ScalingStacks

6 Trudinger Inequalities [057Z]

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

6 Trudinger Inequalities

In this section, we illustrate the versatility of our approach by establishing also inequalities of Trudinger type for general non-linear energies. Let f⁡(λ)f(\lambda) be a fully nonlinear operator satisfying the same structural conditions as in Theorem 1, and define for each p>0p>0,

Ep​(φ)=1V​∫X(−φ)p​fn​(λ⁡[hφ])​ωXnE_{p}(\varphi)=\frac{1}{V}\int_{X}(-\varphi)^{p}f^{n}(\lambda[h_{\varphi}])\omega_{X}^{n} (6.1)

(this notation is slightly different from the notation EtE_{t} used earlier for degenerating background metrics, but there should be no confusion, as the background metric is here fixed, and the index tt can be dropped). Recall that we had established before integral estimates for φ\varphi in terms of the entropy Entp{\rm Ent}_{p}. Trudinger estimates are also exponential estimates, but in terms of the energy EpE_{p}. We have

Theorem 7

Let φ\varphi be a plurisubharmonic function such that supφ=−1\sup\varphi=-1. Then

∫Xexp⁡{c⁡(n,p,γ)​α​(−φEp​(φ)1n+p)n+pn}​ωXn≤2​Cα\int_{X}{\rm exp}\Big\{c(n,p,\gamma)\alpha\Big(\frac{-\varphi}{E_{p}(\varphi)^{\frac{1}{n+p}}}\Big)^{\frac{n+p}{n}}\Big\}\omega_{X}^{n}\leq 2C_{\alpha} (6.2)

where α\alpha and CαC_{\alpha} are the constants coming from the α\alpha-invariant estimate of (X,ωX)(X,\omega_{X}).

We note that when specialized to the case when f⁡(λ)f(\lambda) is the Monge-Ampère operator f⁡(λ)=(∏k=1nλk)1nf(\lambda)=(\prod_{k=1}^{n}\lambda_{k})^{\frac{1}{n}}, our theorem recovers an inequality proved in [2, 10]. Moreover, our estimate has the major advantage that all constants there depend only on the α\alpha-invariant of the underlying manifold, hence is uniform over degenerating families with uniform α\alpha-invariants.

Proof of Theorem 7: Let us solve the following auxiliary Monge-Ampère equation with supψ=0\sup\psi=0, which is solvable due to Yau’s theorem [25].

(ωX+i​∂∂¯​ψ)n=(−φ)p​en​FEp​(φ)​ωXn(\omega_{X}+i\partial\bar{\partial}\psi)^{n}=\frac{(-\varphi)^{p}e^{nF}}{E_{p}(\varphi)}\omega_{X}^{n} (6.3)

then by the same maximum argument as in Lemma 1, we obtain the inequality

c⁡(n,p,γ)​(−φEp​(φ)1/(n+p))n+pn≤−ψ+C⁡(n,p,γ)​Ep​(φ)1p{c(n,p,\gamma)\Big(\frac{-\varphi}{E_{p}(\varphi)^{1/(n+p)}}\Big)^{\frac{n+p}{n}}}\leq-\psi+C(n,p,\gamma)E_{p}(\varphi)^{\frac{1}{p}} (6.4)

Now we pick κ=2nn+p​Cnn+p​c−nn+p\kappa=2^{\frac{n}{n+p}}C^{\frac{n}{n+p}}c^{-\frac{n}{n+p}}, where CC and cc are constants in estimate above, which only depends on n,pn,p and γ\gamma. Now set Uκ={−φ≤κEp(φ)1p}U_{\kappa}=\{-\varphi\leq\kappa E_{p}(\varphi)^{\frac{1}{p}}\}. Then on X∖UκX\setminus U_{\kappa}, we have by our choice of κ\kappa,

12​c​(n,p,γ)​(−φEp​(φ)1/(n+p))n+pn≤−ψ\frac{1}{2}c(n,p,\gamma)\Big(\frac{-\varphi}{E_{p}(\varphi)^{1/(n+p)}}\Big)^{\frac{n+p}{n}}\leq-\psi (6.5)

and on UκU_{\kappa}, we have

c⁡(n,p,γ)​(−φEp​(φ)1/(n+p))n+pn≤−c⁡(n,p,γ)​κpn​φc(n,p,\gamma)\Big(\frac{-\varphi}{E_{p}(\varphi)^{1/(n+p)}}\Big)^{\frac{n+p}{n}}\leq-c(n,p,\gamma)\kappa^{\frac{p}{n}}\varphi (6.6)

now multiplying by min⁡(c−1​κ−pn,1/2)​α\min(c^{-1}\kappa^{-\frac{p}{n}},1/2)\alpha and integrating, we get

∫Xexp⁡{c′​(n,p,γ)​α​(−φEp​(φ)1/(n+p))n+pn}​ωXn≤∫Uκe−α​ψ​ωXn+∫X∖Uκe−α​φ​ωXn≤2​Cα\int_{X}{\rm exp}\Bigg\{c^{\prime}(n,p,\gamma)\alpha\Big(\frac{-\varphi}{E_{p}(\varphi)^{1/(n+p)}}\Big)^{\frac{n+p}{n}}\Bigg\}\omega_{X}^{n}\leq\int_{U_{\kappa}}e^{-\alpha\psi}\omega_{X}^{n}+\int_{X\setminus U_{\kappa}}e^{-\alpha\varphi}\omega_{X}^{n}\leq 2C_{\alpha} (6.7)

which is the desired result.

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