ScalingStacks

Proof. [00F7]

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

Proof.

Choose x0∈∂B⁡(p,R)x_{0}\in\partial B(p,R), so that d⁡(x0,p)=Rd(x_{0},p)=R, and denote by ρ⁡(x)=d⁡(x,x0)\rho(x)=d(x,x_{0}). The Laplacian comparison theorem gives Δ​ρ2≤4​n\Delta\rho^{2}\leq 4n in the sense of distributions. Let φ⁡(x)=ψ⁡(ρ⁡(x))\varphi(x)=\psi(\rho(x)) where

ψ⁡(t)={1if 0≤t≤R−1,12​(R+1−t)if R−1<t<R+1,0if t≥R+1.\psi(t)=\left\{\begin{array}[]{lll}1&\mbox{if $0\leq t\leq R-1$},\\ \frac{1}{2}(R+1-t)&\mbox{if $R-1<t<R+1$},\\ 0&\mbox{if $t\geq R+1$}.\end{array}\right.

Then φ\varphi is a nonnegative Lipschitz function supported in B⁡(x0,R+1)B(x_{0},R+1), and we have that

∫Mφ​Δ​ρ2​d​Vg=−∫B⁡(x0,R+1)∇φ⋅∇ρ2dVg=−2∫B⁡(x0,R+1)ρ|∇ρ|2ψ′(ρ(x))dVg=∫B⁡(x0,R+1)\B⁡(x0,R−1)ρ​d​Vg≥(R−1)​Vol​(B⁡(x0,R+1)\B⁡(x0,R−1)),\begin{split}\int_{M}\varphi\Delta\rho^{2}dV_{g}&=-\int_{B(x_{0},R+1)}\nabla\varphi\cdot\nabla\rho^{2}dV_{g}=-2\int_{B(x_{0},R+1)}\rho|\nabla\rho|^{2}\psi^{\prime}(\rho(x))dV_{g}\\ &=\int_{B(x_{0},R+1)\backslash B(x_{0},R-1)}\rho dV_{g}\\ &\geq(R-1)\mathrm{Vol}(B(x_{0},R+1)\backslash B(x_{0},R-1)),\end{split}

and also

∫Mφ​Δ​ρ2​d​Vg≤4​n​∫B⁡(x0,R+1)φ​d​Vg≤4​n​Vol​(B⁡(x0,R+1)).\int_{M}\varphi\Delta\rho^{2}dV_{g}\leq 4n\int_{B(x_{0},R+1)}\varphi dV_{g}\leq 4n\mathrm{Vol}(B(x_{0},R+1)).

Notice that B⁡(p,1)⊂B⁡(x0,R+1)\B⁡(x0,R−1)B(p,1)\subset B(x_{0},R+1)\backslash B(x_{0},R-1) and so the previous two equations give

(R−1)​Vol​(B⁡(p,1))≤4​n​Vol​(B⁡(x0,R+1)).(R-1)\mathrm{Vol}(B(p,1))\leq 4n\mathrm{Vol}(B(x_{0},R+1)).

The conclusion follows from the fact that B⁡(x0,R+1)⊂B⁡(p,2​(R+1))B(x_{0},R+1)\subset B(p,2(R+1)). ∎

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