Boundedness in asymmetric oscillations under the non-resonant case (Q2217327)

In the paper, the authors concerned with the boundedness of all solutions of the equation \[ x''+ax^+-bx^-=p(t),\qquad a,b\in{\mathbb R}^+, \qquad (a\ne b) \tag{1}\] where \(x^\pm=\max\{\pm x,0\}\), \(p(t)\) is a real-valued analytic \(2\pi\)-periodic function, \(a\) and \(b\) satisfy \[ \omega_0=(1/\sqrt{a}+1/\sqrt{b})/2\in{\mathbb I} \] and the condition \[ |k\omega_0-l|\geq\frac{c_0}{\Omega(|k|)},\qquad k,l\in{\mathbb Z},\quad (k\ne0) \tag{2}\] where, \(c_0\) is a small positive constant, and that \({\mathbb I}\) is the set of irrational numbers, i.e. \({\mathbb I}={\mathbb Z}\setminus{\mathbb Q}\). \(\Omega(\cdot)\) in (2) is an approximation function (see [\textit{H. Rüssmann}, ``On the one-dimensional Schrödinger equation with a quasi-periodic potential'', Ann. N.Y. Acad. Sci. 357 90--107 (1980; \url{doi:10.1111/j.1749-6632.1980.tb29679.x})]), i.e., a non-decreasing function \(\Omega:[1,\infty)\to[1,\infty)\) satisfying \[ \frac{\log\Omega(t)}{t}\to0,\qquad1\leq t\to\infty \] and \[ \int_1^\infty\frac{\log\Omega(t)}{t^2}\mathrm{d}t<\infty. \] Investigations of Eq. (1) can be found in [\textit{E. N. Dancer}, Proc. R. Soc. Edinb., Sect. A, Math. 76, 283--300 (1977; Zbl 0351.35037); \textit{S. Fučík}, Solvability of nonlinear equations and boundary value problems. Dordrecht-Boston-London: D. Reidel Publishing Company (1980; Zbl 0453.47035)], and for recent developments it can be referred the papers [\textit{T. Gallouet} and \textit{O. Kavian}, Commun. Partial Differ. Equations 7, 325--342 (1982; Zbl 0497.35080); \textit{P. Habets} et al., Nonlinear Anal., Theory Methods Appl. 20, No. 5, 533--549 (1993; Zbl 0772.34025); \textit{A. C. Lazer} and \textit{P. J. Mckenna}, Proc. Am. Math. Soc. 106, No. 1, 119--125 (1989; Zbl 0683.34027); \textit{M. Zhang}, J. Differ. Equations 145, No. 2, 332--366 (1998; Zbl 0913.34051)] and the references therein. \textit{R. Ortega} [Adv. Nonlinear Stud. 1, No. 1, 14--39 (2001; Zbl 1136.37336)] proved the boundedness of all solutions of Eq. (1) under the conditions \(2\omega_0\in{\mathbb I}\) and \[ \int_0^{2\pi}p(t)\mathrm{d}t\ne0. \tag{3}\] In this paper, the authors show that under the condition (2), the condition (3) is not necessary. For this purpose, they first study the stability of the trivial solution \(r=0\) of a class of abstract mappings defined on the plane. The results obtained by the process of normalization, the result in [loc. cit.] and the Moser's twist theorem [\textit{J. Moser}, Nachr. Akad. Wiss. Gött., II. Math.-Phys. Kl. 1962, 1--20 (1962; Zbl 0107.29301)].