Modulus of continuity for solutions to complex Monge-Ampère equations on Hermitian manifolds
Junbang Liu · 2025
Read Modulus of continuity for solutions to complex Monge-Ampère equations on Hermitian manifolds by Junbang Liu (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2510-22774.txt.
Introduction
Junbang Liu, writing alone in 2025, proves a continuity result for a hard equation. The equation is the complex Monge-Ampere equation. It is set on compact Hermitian manifolds. Such manifolds carry both complex and metric structure. The claim concerns how rough the solution can be.
Liu proves uniform log-continuity of the solution. Log-continuity is a weak but uniform control on change. The result extends a prior theorem. That theorem covered the Kahler case only. Hermitian manifolds form a broader class. The note gives a full proof of the wider claim. It adds to the theory of complex geometric equations.
Catalog record
| CREATOR | Junbang Liu |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 35 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2510-22774
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Modulus of continuity for solutions to complex Monge-Ampère equations on Hermitian manifolds Junbang Liu Abstract . In this note, we give a proof of the uniform log-continuity of the solution to complex Monge-Ampère equations on compact Hermitian manifolds, which is a generalization of the result of [ 13 ] in the Kähler case. 1. introduction Let (M^{n},\omega_{M}) be a n -dim compact Hermitian manifold with a fixed Hermitian metric \omega_{M} . Let F\in C^{\infty}(M,\mathbb{R}) be a smooth real function. Then it’s proved by Tosatti and Weinkove in [ 21 ] that the following volume conjecture holds: there exists an unique \varphi\in C^{\infty}(M,\mathbb{R}) and unique constant c such that (1.1) (\omega_{M}+\sqrt{-1}\partial\bar{\partial}u)^{n}=e^{F+c}\omega_{M}^{n},\quad\omega_{M}+\sqrt{-1}\partial\bar{\partial}u>0,\quad\sup_{M}u=0. The theorem we proved in this note is Theorem 1.1 . Given p>n , for any 0<\alpha<\frac{p}{n}-1 , there exists a constant C depending on M,\omega_{M},p,\alpha , and upper bound of \int_{M}(|F|^{p}+1)e^{F}\omega_{M}^{n} , positive lower bound of \int_{M}e^{\frac{1}{n}F}\omega_{M}^{n} , such that solutions to ( 1.1 ) satisfies |u(x)-u(y)|\leq\frac{C}{|\log dist_{\omega_{M}}(x,y)|^{\alpha}}. It’s a generalization of the theorem in [ 13 , 10 ] to the Hermitian manifolds. We remark that in the Kähler case, the exponent \alpha in the \log -continuity can be \frac{p}{n}-1 . The Hölder regularity of the solutions to complex Monge-Ampère equation on compact Hermitian manifolds have been studied by many authors, see [ 19 , 15 ] and the reference therein, after the internsively developed pluripotential theorey on Hermitian manifolds(see [ 8 , 20 , 16 ] ). Here we use a purely PDE method developed by Guo-Phong-Tong in [ 12 ] . In the following, we say a constant is uniform if it depends on the parameters stated in the above theorem. 2. stability estimate The following stability estimate is crucial to our proof. Proposition 2.1 . For any v\in Psh(\omega_{M})\cap C^{\infty}(M,\mathbb{R}^{+}) , there exists a uniform constant C additionally depending on ||v||_{L^{\infty}} such that \sup(v-u)\leq\frac{C}{|\log(||(v-u)_{+}||_{L^{1}})|^{\alpha}}. To begin with, we need a lemma to estimate the constant c under the assumption of a weaker bound on the volume density. Lemma 2.2 . Let (u,c) be a smooth solution to equation ( 1.1 ) and \Phi:\mathbb{R}\rightarrow\mathbb{R}^{+} be a positive increasing function with \int_{1}^{\infty}\Phi^{-\frac{1}{n}}(t)dt<\infty . Then there exists a constant C_{0} depending on M,\omega_{M},\Phi and upper bounds for \int_{M}\Phi(F)e^{F}\omega_{M}^{n}, and positive lower bound of \int_{M}e^{\frac{F}{n}}\omega_{M}^{n} , such that e^{c}+e^{-c}+||u||_{L^{\infty}}\leq C_{0}. Proof. The lower bound c is proved in [ 18 ] by the complex ABP estimate. The upper bound of c is proved in [ 16 ] lemma 5.9, and it depends on the positive lower bound of \int_{M}e^{\frac{F}{n}}\omega_{M}^{n},M,\omega_{M} . We only need to prove the L^{\infty} -estimate. Here, we introduce another method to construct a good comparison metric. We cover the manifold (M,\omega_{M}) by coordinates balls \{B(x_{i},2r_{i})|i=1,\cdots,n\} such that B_{i}:=B(x_{i},r_{i}) still covers M . Let \rho_{i} be smooth functions on M , such that 2B_{i}=\{\rho_{i}\leq 0\} and \rho_{i}=|z|^{2}-4r_{i}^{2} on 2B_{i} . On each 2B_{i} , by the work of Caffarelli-Kohn-Nirenberg-Spruck [ 2 ] , one can solve the following equation (2.1) (\sqrt{-1}\partial\bar{\partial}\psi_{i})^{n}=\Phi(F)e^{F}\omega_{M}^{n},\quad\sqrt{-1}\partial\bar{\partial}\psi_{i}>0,\quad\psi_{i}|_{\partial 2B_{i}}=-1. By the complex ABP estimate in [ 18 ] , ||\psi_{i}||_{L^{\infty}} is uniformly bounded by a constant C_{1} . Then by choosing C_{2} large enough with 3C_{2}r_{i}^{2}>C_{1}+1 and \epsilon small, let \tilde{\psi}_{i}:=\max_{\epsilon}\{C_{2}\rho_{i},\psi_{i}\} , where \max_{\epsilon} is the regularized max function, c.f. lemma 5.18 in Demailly’s book [ 5 ] . Then the function \tilde{\psi}_{i} satisfies: • \tilde{\psi}_{i}=\psi_{i} on B_{i} (since C_{2}\rho_{i}<-C_{1}-1<\psi_{i} on B_{i} with our choice of C_{2} ); • \sqrt{-1}\partial\bar{\partial}\tilde{\psi}_{i}\geq-C_{3}\omega_{M} on M , for some uniform constant C_{3} depending on M,\omega_{M} (since \rho_{i} is smooth on M , and \tilde{\psi}_{i}=\rho_{i} in a neighborhood of M\setminus 2B_{i} ). Let \psi:=\frac{1}{C_{3}N}\sum_{i=1}^{N}\tilde{\psi}_{i} . Therefore, one has on each B_{i} , (\omega_{M}+\sqrt{-1}\partial\bar{\partial}\psi)^{n}\geq\frac{1}{(NC_{3})^{n}}(\sqrt{-1}\partial\bar{\partial}\psi_{i})^{n}\geq\epsilon_{0}\Phi(F)e^{F}\omega_{M}^{n} for some uniform constant \epsilon_{0} . Once one has such a comparison metric, the proof of L^{\infty} -estimate follows from [ 17 ] . Remark 2.3 . It’s interesting to know whether the constant c can be estimated from below when we remove the condition \int_{1}^{\infty}\Phi^{-\frac{1}{n}}<\infty ? If so, it can be applied to get Trudinger type estimates for solutions to Monge-Ampère equations on Hermitian manifolds like in the Kähler case. ∎ With this lemma, we can construct appropriate comparison Hermitian metrics to reduce the proof to L^{p} -case. Let \Phi(x)=x^{n\alpha} for x>1 , and extend it smoothly to be a positive increasing function on \mathbb{R} . Then \int_{M}(|\log\Phi(F)+F|^{p-n\alpha}+1)\Phi(F)e^{F}\omega_{M}^{n}\leq C\int_{M}(|F|+1)^{p}e^{F}\omega_{M}^{n}. Let \psi_{1} be the solution to equation (2.2) (\omega_{M}+\sqrt{-1}\partial\bar{\partial}\psi_{1})^{n}=e^{c_{1}}\Phi(F)e^{F}\omega_{M}^{n},\quad\omega_{M}+\sqrt{-1}\partial\bar{\partial}\psi_{1}>0,\quad\sup\psi_{1}=0. Then by lemma 2.2 , the constant e^{-c_{1}} and ||\psi_{1}||_{L^{\infty}} is uniformly bounded from above by a uniform constant C_{0} . Let \alpha_{M} be Tian’s \alpha -invariant of (M,\omega_{M}) . For any q,\Lambda_{1}>0,a\in(0,1) , set h(s)=-\frac{q}{a\alpha_{M}}e^{\frac{c-c_{1}}{n}}\int_{s}^{C_{0}+\Lambda_{1}}\Phi^{-\frac{1}{n}}(t)dt , and \psi_{2}=-h(-\frac{\alpha_{M}}{q}\psi_{1}+\Lambda_{1}) . Without loss of generality, we can assume that \sup(-\frac{\alpha_{M}}{q}\psi)\leq C_{0} , so in this case, we get that \psi_{2} is a nonnegative function. We have Lemma 2.4 . The inequality holds (2.3) \omega_{u}^{n}\leq a^{n}\omega_{\psi_{2}}^{n}+f\omega_{M}^{n}, with f=\exp{(-\frac{\alpha}{q}\psi_{1}+\Lambda_{1}+c)} and \Lambda_{1} is chosen so that \Phi(\Lambda_{1})^{\frac{1}{n}}=\frac{1}{a}e^{\frac{c-c_{1}}{n}}. Proof. From our choice of \Lambda_{1} , we have \frac{\alpha}{q}h^{\prime}(\Lambda_{1})=1 . Note that h is an increasing concave function, we have \displaystyle\omega_{M}+\sqrt{-1}\partial\bar{\partial}\psi_{2}= \displaystyle\omega_{M}+\frac{\alpha}{q}h^{\prime}\sqrt{-1}\partial\bar{\partial}\psi_{1}-\frac{\alpha^{2}}{q^{2}}h^{\prime\prime}\sqrt{-1}\partial\psi_{1}\wedge\bar{\partial}\psi_{1}\geq\frac{\alpha}{q}h^{\prime}(\omega_{M}+\sqrt{-1}\partial\bar{\partial}\psi_{1}) In the above inequality we have used that \frac{\alpha}{q}h^{\prime}(-\frac{\alpha}{q}\psi_{1}+\Lambda_{1})\leq\frac{\alpha}{q}h^{\prime}(\Lambda_{1})=1. Hence \omega_{u}^{n}=e^{F+c}\omega_{M}^{n}\leq\frac{q^{n}}{\alpha^{n}}\frac{1}{h^{\prime n}}\frac{e^{c}}{e^{c_{1}}}\frac{1}{\Phi(F)}\omega_{\psi_{2}}^{n}. Next, for any a\in(0,1) , if \frac{q^{n}}{\alpha^{n}}\frac{1}{h^{\prime n}}\frac{e^{c}}{e^{c_{1}}}\frac{1}{\Phi(F)}\leq a^{n} , we get \omega_{u}^{n}\leq a^{n}\omega_{\psi_{2}}. Otherwise, we have \frac{q^{n}}{\alpha^{n}}\frac{1}{h^{\prime n}}\frac{e^{c}}{e^{c_{1}}}\frac{1}{\Phi(F)}>a^{n} , which is equivalent to \Phi(F)<\frac{q^{n}}{a^{n}\alpha^{n}}\frac{1}{h^{\prime n}(-\alpha q^{-1}\psi_{1}+\Lambda_{1})}e^{c-c_{1}}=\Phi(-\frac{\alpha}{q}\psi_{1}+\Lambda_{1}), where we use the definition of h in the last equality. From the monotonicity of \Phi , we get F\leq-\frac{\alpha}{q}\psi_{1}+\Lambda_{1} . So, in this case, we have e^{F+c}\leq f with f stated in the lemma. ∎ For a\in(0,1) and {p_{1}}>1 , which will be specified later, let \tilde{u}:=-u+(1-2a)v+a\psi_{2} , and \Omega_{s}:=\{\tilde{u}-s>0\}. We use a trick of Cheng-Xu in [ 4 ] , and consider the auxiliary equation (\omega_{M}+\sqrt{-1}\partial\bar{\partial}\psi_{3})^{n}=\frac{e^{c_{s,j,{p_{1}}}}\eta_{j}(\tilde{u}-s)f\omega_{M}^{n}}{A_{s,j,{p_{1}}}},\quad\omega_{M}+\sqrt{-1}\partial\bar{\partial}\psi_{3}>0,\quad\sup\psi_{3}=0, where A_{s,j,{p_{1}}}=\left(\int_{M}\eta_{j}^{{p_{1}}}(\tilde{u}-s)f^{{p_{1}}}\omega_{M}^{n}\right)^{\frac{1}{{p_{1}}}}, and \eta_{j}(x) is a sequence of positive smooth function approximating \max\{x,0\} from above in L^{\infty} sense. From dominated convergence theorem, it’s easy to see \lim_{j\rightarrow\infty}A_{s,j,{p_{1}}}=\left(\int_{\Omega_{s}}(\tilde{u}-s)_{+}^{{p_{1}}}f^{{p_{1}}}\omega_{M}^{n}\right)^{\frac{1}{{p_{1}}}}=:A_{s,{p_{1}}}. Then apply lemma 2.2 again with \Phi(x)=e^{({p_{1}}-1)x} , the constant c_{s,j,{p_{1}}} and solution \psi_{3} are uniformly bounded. Choose constant \epsilon_{3},\Lambda_{3} suitably, then we have Lemma 2.5 . The inequality holds: \tilde{u}-s\leq\epsilon_{3}(\psi_{3}+\Lambda_{3})^{\frac{n}{n+1}} for \epsilon_{3}=\left(\frac{n+1}{n}\right)^{\frac{n}{n+1}}e^{-c_{s,j,{p_{1}}}}A_{s,j,{p_{1}}}^{\frac{1}{n+1}},\quad\Lambda_{3}=\frac{n}{n+1}\frac{A_{s,j,{p_{1}}}}{a^{n+1}}e^{-(n+1)c_{s,j,{p_{1}}}}. Proof. Without loss of generality, we can assume the maximum of \tilde{u}-s-\epsilon_{3}(-\psi_{3}+\Lambda_{3})^{\frac{n}{n+1}} is attained at some point x_{0}\in\Omega_{s} ; otherwise the inequality holds automatically. Then at x_{0} , by maximum principle, we have (2.4) \displaystyle 0\geq \displaystyle\sqrt{-1}\partial\bar{\partial}\tilde{u}+\epsilon_{3}\frac{n}{n+1}(-\psi_{3}+\Lambda_{3})^{-\frac{1}{n+1}}\sqrt{-1}\partial\bar{\partial}\psi_{3}+\epsilon\frac{n^{2}}{n+1}\sqrt{-1}\partial\psi_{3}\wedge\bar{\partial}\psi_{3} \displaystyle\geq \displaystyle-\omega_{u}+(1-2a)\omega_{v}+a\omega_{\psi_{2}}+\epsilon_{3}\frac{n}{n+1}(-\psi_{3}+\Lambda_{3})^{-\frac{1}{n+1}}\omega_{\psi_{3}}+(a-\epsilon_{3}\frac{n}{n+1}(\Lambda_{3})^{-\frac{1}{n+1}})\omega_{M} From our choice of \Lambda_{3} , the last term above is nonnegative. So we have (2.5) \displaystyle\omega_{u}^{n}\geq \displaystyle a^{n}\omega_{\psi}^{n}+\left(\epsilon_{3}\frac{n}{n+1}\right)^{n}(-\psi_{3}+\Lambda_{3})^{-\frac{n}{n+1}}\omega_{\psi_{3}}^{n} \displaystyle= \displaystyle a^{n}\omega_{\psi}^{n}+\left(\epsilon_{3}\frac{n}{n+1}\right)^{n}\frac{\eta_{j}(\tilde{u}-s)}{(-\psi_{3}+\Lambda_{3})^{\frac{n}{n+1}}}\frac{e^{c_{s,j,{p_{1}}}}f\omega_{M}^{n}}{A_{s,j,{p_{1}}}} Then apply lemma 2.4 to the left-hand side of ( 2.5 ), we get \eta_{j}(\tilde{u}-s)\leq\left(\epsilon_{3}\frac{n}{n+1}\right)^{-n}e^{-c_{s,j,{p_{1}}}}A_{s,j,{p_{1}}}(-\psi_{3}+\Lambda_{3})^{\frac{n}{n+1}}. Note that \eta_{j}(x)\geq\max\{x,0\} and from our choice of \epsilon_{3} , we have \epsilon_{3}=\left(\epsilon_{3}\frac{n}{n+1}\right)^{-n}e^{-c_{s,j,{p_{1}}}}A_{s,j,{p_{1}}} . The inequality in the lemma follows. ∎ Let \phi(s)=\int_{\Omega_{s}}f\omega_{M}^{n} , one can estimate A_{s,{p_{1}}} as Lemma 2.6 . If \frac{A_{s,{p_{1}}}}{a^{n+1}}\leq 1 , and fixed p_{1}>1 close to 1 , one has A_{s,{p_{1}}}\leq Ce^{A_{1}\Lambda_{1}}\phi(s)^{1+\delta_{n}} for some uniform constant C , and dimentional constant A_{1},\delta_{n}>0 . Proof. When \frac{A_{s,{p_{1}}}}{a^{n+1}}\leq 1 , the constant \Lambda_{3} is uniformly bounded from above. Therefore (2.6) \displaystyle A_{s,{p_{1}}}= \displaystyle\left(\int_{M}(\tilde{u}-s)_{+}^{p_{1}}f^{p_{1}}\omega_{M}^{n}\right)^{\frac{1}{{p_{1}}}} \displaystyle\leq \displaystyle C_{n}A_{s,j,{p_{1}}}^{\frac{1}{n+1}}\left(\int_{\Omega_{s}}(-\psi_{3}+\Lambda_{3})^{\frac{n{p_{1}}}{n+1}}f^{{p_{1}}}\omega_{M}^{n}\right)^{\frac{1}{{p_{1}}}}\quad\text{ (for $j$ large enough)} \displaystyle\leq \displaystyle CA_{s,j,{p_{1}}}^{\frac{1}{n+1}}\left(\int_{M}(-\psi_{3}+1)^{\frac{n{p_{1}}p_{2}}{n+1}}f^{({p_{1}}-1)p_{2}+1}\omega_{M}^{n}\right)^{\frac{1}{{p_{1}}p_{2}}}\left(\int_{\Omega_{s}}f\omega_{M}^{n}\right)^{(1-\frac{1}{p_{2}})\frac{1}{p_{1}}} We note that form our construction of f , we have ||f||_{L^{q}}\leq Ce^{\Lambda_{1}} for some uniform constant C . Note that \psi_{3} is uniformly bounded, one can estimate the term (2.7) \displaystyle\left(\int_{M}(-\psi_{3}+1)^{\frac{n{p_{1}}p_{2}}{n+1}}f^{({p_{1}}-1)p_{2}+1}\omega_{M}^{n}\right)^{\frac{1}{{p_{1}}p_{2}}}\leq \displaystyle C\left(\int_{M}f^{(p_{1}-1)p_{2}+1}\omega_{M}^{n}\right)^{\frac{1}{{p_{1}}p_{2}}}\leq Ce^{\frac{\Lambda_{1}((p_{1}-1)p_{2}+1)}{{p_{1}}p_{2}}} Hence equation ( 2.6 ) becomes (2.8)