Solutions for autonomous semilinear elliptic equations

Alexis Molino, Salvador Villegas · 2025

Read Solutions for autonomous semilinear elliptic equations by Alexis Molino, Salvador Villegas (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-2504-18877.txt.

Introduction

Alexis Molino and Salvador Villegas study a classic partial differential equation in a 2025 paper. Their problem couples diffusion with a linear term and a nonlinear reaction. They ask for solutions that vanish at the boundary of a smooth region. The reaction function is only assumed to be locally smooth, with a primitive that never goes positive. For one dimension, they describe the full solution set. For higher dimensions, they prove partial results.

Catalog record

CREATORAlexis Molino, Salvador Villegas
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE9 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2504-18877

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Solutions for autonomous semilinear elliptic equations Alexis Molino, Salvador Villegas Abstract . We study existence of nontrivial solutions to problem { − Δ ​ u = λ ​ u + f ⁡ ( u ) in ​ Ω , u = 0 on ​ ∂ Ω , \left\{\begin{array}[]{rcll}-\Delta u&=&\lambda u+f(u)&\text{ in }\Omega,\\ u&=&0&\text{ on }\partial\Omega,\end{array}\right. where Ω ⊂ ℝ N \Omega\subset\mathbb{R}^{N} is a smooth bounded domain, N ≥ 1 N\geq 1 , λ ∈ ℝ \lambda\in\mathbb{R} and f : ℝ → ℝ f:\mathbb{R}\to\mathbb{R} is any locally Lipschitz function with nonpositive primitive. A complete description is obtained for N = 1 N=1 and partial results for N ≥ 2 N\geq 2 . † † footnotetext: MSC : 35J05, 35J15, 35J25. Keywords: Elliptic Equation, Dirichlet Problem, Existence. 1. Introduction In this paper we deal about the existence of nontrivial solutions for the following autonomous semilinear elliptic differential equation ( P ) \left\{\begin{array}[]{rcll}-\Delta u&=&\lambda u+f(u)&\text{ in }\Omega,\\ u&=&0&\text{ on }\partial\Omega,\end{array}\right. being \Omega\subset\mathbb{R}^{N} a smooth bounded domain, N\geq 1 , \lambda\in\mathbb{R} and f:\mathbb{R}\to\mathbb{R} a locally Lipschitz function satisfying ( H ) F(s):=\int_{0}^{s}f(t)dt\leq 0\mbox{ for every }s\in\mathbb{R}. Typical example are f(u)=-\sin u or f(u)=-u(u-1)(u-2) A bounded weak solution (solution from now on) to problem ( P ) is a function u\in H_{0}^{1}(\Omega)\cap L^{\infty}(\Omega) such that \int_{\Omega}\nabla u\nabla\varphi=\lambda\int_{\Omega}u\varphi+\int_{\Omega}f(u)\varphi,\quad\textit{for all}\,\,\varphi\in H_{0}^{1}(\Omega). Observe that, by regularity results, every solution is a classical solution to this problem. Note that hypothesis ( H ) implies that u\equiv 0 is always a solution to problem ( P ). In [ 2 ] the authors establish the uniqueness of trivial solution for the case \lambda=0 . Therefore, we are interested in existence of nontrivial solutions by adding the term \lambda u to the source data f(u) . As will be seen below, this will depend both on the dimension of the space and on the different values of \lambda . Besides, \lambda_{1} (the first eigenvalue of the Laplace operator with Dirichlet boundary conditions on \partial\Omega ) plays a crucial role. That is, an effect similar to that found in the well-known Brézis-Nirenberg work [ 1 ] . However, as can be seen in the next section, partial results will be obtained, leaving open questions reflected in the last section. 2. Main results Lemma 2.1 . Let N\geq 1 . If \lambda\leq 0 , then u\equiv 0 is the unique solution to ( P ), while for \lambda\geq\lambda_{1} there exists f satisfying ( H ) such that there exists a nontrivial solution to ( P ). Proof. Case \lambda\leq 0 is a fairly straightforward result of [ 2 ] , since the term \lambda u+f(u) has nonpositive primitive. On the other hand, for case \lambda\geq\lambda_{1} it is sufficient to take f(u)=(\lambda_{1}-\lambda)u and observe that u=\varphi_{1} (the first eigenfunction of the Laplace operator with Dirichlet boundary conditions on \partial\Omega ) is a classical solution to ( P ). ∎ The previous lemma focuses the problem for values of \lambda\in(0,\lambda_{1}) . Next, a partial result is obtained for dimensions greater than 2 . Theorem 2.2 . Let N\geq 3 and \Omega\subset\mathbb{R}^{N} a smooth bounded starshaped domain and \lambda\leq\frac{N-2}{N}\lambda_{1} . Then, problem ( P ) has no nontrivial solutions. Proof. Without loss of generality we can assume, up to a translation, that \Omega is starshaped respect to the origin. Therefore x\cdot\nu(x)\geq 0 on \partial\Omega , where \nu denotes the unit vector normal to \partial\Omega pointing outwards. By the well-known Pohožaev identity [ 3 ] , every solution u to ( P ) must satisfy \displaystyle\frac{1}{2}\int_{\partial\Omega}|\nabla u(x)|^{2}\,x\cdot\nu(x)\,dx+\frac{N-2}{2}\int_{\Omega}|\nabla u(x)|^{2}\,dx \displaystyle=\lambda\,\frac{N}{2}\int_{\Omega}u(x)^{2}\,dx+N\int_{\Omega}F(u(x))\,dx, which implies \frac{N-2}{2}\int_{\Omega}|\nabla u(x)|^{2}\,dx\leq\lambda\,\frac{N}{2}\int_{\Omega}u(x)^{2}\,dx+N\int_{\Omega}F(u(x))\,dx, and by using Poincaré inequality \lambda_{1}\,\frac{N-2}{2}\int_{\Omega}u(x)^{2}\,dx\leq\lambda\,\frac{N}{2}\int_{\Omega}u(x)^{2}+N\int_{\Omega}F(u(x))\,dx. We claim that \int_{\Omega}F(u(x))\,dx<0 . Conversely, since f satisfies ( H ) and F\circ u is continuous, then F(u(x))=0 for every x\in\Omega . Therefore F(s)=0 for every s\in I=\{u(x):x\in\Omega\} . Since u is nontrivial it follows that I is a nontrivial interval and hence f(s)=F^{\prime}(s)=0 for every s\in I . Thus u satisfies -\Delta u=\lambda u in \Omega and consequently \lambda is an eigenvalue of the Laplace operator with Dirichlet boundary conditions on \partial\Omega , contradicting hypothesis \lambda\leq\frac{N-2}{N}\lambda_{1} . Finally, from the above inequality it follows that \lambda_{1}\,\frac{N-2}{2}\int_{\Omega}u(x)^{2}\,dx<\lambda\,\frac{N}{2}\int_{\Omega}u(x)^{2}\,dx. This leads to the fact that any nontrivial solution to the problem ( P ) must satisfy \lambda>\lambda_{1}\,\frac{N-2}{N} and the proof is concluded. ∎ However, as can be seen in the following theorem, in dimension 1 a drastic change occurs: the existence of nontrivial solutions for every positive \lambda . Theorem 2.3 . Let \Omega=(a,b)\subset\mathbb{R} and set \lambda>0 . Then, there exists f\in C^{\infty}(\mathbb{R}) satisfying ( H ), such that problem ( P ) admits a nontrivial solution u\in C^{\infty}[a,b] . For its proof, we will first state the following technical lemma: Lemma 2.4 . Let a<b and \lambda real numbers satisfying 0<\lambda<\pi^{2}/(b-a)^{2} . Then, there exists a unique M>0 , depending on \lambda , such that \int_{0}^{1}\frac{ds}{\sqrt{Ms^{2}(1-s)^{2}+\lambda(1-s^{2})}}=\frac{b-a}{2}. Proof. Define the function \Phi:(0,+\infty)\to\mathbb{R} by \Phi(k):=\int_{0}^{1}\frac{ds}{\sqrt{ks^{2}(1-s)^{2}+\lambda(1-s^{2})}},\ \ \ k>0. It is obvious that \frac{1}{\sqrt{ks^{2}(1-s)^{2}+\lambda(1-s^{2})}}\leq\frac{1}{\sqrt{\lambda(1-s^{2})}}\in L^{1}(0,1)\mbox{ for every }k>0. Hence, applying the Convergence Dominated Theorem we can assert that \Phi is a well defined decreasing continuous function satisfying \lim_{k\to+\infty}\Phi(k)=\int_{0}^{1}0\,ds=0, \lim_{k\to 0}\Phi(k)=\int_{0}^{1}\frac{ds}{\sqrt{\lambda(1-s^{2})}}=\frac{\pi}{2\sqrt{\lambda}}. Applying 0<\lambda<\pi^{2}/(b-a)^{2} , we deduce that \frac{b-a}{2}\in\left(\lim_{k\to+\infty}\Phi(k),\lim_{k\to 0}\Phi(k)\right), which gives \Phi(M)=(b-a)/2 for a unique M>0 , and the proof is complete. ∎ Proof of Theorem 2.3 If \lambda\geq\lambda_{1}=\pi^{2}/(b-a)^{2} , by Lemma 2.1 there exists a nontrivial solution to problem ( P ) for f(s)=\left(\pi^{2}/(b-a)^{2}-\lambda\right)s,\,\,s\in\mathbb{R} . For the rest of the proof we will suppose that 0<\lambda<\pi^{2}/(b-a)^{2} . Define the function f:\mathbb{R}\to\mathbb{R} by f(s)=-Ms(s-1)(2s-1),\ \ \ s\in\mathbb{R}, where M>0 is the value obtained in Lemma 2.4 . It is immediate that f\in C^{\infty}(\mathbb{R}) and satisfies F(s):=\int_{0}^{s}f(t)dt=-Ms^{2}(1-s)^{2}/2\leq 0 , for every s\in\mathbb{R} . Now, we will construct a nontrivial solution u\in C^{\infty}(a,b) to problem ( P ). For this purpose; define \Psi:[0,1]\to[a,(a+b)/2] by \Psi(t):=a+\int_{0}^{t}\frac{ds}{\sqrt{Ms^{2}(1-s)^{2}+\lambda(1-s^{2})}},\ \ \ t\in[0,1]. Clearly \Psi is a C^{\infty}[0,1)\cap C[0,1] bijection between [0,1] and [a,(a+b)/2] , with \Psi^{\prime}(t)=1/\sqrt{Mt^{2}(1-t)^{2}+\lambda(1-t^{2})}>0 , t\in[0,1). Therefore, we can define u:=\Psi^{-1} , obtaining u\in C^{\infty}[a,(a+b)/2)\cap C[a,(a+b)/2] with u^{\prime}(x)=\sqrt{Mu(x)^{2}(1-u(x))^{2}+\lambda(1-u(x)^{2})},\ \ \ x\in[a,(a+b)/2). Since u\left((a+b)/2\right)=1 , we obtain that \lim_{x\to(a+b)/2}u^{\prime}(x)=0 and consequently u\in C^{1}[a,(a+b)/2] . Moreover, from the above expression we can assert that u^{\prime}(x)^{2}=Mu(x)^{2}(1-u(x))^{2}+\lambda(1-u(x)^{2}),\ \ \ x\in[a,(a+b)/2), and differentiating yields 2u^{\prime}(x)u^{\prime\prime}(x)=u^{\prime}(x)\left(2Mu(x)(u(x)-1)(2u(x)-1)-2\lambda u(x)\right),\ \ \ x\in[a,(a+b)/2). Since u^{\prime}(x)\neq 0 for x\in[a,(a+b)/2) we conclude that -u^{\prime\prime}=f(u)+\lambda u,\,\,x\in[a,(a+b)/2). In fact, the continuity of f implies that there exists \lim_{x\to(a+b)/2}u^{\prime\prime}(x) , and consequently u\in C^{2}[a,(a+b)/2] . Finally, we define u for x\in((a+b)/2,b] by symmetry. That is, u(x):=u(a+b-x),\,\,x\in((a+b)/2,b]. Taking into account that u(a)=0 , u^{\prime}((a+b)/2)=0 and u\in C^{2}[a,(a+b)/2] . It follows that -u^{\prime\prime}=f(u)+\lambda u in [a,(a+b)/2] . So, we deduce that u\in C^{2}[a,b] is a solution of problem ( P ). Finally, since f\in C^{\infty}(\mathbb{R}) we conclude u\in C^{\infty}[a,b] and the result follows. ∎ 3. Summary and open problems Our results are summarized as follows. Uniqueness/non-uniqueness of trivial solution to problem ( P ). • N=1 ; Uniqueness for \lambda\leq 0 and non-uniqueness for \lambda>0 . • N=2 ; Uniqueness for \lambda\leq 0 and non-uniqueness for \lambda\geq\lambda_{1} . • N\geq 3 ; Uniqueness for \lambda\leq\lambda_{1}\frac{N-2}{N} (in case that the domain is starshaped) and non-uniqueness for \lambda\geq\lambda_{1} . We finish our discussion with the following open questions: (1) Does ( P ) admit a nontrivial solution if N=2 and 0<\lambda<\lambda_{1} ? (2) Does ( P ) admit a nontrivial solution if N\geq 3

Read the complete text