ScalingStacks

Isoperimetric inequality [04EL]

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

Isoperimetric inequality

Lemma 5.10.

(Isoperimetric inequality cf. [63, Lem 3.10]) Let LL be a closed Lagrangian integral current in (X,ω,Ω)(X,\omega,\Omega), and consider a Euclidean coordinate ball B2​RB_{2R} on which the regularity scale is at least O⁡(R)O(R). Assume the quantitative almost calibrated condition cos⁡θ≥sin⁡ϵ\cos\theta\geq\sin\epsilon. Then there is a universal constant CC depending only on ϵ,n\epsilon,n and the metric uniform equivalence constant in (55), so that

M​a​s​s​(A)(n−1)/n≤C​M​a​s​s​(∂A)Mass(A)^{(n-1)/n}\leq CMass(\partial A)

for all closed subsets AA of supp​(L)∩BR\text{supp}(L)\cap B_{R} with rectifiable boundary. Here the Hausdorff measures are computed using the Calabi-Yau metric.

Proof.

The isoperimetric theorem [71, Thm 6.1] guarantees the existence of an integral current A′A^{\prime} supported in BRB_{R} such that ∂A′=∂A\partial A^{\prime}=\partial A and for which

M​a​s​s​(A′)(n−1)/n≤C​M​a​s​s​(∂A).Mass(A^{\prime})^{(n-1)/n}\leq CMass(\partial A).

Notice the metric uniform equivalence means we do not need to be careful to distinguish the Hausdorff measure for the Euclidean metric and the Calabi-Yau metric. Let TT denote the cone over the current A−A′A-A^{\prime}, then ∂T=A−A′\partial T=A-A^{\prime}, and thus by the quantitative calibrated condition,

M​a​s​s​(A)≤1sin⁡ϵ​∫ARe​Ω=1sin⁡ϵ​∫A′Re​Ω+1sin⁡ϵ​∫∂TRe​Ω≤1sin⁡ϵ​M​a​s​s​(A′)+1sin⁡ϵ​∫Td​Re​Ω≤1sin⁡ϵ​C​M​a​s​s​(∂A)n/(n−1),\begin{split}&Mass(A)\leq\frac{1}{\sin\epsilon}\int_{A}\text{Re}\Omega=\frac{1}{\sin\epsilon}\int_{A^{\prime}}\text{Re}\Omega+\frac{1}{\sin\epsilon}\int_{\partial T}\text{Re}\Omega\\ \leq&\frac{1}{\sin\epsilon}Mass(A^{\prime})+\frac{1}{\sin\epsilon}\int_{T}d\text{Re}\Omega\leq\frac{1}{\sin\epsilon}CMass(\partial A)^{n/(n-1)},\end{split}

which is the isoperimetric inequality. ∎

Remark 5.11.

The standard isoperimetric theorem inside the Euclidean space [71, Thm 6.1] does not state supp​(A′)⊂BR\text{supp}(A^{\prime})\subset B_{R}. Once we have found some A′A^{\prime} with ∂A′=∂A\partial A^{\prime}=\partial A with mass control, we can pushforward by a Lipschitz retraction map F:ℝ2​n→B1F:\mathbb{R}^{2n}\to B_{1}:

F⁡(x)={x,x∈BR,R​x|x|,x∈ℝ2​n∖BR.F(x)=\begin{cases}x,\quad x\in B_{R},\\ \frac{Rx}{|x|},\quad x\in\mathbb{R}^{2n}\setminus B_{R}.\end{cases}

Replacing A′A^{\prime} by F∗​A′F_{*}A^{\prime}, the mass cannot increase, and ∂F∗​(A′)=F∗​(∂A′)=F∗​(∂A)=∂A\partial F_{*}(A^{\prime})=F_{*}(\partial A^{\prime})=F_{*}(\partial A)=\partial A, and we have ensured the support is contained in BRB_{R}.

The following version of the isoperimetric theorem should be well known to experts, but for lack of a reference we include a proof below.

Proposition 5.11.

(Isoperimetric theorem on complete manifolds) Let XX be a complete Riemannian manifold, and ∂A\partial A be an mm-dimensional exact integral current supported in a fixed bounded open subset U⊂XU\subset X. Then there is an integral current A′A^{\prime} supported in a fixed large bounded subset of XX, with ∂A=∂A′\partial A=\partial A^{\prime} and

M​a​s​s​(A′)≤Const⋅min⁡{M​a​s​s​(∂A)(m+1)/m,M​a​s​s​(∂A)}.Mass(A^{\prime})\leq\text{Const}\cdot\min\{Mass(\partial A)^{(m+1)/m},Mass(\partial A)\}.
Proof.

We first isometrically embed XX into an ambient Euclidean space ℝN\mathbb{R}^{N}, so AA can be regarded as an integral current compactly supported in XX. Fix a small number ρ0>0\rho_{0}>0 such that over UU the ρ0\rho_{0}-neighbourhood in ℝN\mathbb{R}^{N} is isomorphic to the normal bundle, so there is a smooth retraction map FF back to UU. The Lipschitz norm of FF is approximately one.

Applying the deformation theorem for ℝN\mathbb{R}^{N} [71, section 5.3] to the current ∂A\partial A, with a parameter ρ≤ρ0\rho\leq\rho_{0} to be fixed, we can write

∂A=P+∂R,\partial A=P+\partial R,

where P,RP,R are integral currents inside ℝN\mathbb{R}^{N}, supported in the O⁡(ρ)O(\rho) neighbourhood of supp​(∂A)\text{supp}(\partial A), with

M​a​s​s​(R)≤C​ρ​M​a​s​s​(∂A),M​a​s​s​(P)≤C​M​a​s​s​(∂A),Mass(R)\leq C\rho Mass(\partial A),\quad Mass(P)\leq CMass(\partial A),

where the constant CC depends only on N,mN,m. Morever, PP is an integral linear sum of mm-dimensional faces in the standard grid decomposition of ℝN\mathbb{R}^{N} with cube size ρ\rho. We now push forward via FF:

F∗​P+∂F∗​R=F∗​(∂A)=∂A,F_{*}P+\partial F_{*}R=F_{*}(\partial A)=\partial A,

since ∂A⊂U⊂X\partial A\subset U\subset X is fixed by FF. Note that F∗​P,F∗​RF_{*}P,F_{*}R both live inside UU, and their mass bounds are essentially the same as P,RP,R respectively.

Suppose first that M​a​s​s​(∂A)≪1Mass(\partial A)\ll 1. If PP is nonzero, then by the grid description of PP,

ρm≤M​a​s​s​(P)≤C​M​a​s​s​(∂A)\rho^{m}\leq Mass(P)\leq CMass(\partial A)

So by choosing ρ=2​(C​M​a​s​s​(∂A))1/m\rho=2(CMass(\partial A))^{1/m} in the above, we force P=0P=0, so ∂A=∂F∗​R\partial A=\partial F_{*}R, with mass bound M​a​s​s​(F∗​R)≤const​M​a​s​s​(∂A)(m+1)/mMass(F_{*}R)\leq\text{const}Mass(\partial A)^{(m+1)/m}, so it suffices to take A′=F∗​RA^{\prime}=F_{*}R.

Now suppose M​a​s​s​(∂A)≳1Mass(\partial A)\gtrsim 1, then we choose ρ=ρ0\rho=\rho_{0}. Without loss of generality, we can replace ∂A\partial A by F∗​PF_{*}P, and pretend R=0R=0. We know

  • •

    PP is an integral linear combination of grid cube faces, where all the cubes lie in a bounded region of a fixed grid,

  • •

    F∗​PF_{*}P is an exact current on XX.

The set of all such PP form a finitely generated abelian group, which by classification is isomorphic to the direct sum of ⊕1rℤei\oplus_{1}^{r}\mathbb{Z}e_{i} and a finite abelian group. For any given element

P=∑ai​ei+finite group part,P=\sum a_{i}e_{i}+\text{finite group part},

the linear coefficients aia_{i} of eie_{i} for i=1,…​ri=1,\ldots r are bounded by |ai|≲M​a​s​s​(P)≲M​a​s​s​(∂A)|a_{i}|\lesssim Mass(P)\lesssim Mass(\partial A). Each eie_{i} gives rise to an exact simplicial chain F∗​eiF_{*}e_{i} inside XX, which is the boundary of a finite mass integral current QiQ_{i}. Thus

M​a​s​s​(∑1rai​Qi)≤∑1r|ai|​M​a​s​s​(Qi)≤const⋅M​a​s​s​(∂A).Mass(\sum_{1}^{r}a_{i}Q_{i})\leq\sum_{1}^{r}|a_{i}|Mass(Q_{i})\leq\text{const}\cdot Mass(\partial A).

The finite group part gives rise to another simplicical chain inside XX which is the boundary of some finite mass integral current. Thus we have produced an integral current A′A^{\prime} with ∂A′=∂A\partial A^{\prime}=\partial A, and mass bound

M​a​s​s​(A′)≤C⁡(M​a​s​s​(∂A))+C≤const ​M​a​s​s​(∂A).Mass(A^{\prime})\leq C(Mass(\partial A))+C\leq\text{const }Mass(\partial A).

since we are in the M​a​s​s​(∂A)≳1Mass(\partial A)\gtrsim 1 case. ∎

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