Infinitely many solutions for a class of resonant problems
Philip Korman · 2025
Read Infinitely many solutions for a class of resonant problems by Philip Korman (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-2512-18562.txt.
Introduction
Philip Korman studies a resonant equation on a unit ball, where a vibrating system is driven at its own natural frequency. The equation couples the Laplacian of u with the principal eigenvalue term and a periodic, mean-zero nonlinearity g(u). Radial solutions are the focus of his paper, posted to arXiv in December 2025 and published in Nonlinear Differential Equations and Applications.
The headline result is dimensional: whether the problem has infinitely many solutions or only finitely many depends on the space dimension n. The pattern is not uniform. Dimensions one through three behave one way, while four, five, and six each differ, and seven and above form another case. Such dimension-by-dimension maps matter in nonlinear analysis, where resonance can flood a problem with solutions or shut them down.
Catalog record
| CREATOR | Philip Korman |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 42 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2512-18562
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Infinitely many solutions for a class of resonant problems Philip Korman Abstract We consider radially symmetric solutions for a class of resonant problems on a unit ball B ⊂ R n B\subset R^{n} around the origin Δ u + λ 1 u + g ( u ) = f ( r ) for x ∈ B , u = 0 on ∂ B . \Delta u+\lambda_{1}u+g(u)=f(r)\;\;\mbox{for $x\in B$},\;\;u=0\;\;\mbox{on $\partial B$}\,. Here the function g ( u ) g(u) is periodic of mean zero, x ∈ R n x\in R^{n} , r = | x | r=|x| , λ 1 \lambda_{1} is the principal eigenvalue of Δ \Delta on B B . The problem has either infinitely many or finitely many solutions depending on the space dimension n n . The situation turns out to be different for each of the following cases: 1 ≤ n ≤ 3 1\leq n\leq 3 , n = 4 n=4 , n = 5 n=5 , n = 6 n=6 , and n ≥ 7 n\geq 7 . Key words: Resonant nonlinear problems, solution curves, oscillatory integrals. AMS subject classification: 35J61, 34B15. 1 Introduction Consider a resonant semilinear problem (1.1) \;\;\;\;\;\;\Delta u+\lambda_{1}u+g(u)=\mu_{1}\varphi_{1}(x)+e(x)\;\;\mbox{for $x\in D$},\;\;u=0\;\;\mbox{on $\partial D$}\,, on a bounded domain D\subset R^{n} . Here x\in R^{n} , and (\lambda_{1},\varphi_{1}(x)) is the principal eigenpair of the Laplacian on D , with zero boundary conditions, \mu_{1}\in R , e(x)\in\varphi_{1}^{\perp} , where \varphi_{1}^{\perp} denotes the orthogonal complement of \varphi_{1}(x) in L^{2}(D) . Solutions of ( 3.1 ) are also decomposed as u(x)=\xi_{1}\varphi_{1}(x)+U(x) , with U(x)\in\varphi_{1}^{\perp} . This problem is at resonance , which is one of the basic concepts of science and engineering. The study of resonance for elliptic boundary value problem was originated in the classical papers of E.M. Landesman and A.C. Lazer [ 10 ] , and A.C. Lazer and D.E. Leach [ 11 ] . There is now a huge literature on this topic. In this paper we use a novel approach, based on the theory of global solution curves that are parameterized by the first harmonic of the solution, that we developed in [ 7 ] . Another new ingredient is the use of the stationary phase method to study oscillating integrals. Our approach is uniquely suitable for numerical computations, that we perform and present below. The Mathematica program was written jointly with D.S. Schmidt, and it is presented with detailed explanations in [ 9 ] . In [ 7 ] we gave conditions (see ( 3.1 ), ( 3.2 ) below) under which the solution set of ( 1.1 ) is exhausted by a single continuous solution curve (u(x),\mu)(\xi_{1}) with the first harmonic of the solution, \xi_{1} , acting as a global parameter . Namely, for each \xi_{1}\in R there is a unique solution pair (\lambda_{1},\varphi_{1}(x)) . A section of this curve, \mu_{1}=\mu_{1}(\xi_{1}) , governs the multiplicity of solutions. In particular, if \mu_{1}(\xi_{1}) has infinitely many roots, then the problem (1.2) \;\;\;\;\;\;\Delta u+\lambda_{1}u+g(u)=e(x)\;\;\mbox{for $x\in D$},\;\;u=0\;\;\mbox{on $\partial D$} has infinitely many solutions. For periodic g(u) of mean zero (like g(u)=\sin u ) a very detailed result was obtained in R. Schaaf and K. Schmitt [ 13 ] . They defined g_{1}(u),g_{2}(u),g_{3}(u) to be the unique periodic functions of mean zero, so that g^{\prime}_{1}(u)=g(u) , g^{\prime}_{2}(u)=g_{1}(u) , g^{\prime}_{3}(u)=g_{2}(u) , and showed that the multiplicity depends on the dimension n as follows. For 1\leq n\leq 3 , the problem ( 1.2 ) has infinitely many solutions. For n=4 , a condition on g_{2}(u) was given for the existence of infinitely many solutions, and a complementary condition was provided under which the number of solutions is finite. For n\geq 5 and g_{2}(0)\neq 0 , the number of solutions of ( 1.2 ) was proved to be finite. Only the case n\geq 5 and g_{2}(0)=0 was left open. (That includes g(u)=\sin u , with g_{2}(u)=-\sin u , g_{2}(0)=0 .) In this paper we investigate the case g_{2}(0)=0 for a special class of problems involving radial solutions on a ball around the origin D=B=\left\{x\in R^{n}\;{\rm with}\;||x||<1\right\} . It turned out that for n=5 and g_{2}(0)=0 the number of solutions is infinite, while for n\geq 6 and g_{2}(0)=0 the number of solutions depends on g_{3}(0) . (We also rederive the results of [ 13 ] for the radial case by a different method.) Unlike [ 13 ] , we use the stationary phase method to study oscillating integrals. The use of global solution curves described above (rather than continuum of solutions as in [ 13 ] ) allowed us to conclude the radial symmetry of solutions of ( 1.1 ) on B , and also that the problem ( 1.1 ) has no solutions for |\mu_{1}| large (global solution curves also provide a basis for numerical computations that we present). Once it is established that solutions are radially symmetric on B , we proceed similarly to A. Galstyan et al [ 4 ] . In the study of oscillating integrals we perform up to three integrations by parts, depending on the dimension n , and then use the stationary phase method. We begin with g(u)=\sin u , and then generalize. More general results, without requiring g(u) to be periodic, are obtained in the one-dimensional case, by using geometrical arguments instead of stationary phase method. 2 Radially symmetric oscillatory integrals We study oscillating integrals of the form (2.1) I(\xi)=\int_{0}^{1}g(\xi v(r))\varphi(r)r^{n-1}\,dr\,, depending on a parameter \xi . Various choices of the functions v(r) and \varphi(r) will be considered, beginning with v(r)=\varphi(r)=\varphi_{1}(r) , the principal eigenfunction of the Laplacian on the unit ball B , with \varphi_{1}=0 on \partial B . We assume that g(u) is a periodic function of mean zero, which implies that g(u) changes sign infinitely many times, and the issue is whether I(\xi) changes sign infinitely many times, as \xi\rightarrow\infty . It turns out that the answer depends on the dimension n . Depending on the dimension n , we shall need to perform up to three integrations by parts for I(\xi) . Following [ 13 ] , define g_{1}(u),g_{2}(u),g_{3}(u) to be the unique periodic functions of mean zero, such that g^{\prime}_{1}(u)=g(u) , g^{\prime}_{2}(u)=g_{1}(u) , g^{\prime}_{3}(u)=g_{2}(u) . (In case g(u)=\sin u , g_{1}(u)=-\cos u , g_{2}(u)=-\sin u , g_{3}(u)=\cos u .) Denoting f_{1}(r)=\frac{\varphi(r)r^{n-1}}{v^{\prime}(r)} , obtain (2.2) \displaystyle I(\xi)=\frac{1}{\xi}\int_{0}^{1}\frac{\varphi(r)r^{n-1}}{v^{\prime}(r)}d\left(g_{1}(\xi v(r))\right)=\frac{1}{\xi}\int_{0}^{1}f_{1}(r)d\left(g_{1}(\xi v(r))\right) \displaystyle=\frac{1}{\xi}f_{1}(r)g_{1}(\xi v(r)){\Large|}_{{}_{0}}^{{}^{1}}-\frac{1}{\xi}\int_{0}^{1}f^{\prime}_{1}(r)g_{1}(\xi v(r))\,dr\,. Writing g_{1}(\xi v)=\frac{1}{\xi v^{\prime}}\frac{d}{dr}\left(g_{2}(\xi v(r))\right) , and denoting f_{2}(r)=\frac{f^{\prime}_{1}(r)}{v^{\prime}(r)} , integrate by parts again to get (2.3) \displaystyle I(\xi)=\frac{1}{\xi}f_{1}(r)g_{1}(\xi v(r)){\Large|}_{{}_{0}}^{{}^{1}}-\frac{1}{\xi^{2}}f_{2}(r)g_{2}(\xi v(r)){\Large|}_{{}_{0}}^{{}^{1}} \displaystyle+\frac{1}{\xi^{2}}\int_{0}^{1}f_{2}^{\prime}(r)g_{2}(\xi v(r))\,dr\,. Denoting f_{3}(r)=\frac{f^{\prime}_{2}(r)}{v^{\prime}(r)} , and writing g_{2}(\xi v)=\frac{1}{\xi v^{\prime}}\frac{d}{dr}\left(g_{3}(\xi v(r))\right) , integrate by parts once more to get (2.4) \displaystyle I(\xi)=\frac{1}{\xi}f_{1}(r)g_{1}(\xi v(r)){\Large|}_{{}_{0}}^{{}^{1}}-\frac{1}{\xi^{2}}f_{2}(r)g_{2}(\xi v(r)){\Large|}_{{}_{0}}^{{}^{1}} \displaystyle+\frac{1}{\xi^{3}}f_{3}(r)g_{3}(\xi v(r)){\Large|}_{{}_{0}}^{{}^{1}}-\frac{1}{\xi^{3}}\int_{0}^{1}f_{3}^{\prime}(r)g_{3}(\xi v(r))\,dr\,. We shall use the following lemma, based on the stationary phase method, see e.g., A. Galstian et al [ 4 ] or P. Korman [ 6 ] . Lemma 2.1 Assume that the functions f(x) and \varphi(x)>0 are of class C^{2}[0,1] , and satisfy \varphi^{\prime}(x)<0\quad{\rm for\,all}\quad x\in(0,1],\quad{\rm and}\quad\varphi^{\prime}(0)=0,\,\,\varphi^{\prime\prime}(0)<0\,. Then, as \xi\to\infty , \int_{0}^{1}f(x)e^{i\xi\varphi(x)}dx=e^{i(\xi\varphi(0)-\frac{\pi}{4})}\sqrt{\frac{\pi}{2\xi|\varphi^{\prime\prime}(0)|}}f(0)+O\left(\frac{1}{\xi}\right)\,. The proof of this lemma can be found in e.g., [ 4 ] , however the proof is sketched next, since a similar idea is used later on to prove a more general result. The term e^{i\xi\varphi(x)} involves fast oscillations about zero, which are mutually cancelling, except near