ScalingStacks

Proof. [02C8]

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.

The argument in the previous subsection implies that TT maps the smooth set in X∞X_{\infty} to the regular set in WW. So we need to show that if T⁡(p)T(p) is a smooth point of WW, then the limit metric on X∞X_{\infty} is also smooth at pp. Denote by ωi\omega_{i} the Kähler-Einstein metric on XiX_{i}, and ωi′\omega_{i}^{\prime} the induced Fubini-Study metric. Then we have ωi′=ωi+−1​∂∂¯​ϕi\omega_{i}^{\prime}=\omega_{i}+\sqrt{-1}\partial\bar{\partial}\phi_{i} with ϕi=k−1​log⁡ρk​(ωi)\phi_{i}=k^{-1}\log\rho_{k}(\omega_{i}). By our main Theorem 1.1 and Proposition 2.1 there is a constant C1>0C_{1}>0 such that |ϕi|L∞≤C1|\phi_{i}|_{L^{\infty}}\leq C_{1} for all ii. Also by arguments similar to the proof of Lemma 4.3 we see that there is a constant C2>0C_{2}>0 such that for all ii we have |∇ωiϕi|L∞≤C2|\nabla_{\omega_{i}}\phi_{i}|_{L^{\infty}}\leq C_{2}, and ωi′≤C2​ωi\omega_{i}^{\prime}\leq C_{2}\omega_{i}. Now write R​i​c​(ωi′)=λ​ωi′+−1​∂∂¯​hiRic(\omega_{i}^{\prime})=\lambda\omega_{i}^{\prime}+\sqrt{-1}\partial\bar{\partial}h_{i}, where λ\lambda is 11, −12-\frac{1}{2} or 00. So with suitable normalization of hih_{i} we have the equation

(4.1) ωin=ehi+λ​ϕi​ωi′n.\omega_{i}^{n}=e^{h_{i}+\lambda\phi_{i}}\omega_{i}^{\prime n}.

Then it is not hard to see that ∫Xihi2​ωi′n≤C3\int_{X_{i}}h_{i}^{2}\omega_{i}^{\prime n}\leq C_{3} for some constant C3>0C_{3}>0. Now for any pp in Wr​e​gW^{reg}, we choose a small neighborhood B⁡(p,δ)⊂Wr​e​gB(p,\delta)\subset W^{reg}. Then there are corresponding points pi∈Xip_{i}\in X_{i}, such that B⁡(pi,δ)B(p_{i},\delta) converges smoothly to B⁡(p,δ)B(p,\delta) in ℂ​ℙNk\mathbb{C}\mathbb{P}^{N_{k}}. By standard elliptic estimate we see that |hi|C1​(B⁡(pi,δ/2),ωi′)|h_{i}|_{C^{1}(B(p_{i},\delta/2),\omega_{i}^{\prime})} is uniformly bounded. Then by (4.1) there is a C4>0C_{4}>0 such that C4−1​ωi≤ωi′≤C4​ωiC_{4}^{-1}\omega_{i}\leq\omega_{i}^{\prime}\leq C_{4}\omega_{i} in B⁡(pi,δ/2)B(p_{i},\delta/2). Thus |∇ωi′ϕi|L∞​(B⁡(pi,δ/2))≤C4​|∇ωiϕi|L∞​(B⁡(pi,δ/2))≤C4​C2|\nabla_{\omega_{i}^{\prime}}\phi_{i}|_{L^{\infty}(B(p_{i},\delta/2))}\leq C_{4}|\nabla_{\omega_{i}}\phi_{i}|_{L^{\infty}(B(p_{i},\delta/2))}\leq C_{4}C_{2}. Then in B⁡(pi,δ/2)B(p_{i},\delta/2) with respect to the metric ωi′\omega_{i}^{\prime}, the right hand side of (4.1) has a uniform C1C^{1} bound. Therefore we can apply the Evans-Krylov theory(see for example [3]) to conclude that |ϕi||\phi_{i}| has a uniform C2,αC^{2,\alpha} bound in B⁡(pi,δ/4)B(p_{i},\delta/4). Then standard arguments show that all covariant derivatives of ϕi\phi_{i}(with respect to ωi′\omega_{i}^{\prime}) are uniformly bounded, so the Kähler-Einstein metrics ωi\omega_{i} converge smoothly in a neighborhood of pp.

∎

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