Global solution curves in harmonic parameters, and multiplicity of solutions
Philip Korman · 2026
Read Global solution curves in harmonic parameters, and multiplicity of solutions by Philip Korman (2026) 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-2601-14581.txt.
Introduction
Philip Korman studies solution curves for a boundary value problem in this 2026 math paper. The problem is a semilinear Dirichlet problem. In it, the Laplacian of u plus a nonlinear term g of u equals a given function f. The value u is zero on the boundary. Korman decomposes f into parts. One part is a multiple of the first eigenfunction. The rest is a remainder term.
His title names the approach: global solution curves in harmonic parameters, and multiplicity of solutions. Multiplicity asks how many solutions exist for one set of data. Readers get a picture of how solutions branch and multiply as the harmonic parameter varies, a core question in nonlinear analysis.
Catalog record
| CREATOR | Philip Korman |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 65 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2601-14581
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Global solution curves in harmonic parameters, and multiplicity of solutions Philip Korman Abstract For the semilinear Dirichlet problem Δ u + g ( u ) = f ( x ) for x ∈ Ω , u = 0 on ∂ Ω \Delta u+g(u)=f(x)\;\;\mbox{for $x\in\Omega$},\;\;u=0\;\;\mbox{on $\partial\Omega$} decompose f ( x ) = μ 1 φ 1 + e ( x ) f(x)=\mu_{1}\varphi_{1}+e(x) , where φ 1 \varphi_{1} is the principal eigenfunction of the Laplacian with zero boundary conditions, and e ( x ) ⟂ φ 1 e(x)\perp\varphi_{1} in L 2 ( Ω ) L^{2}(\Omega) , and similarly write u ( x ) = ξ 1 φ i + U ( x ) u(x)=\xi_{1}\varphi_{i}+U(x) , with U ⟂ φ 1 U\perp\varphi_{1} in L 2 ( Ω ) L^{2}(\Omega) . We study properties of the solution curve ( u ( x ) , μ 1 ) ( ξ 1 ) (u(x),\mu_{1})(\xi_{1}) , and in particular its section μ 1 = μ 1 ( ξ 1 ) \mu_{1}=\mu_{1}(\xi_{1}) , which governs the multiplicity of solutions. We consider both general nonlinearities, and some important classes of equations, and obtain detailed description of solution curves under the assumption g ′ ( u ) < λ 2 g^{\prime}(u)<\lambda_{2} . We obtain particularly detailed results in case of one dimension. This approach is well suited for numerical computations, which we perform to illustrate our results. Key words: Continuation in harmonic parameters, multiplicity of solutions. AMS subject classification: 35J60. 1 Introduction We consider the semilinear Dirichlet problem (1.1) \;\;\;\;\Delta u+g(u)=f(x)\;\;\mbox{for $x\in\Omega$},\;\;u=0\;\;\mbox{on $\partial\Omega$}\,, on a smooth and bounded domain \Omega\in R^{m} . We are interested in the existence, multiplicity and numerical computation of solutions. On the surface the problem ( 1.1 ) does not contain any parameters. It turns out there are parameters (somewhat hidden), which allow one to understand the solution set. These are the harmonics, particularly the first harmonic, of f(x) and of the solution u(x) . Represent the given function f(x)\in L^{2}(\Omega) by its Fourier series f(x)=\mu_{1}\varphi_{1}+\cdots+\mu_{n}\varphi_{n}+e(x) where \varphi_{k} is the k -th eigenfunction of the Laplacian on \Omega with zero boundary conditions, and e(x)\perp\varphi_{k} in L^{2}(\Omega) , k=1,\ldots,n . Likewise decompose the solution in the form u(x)=\Sigma_{i=1}^{n}\xi_{i}\varphi_{i}+U(x) , with U\perp\varphi_{k} , k=1,\ldots,n in L^{2}(\Omega) . In [ 15 ] and [ 17 ] we studied the map \xi=\left(\xi_{1},\xi_{2},\ldots,\xi_{n}\right)\rightarrow\mu=\left(\mu_{1},\mu_{2},\ldots,\mu_{n}\right) which determines the multiplicity of solutions of ( 1.1 ). Indeed, if at some \mu_{0} this map has k pre-images, then the problem ( 1.1 ) has exactly k solutions at \mu=\mu_{0} . In this paper we concentrate on the solution curves , when the parameters \xi_{i} are varied one at a time. Particularly, we shall study the curve \mu_{1}=\mu_{1}(\xi_{1}) , which we call the principal solution curve . We shall also consider the curves \mu_{k}=\mu_{k}(\xi_{k}) for a class of oscillatory problems. We now describe one of our results which demonstrates some of the advantages of working with curves of solutions. For a class of problems (1.2) \;\;\;\;\Delta u+g(u)=\mu_{1}\varphi_{1}+e(x)\;\;\mbox{for $x\in\Omega$},\;\;u=0\;\;\mbox{on $\partial\Omega$}\,, with the solution decomposed as u(x)=\xi_{1}\varphi_{1}+U(x) ( U(x)\perp\varphi_{1} ), we showed the existence of a continuous solution curve (u(x),\mu_{1})(\xi_{1}) which exhausts the solution set of ( 1.2 ) . A section of this curve \mu_{1}=\mu_{1}(\xi_{1}) (which is a planar curve) will determine the multiplicity of solutions. (If the value of \mu_{1}^{0} is achieved at k values of \xi_{1} , the problem ( 1.2 ) has k solutions at \mu_{1}=\mu_{1}^{0} ). Assuming that \frac{g(u)}{u} crosses the first eigenvalue on the interval (-\infty,\infty) (see the Theorem 4.2 below) it was possible to show that \mu_{1}(\xi_{1})\rightarrow\infty as \xi_{1}\rightarrow\pm\infty . It follows that the continuous function \mu_{1}(\xi_{1}) has a global minimum value \mu_{0} , and it takes on at least twice any value in (\mu_{0},\infty) . We conclude that the problem ( 1.2 ) has either zero, at least one, or at least two solutions depending on whether \mu_{1}<\mu_{0} , \mu_{1}=\mu_{0} or \mu_{1}>\mu_{0} respectively, which is an extension of the well known result of H. Amann and P. Hess [ 1 ] , since we do not require the limits \lim_{u\rightarrow\pm\infty}\frac{g(u)}{u} to exist. The classical result of A. Ambrosetti and G. Prodi [ 3 ] , in the form of M.S. Berger and E. Podolak [ 6 ] , also follows along these lines. We derive similar results if \frac{g(u)}{u} crosses the first eigenvalue on the interval (0,\infty) . A major advantage of our approach is in establishing that all solutions of the problem ( 1.2 ) lie on a single solution curve, so that they can be computed by a curve following algorithm. Such algorithms (based on Newton’s method) are very efficient (fast and accurate), and relatively easy to implement. We present some of our numerical computations. We obtained detailed results for two specific classes of equations. The first one involves the following model problem \Delta u+\lambda u-u^{3}=\mu_{1}\varphi_{1}+e(x)\,,\,\;\;\mbox{for $x\in\Omega$}\,,\;\;\;\;u=0\;\;\mbox{on $\partial\Omega$}\,, with a parameter \lambda\in(0,\lambda_{2}) , and \int_{\Omega}e(x)\varphi_{1}(x)\,dx=0 . Decompose u(x)=\xi_{1}\varphi_{1}(x)+U(x) , with \int_{\Omega}U(x)\varphi_{1}(x)\,dx=0 , as above. We show that a typical solution curve \mu_{1}=\mu_{1}(\xi_{1}) is either monotone or S -shaped. This problem has been studied previously, see M.S. Berger et al [ 5 ] and P.T. Church et al [ 9 ] and the references therein, but mostly for the space dimensions n\leq 4 . The second class involves resonant problems (1.3) \Delta u+\lambda_{1}u+g(u)=\mu_{1}\varphi_{1}+e(x)\;\;\mbox{on $\Omega$},\;\;u=0\;\;\mbox{on $\partial\Omega$}\,, with e(x)\in\varphi_{1}^{\perp} in L^{2}(\Omega) . We wish to find a solution pair (u,\mu_{1}) . For bounded g(u) , satisfying ug(u)\geq 0 for all u\in R , and \mu_{1}=0 , D.G. de Figueiredo and W.-M. Ni [ 12 ] have proved the existence of solutions. R. Iannacci, M.N. Nkashama and J.R. Ward [ 13 ] generalized this result to a class of unbounded g(u) . In [ 17 ] we extended the last result to the case \mu_{1}\neq 0 . Compared with our paper [ 17 ] , here we do not try to obtain a bound on \int_{\Omega}|\nabla U|^{2}\,dx uniformly in \xi_{1} in general, but either derive such bounds for specific cases, or work around the issue. In particular, that allowed us to drop the technical condition (3.2) of that paper for the problem ( 1.3 ). The most detailed results are obtained for the one-dimensional case, n=1 . On the interval (0,L) consider the problem (1.4) u^{\prime\prime}+\lambda_{1}\,u+h(u)\sin u=\mu_{1}\varphi_{1}(x)+e(x)\,,\;\;u(0)=u(L)=0\,. Here \lambda_{1}=\frac{\pi^{2}}{L^{2}} , the principal eigenvalue of u^{\prime\prime} on (0,L) corresponding to \varphi_{1}(x)=\sin\frac{\pi}{L}x , \mu_{1}\in R , e(x)\in C(0,L) satisfies \int_{0}^{L}e(x)\sin\frac{\pi}{L}x\,dx=0 . Assume that (1.5) \lim_{u\rightarrow\infty}\frac{h(u)}{u^{p}}=h_{0}\,,\;\;\mbox{with constants $p\in(0,1)$ and $h_{0}>0$}\,. It follows from R. Schaaf and K. Schmitt [ 23 ] that the problem ( 1.4 ) has infinitely many solutions when \mu_{1}=0 . We show that the same is true for all \mu_{1}\in R , provided that p\in(\frac{1}{2},1) . Moreover, we obtain an asymptotic formula \mu_{1}(\xi_{1})\sim\frac{2\sqrt{2}}{\sqrt{\pi\xi_{1}}}\sin\left(\xi_{1}-\frac{\pi}{4}\right)h(\xi_{1}) for large \xi_{1} , where \xi_{1} is the first harmonic of the solution u(x) . Our numerical computations suggest that this formula is very accurate, and that it tends to be accurate for small \xi_{1} as well (see Figure 2 in Section 6 ). In case p\in(0,\frac{1}{2}) we showed that (1.6) u(x)\sim\xi_{1}\sin\frac{\pi}{L}x+E(x)\,,\;\;\mbox{for $\xi_{1}$ large}\,, where E(x) is the unique solution of u^{\prime\prime}+\frac{\pi^{2}}{L^{2}}\,u=e(x)\,,\;\;u(0)=u(L)=0\,,\;\;\int_{0}^{L}u(x)\sin\frac{\pi}{L}x\,dx=0\,. The formula ( 1.6 ) gives an universal