Limit of a class of quasilinear equations and related extremal problems
Yuanlong Ruan · 2025
Read Limit of a class of quasilinear equations and related extremal problems by Yuanlong Ruan (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-2503-06126.txt.
Introduction
Ruan, a solo author, studies the limit of a class of equations in a 2025 mathematics paper. The problems carry fixed values at their edges. The author gives a full account of how the solutions and their energies behave as the limit is approached.
One number from the edge data, a bound on steepness, controls how the natural energies settle. Yet the solutions themselves settle to a limit equation that does not depend on that number. The limit equation lacks a standard grip on its solution, so care is needed. Ruan proves a comparison rule and a single solution of viscosity type. A linked extreme problem with an unbounded operator is also solved, with one solution that varies smoothly.
Catalog record
| CREATOR | Yuanlong Ruan |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 171 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2503-06126
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Limit of a class of quasilinear equations and related extremal problems Yuanlong Ruan Abstract . We perform a complete analysis of the limiting behaviour of a class of quasilinear problems with Dirichlet boundary data g g , − div 𝒜 n ( x , ∇ u ( x ) ) = 0 , x ∈ Ω . -\operatorname{div}\mathcal{A}_{n}\left(x,\nabla u\left(x\right)\right)=0,\text{ }x\in\Omega. We show that the Lipschitz constant of g g plays a role in controlling the Γ \Gamma -convergence of the natural energies. However the solutions converge uniformly to solution of a limiting equation whenever g ∈ L i p ( ∂ Ω ) g\in Lip(\partial\Omega) , irrelevant to the Lipschitz constant of g g . The limiting equation has no coercivity in u u . We prove that the limiting equation admits a weak comparison principle and has a unique viscosity solution. We also obtain a Poincare’s inequality in the Sobolev-Orlicz space for discontinuous 𝒜 n \mathcal{A}_{n} , which paves the way for our study of an extremal problem where its operator 𝒜 ∞ \mathcal{A}_{\infty} becomes unbounded in a subdomain Ω ∞ ⊂ Ω \Omega_{\infty}\subset\Omega . Upon giving proper meaning to its solution, we show that the extremal problem has a unique solution. It turns out the solution belongs to C ( Ω ¯ ) ∩ W 1 , ∞ ( Ω ) C\left(\bar{\Omega}\right)\cap W^{1,\infty}\left(\Omega\right) , although 𝒜 ∞ \mathcal{A}_{\infty} is discontinuous. In the appendix we provide some technical inequalities which play crucial roles in the proof of uniqueness and we believe will be of independent interest. 2010 Mathematics Subject Classification Primary 35A02, 35A15, 35D40 1. Introduction We study the limiting behaviour as n\rightarrow\infty of a class of quasilinear problems in the nonlinear potential theory (1.1) -\operatorname{div}\mathcal{A}_{n}\left(x,\nabla u\left(x\right)\right)=0,\text{ }x\in\Omega, and a related extremal problem (1.2) -\operatorname{div}\mathcal{A}_{\infty}\left(x,\nabla u\left(x\right)\right)=0,\text{ }x\in\Omega, where \mathcal{A}_{\infty} is unbounded in a subdomain of \Omega . These quasilinear problems have been the prototype of many researches. The most studied one being the p -Laplacian for large p>1 , -\Delta_{p}u=\operatorname{div}\left(\left|\nabla u\right|^{p-2}\nabla u\right)=0,\text{ }x\in\Omega. Early studies in this respect dates back to 1980s, e.g. the torsional creep problem [ 4 ] where the main interest lies in the properties of solutions when p is large. An interesting result is that for large p , the p -Laplacian equation is close to the infinity Laplacian equation, -\Delta_{\infty}u=0,\text{ }x\in\Omega. This observation is also exploited in the study of sandpiles [ 2 ] [ 11 ] where p -Laplacian with large p is used as a simplistic approximation of collapsing sandpile. The asymptotic analysis for the p -Laplacian was later extended to the p\left(x\right) -Laplacian [ 24 ] . When p\left(x\right) is unbounded in a subdomain D\subset\Omega [ 25 ] , the p\left(x\right) -harmonic function is related to the infinity Laplacian equation subject to a mixed Dirichlet-Neumann boundary condition. In the subdmain D , the mixed Dirichlet-Neumann problem \left\{\begin{array}[c]{ll}-\Delta_{\infty}u=0,&x\in D;\\ u=g,&x\in\partial D;\\ \dfrac{\partial u}{\partial\nu}=0,&x\in\partial\Omega\backslash\partial D,\end{array}\right. coincides with the characterization of the tug-of-war game [ 26 ] [ 6 ] . In addition to the p\left(x\right) -Laplacian, limit of anisotropic p -Laplacian [ 27 ] and isotropic \varphi -Laplacian [ 28 ] were also considered. Another thread of studies on limiting behaviour of \left(\ref{prototype_quasilinear}\right) is mostly variational and involves the convergence of the associated energies, e.g. [ 5 ] . Motivated by these works, we intend to present a complete study of \left(\ref{prototype_quasilinear_n}\right) and the related extremal problem \left(\ref{prototype_quasilinear}\right) , from the perspectives of both partial differential equations and variational convergences. The quasilinear problems \left(\ref{prototype_quasilinear_n}\right) and the extremal problem \left(\ref{prototype_quasilinear}\right) encompass the p\left(x\right) -Laplacian problem considered in [ 25 ] . However, as noted in [ 15 ] , going from the variable exponent case to the general case requires new techniques on the level of both function spaces and partial differential equations. There are also some new features here. The general structure of the equation \left(\ref{prototype_quasilinear_n}\right) allows complex interplay between the x -variable and the gradient \nabla u\left(x\right) . Many properties and results present for the case of p\left(x\right) -Laplacian are not available for \left(\ref{prototype_quasilinear_n}\right) . Most prominently, several new building blocks, particularly the uniqueness for the limiting equation, some subtle estimates and the embedding in the generalized Sobolev-Orlicz space with discontinuous coefficients \mathcal{A}_{\infty} have to be built in the course of solving \left(\ref{prototype_quasilinear}\right) . We first give a rough sketch of our approaches and results. Here in this article, we do not try to pursue full generality, so we restrict ourselves to quasilinear problems of the following form (1.3) \left\{\begin{array}[c]{ll}-\operatorname{div}\left(\dfrac{a_{n}\left(x,\left|\nabla u\left(x\right)\right|\right)}{\left|\nabla u\left(x\right)\right|}\nabla u\left(x\right)\right)=0,&x\in\Omega;\\ u\left(x\right)=g\left(x\right),&x\in\partial\Omega,\end{array}\right. where \Omega is a bounded smooth domains of \mathbb{R}^{d} , g\in Lip\left(\partial\Omega\right) and a_{n}\left(x,s\right)=\frac{\varphi_{n}\left(x,s\right)}{\varphi_{n}\left(x,1\right)}. The assumptions on \varphi_{n} will be made clear in section 3 (ref. \left(\ref{assumption_phi_x}\right) ). The natural energies associated with this equation is denoted by E_{n}\left(u\right):u\in W^{1,\Phi_{n}\left(\cdot\right)}\left(\Omega\right)\mapsto{\displaystyle\int_{\Omega}}\dfrac{\Phi_{n}\left(x,\left|\nabla u\right|\right)}{\varphi_{n}\left(x,1\right)}dx, where \Phi_{n}\left(x,s\right)=\int_{0}^{s}\varphi_{n}\left(x,t\right)dt. We intend to let n\rightarrow\infty and get limits of solutions u_{n}\ of \left(\ref{general_quasilinear_n_intro}\right) , so require some estimates independent of n . A key ingredient is a uniform gradient estimate for the weak solutions of \left(\ref{general_quasilinear_n_intro}\right) . In obtaining the uniform gradient estimate, the boundary data g seems to play a role. The first evidence for g ’s effects is revealed by the \Gamma -convergence of the extended natural energies \tilde{E}_{n}\left(u\right) associated with \left(\ref{general_quasilinear_n_intro}\right), \tilde{E}_{n}\left(u\right):u\in L^{1}\left(\Omega\right)\mapsto\left\{\begin{array}[c]{ll}{\displaystyle\int_{\Omega}}\dfrac{\Phi_{n}\left(x,\left|\nabla u\right|\right)}{\varphi_{n}\left(x,1\right)}dx,&u\in W^{1,\Phi_{n}\left(\cdot\right)}\left(\Omega\right);\\ \infty,&\text{otherwise.}\end{array}\right. Hereafter when there is no confusion, we will use the same symbol E_{n} to denote the natural energy on W^{1,\Phi_{n}\left(\cdot\right)}\left(\Omega\right) and its extension to L^{1}\left(\Omega\right) . The \Gamma -convergence result suggests that, in order for E_{n} to be bounded, we assume (1.4) \left\|g\right\|_{Lip\left(\partial\Omega\right)}\leqslant 1. Here \left\|g\right\|_{Lip\left(\partial\Omega\right)} is the Lipschitz constant of g\in Lip\left(\partial\Omega\right) , i.e., \left\|g\right\|_{Lip\left(\partial\Omega\right)}\triangleq\sup_{\begin{subarray}{c}x,y\in\partial\Omega\\ x\neq y\end{subarray}}\frac{\left|g\left(x\right)-g\left(y\right)\right|}{\left|x-y\right|}. Previous studies for p\left(x\right) -Laplacian are mostly restricted to the assumption \left(\ref{assumption_g_lip}\right) , e.g. [ 25 ] [ 24 ] . In fact we show that, for convex domains, \left(\ref{assumption_g_lip}\right) is a necessary and sufficient condition for E_{n} to be bounded. We also show that the required uniform gradient estimate can be obtained without the assumption \left\|g\right\|_{Lip\left(\partial\Omega\right)}\leqslant 1 , although \left\|g\right\|_{Lip\left(\partial\Omega\right)}>1 might cause the energies E_{n} to explode as n\rightarrow\infty . Once we get the required estimates, the convergence of solutions can be proved and the limiting equation for \left(\ref{general_quasilinear_n_intro}\right) turns out to be (1.5) \left\{\begin{array}[c]{ll}-\Delta_{\infty}u-\left|\nabla u\right|\left\langle\Lambda\left(x,\left|\nabla u\right|\right),\nabla u\right\rangle=0,&x\in\Omega;\\ u=g,&x\in\partial\Omega.\end{array}\right. Now here is one difficulty. Since this equation is not strictly coercive in u , uniqueness for such equation is usually difficult to obtain. The classical comparison principle does not readily apply and the perturbation technique suggested by the User’s guide [ 7 ] does not seem to work either. So we have to prove a comparison principle for this limiting equation. In the course of proving the uniqueness, some technical inequalities have a crucial role to play. Once the uniqueness is obtained, the convergence of the solutions for \left(\ref{general_quasilinear_n_intro}\right) follows. Next we look into the extremal equation (1.6) \left\{\begin{array}[c]{ll}-\operatorname{div}\left(\dfrac{a_{\infty}\left(x,\left|\nabla u\left(x\right)\right|\right)}{\left|\nabla u\left(x\right)\right|}\nabla u\left(x\right)\right)=0,&x\in\Omega;\\ u\left(x\right)=g\left(x\right),&x\in\partial\Omega,\end{array}\right. where a_{\infty}\left(x,s\right):\Omega\times\left[0,\infty\right)\rightarrow\mathbb{R\cup}\left\{+\infty\right\} is defined piecewise as below. Let \Omega_{\infty} be a smooth subdomain of \Omega , a\left(x,s\right) an appropriate function on \left(\Omega\backslash\Omega_{\infty}\right)\times\left[0,\infty\right) . We write (1.7) a_{\infty}\left(x,s\right)=\left\{\begin{array}[c]{ll}0,&s\leqslant 1,x\in\Omega_{\infty};\\ \infty,&s>1,x\in\Omega_{\infty}.\end{array}\right. and (1.8) a_{\infty}\left(x,s\right)=a\left(x,s\right)\text{ for }x\in\Omega\backslash\Omega_{\infty},\text{ }s\geqslant 0. The basic question about understanding \left(\ref{equ_extremal_intro}\right) is whether it has a solution and in what sense. If it does have a solution in