n} induced by hEh_{E}.

  • ii)

    The action of Γ\Gamma on X~\tilde{X} is a product action, i.e. there are Γ\Gamma-actions on TnT^{n} and ℝn\mathbb{R}^{n} such that γ⋅(x,y)=(γ⋅x,γ⋅y)\gamma\cdot(x,y)=(\gamma\cdot x,\gamma\cdot y) for any x∈Tnx\in T^{n} and y∈ℝny\in\mathbb{R}^{n}. Furthermore, Tn×{0}T^{n}\times\{0\} is Γ\Gamma-invariant, and S=(Tn×{0})/ΓS=(T^{n}\times\{0\})/\Gamma.

  • iii)
    π∗​ω0|Tn×{y}≡0,andπ∗​Im​e−1​θ0​Ω0|Tn×{y}≡0,\pi^{*}\omega_{0}|_{T^{n}\times\{y\}}\equiv 0,\ \ {\rm and}\ \ \pi^{*}{\rm Im}e^{\sqrt{-1}\theta_{0}}\Omega_{0}|_{T^{n}\times\{y\}}\equiv 0,

    for any y∈ℝny\in\mathbb{R}^{n}, and a constant θ0∈ℝ\theta_{0}\in\mathbb{R}.

Note that the Γ\Gamma-action on ℝn\mathbb{R}^{n} preserves hEh_{E}, and, BhE​(0,ρ)B_{h_{E}}(0,\rho), which implies that X~ρ=Tn×BhE​(0,ρ)\tilde{X}_{\rho}=T^{n}\times B_{h_{E}}(0,\rho) are invariant, for any ρ>0\rho>0. Lemma (3.3) shows [Fr,k∗​ω~k|S]=0[F_{r,k}^{*}\tilde{\omega}_{k}|_{S}]=0 in H2​(S,ℝ)H^{2}(S,\mathbb{R}), for k≫1k\gg 1, which implies [π∗​Fr,k∗​ω~k|X~ρ]=0[\pi^{*}F_{r,k}^{*}\tilde{\omega}_{k}|_{\tilde{X}_{\rho}}]=0 in H2​(X~ρ,ℝ)H^{2}(\tilde{X}_{\rho},\mathbb{R}), for any ρ>0\rho>0. By Remark (3.5), there are parallel 1-forms d​x1,⋯,d​xndx_{1},\cdots,dx_{n} on (Tn,h)(T^{n},h), which are pointwise linear independent, and coordinates y1,⋯,yny_{1},\cdots,y_{n} on ℝn\mathbb{R}^{n} such that

π∗​g0=∑(d​xj2+d​yj2),π∗​ω0=∑d​xj∧d​yj,e−1​θ0​π∗​Ω0=⋀j=1n(d​xj+−1​d​yj).\pi^{*}g_{0}=\sum(dx_{j}^{2}+dy_{j}^{2}),\ \ \ \pi^{*}\omega_{0}=\sum dx_{j}\wedge dy_{j},\ \ \ e^{\sqrt{-1}\theta_{0}}\pi^{*}\Omega_{0}=\bigwedge_{j=1}^{n}(dx_{j}+\sqrt{-1}dy_{j}).

Hence Condition 4.1 is satisfied.

Let r>ρ≫σr>\rho\gg\sigma such that Bg0​(p0,σ)⊂π⁡(X~ρ)⊂π⁡(X~2​ρ)⊂Bg0​(p0,r)B_{g_{0}}(p_{0},\sigma)\subset\pi(\tilde{X}_{\rho})\subset\pi(\tilde{X}_{2\rho})\subset B_{g_{0}}(p_{0},r). By Proposition (4.6), for k≫1k\gg 1, there is an open set Wk⊃X~ρW_{k}\supset\tilde{X}_{\rho} such that (Wk,π∗​Fr,k∗​ωk,π∗​Fr,k∗​Ωk)(W_{k},\pi^{*}F_{r,k}^{*}\omega_{k},\pi^{*}F_{r,k}^{*}\Omega_{k}) admits a equivariant special lagrangian fibration fk:Wk⟶Bkf_{k}:W_{k}\longrightarrow B_{k} of phase θk\theta_{k}, where Bk⊂ℝnB_{k}\subset\mathbb{R}^{n}, i.e. there is a Γ\Gamma-action on BkB_{k}, fkf_{k} is a Γ\Gamma-equivariant map, and

π∗​Fr,k∗​ωk|fk−1​(b)≡0,π∗​Fr,k∗​Im​e−1​θk​Ωk|fk−1​(b)≡0,\pi^{*}F_{r,k}^{*}\omega_{k}|_{f_{k}^{-1}(b)}\equiv 0,\ \ \ \pi^{*}F_{r,k}^{*}{\rm Im}e^{\sqrt{-1}\theta_{k}}\Omega_{k}|_{f_{k}^{-1}(b)}\equiv 0,

for any b∈Bkb\in B_{k}. Hence fkf_{k} induces a special lagrangian fibration f¯k:π⁡(Wk)⟶Bk/Γ\bar{f}_{k}:\pi(W_{k})\longrightarrow B_{k}/\Gamma, which implies that (Fr,k∘π⁡(Wk),ωk,Ωk)(F_{r,k}\circ\pi(W_{k}),\omega_{k},\Omega_{k}) admits a special lagrangian fibration, and Fr,k∘π⁡(Wk)⊃Bgk​(xk,σ​igk​(xk))F_{r,k}\circ\pi(W_{k})\supset B_{g_{k}}(x_{k},\sigma i_{g_{k}}(x_{k})). It is a contradiction. We obtain the conclusion. ∎

Original mathematics by the credited authors. Source-backed reader collection; mathematical self-containment is not assessed.