A note on global existence of solutions to nonlinear Klein-Gordon equations in one space dimension (Q1970344): Difference between revisions

In this paper the following Cauchy problem for the nonlinear Klein-Gordon equation is considered: \[ (\square+1)u=F(u,u_t,u_x,u_{tx},u_{xx})\quad \text{for}\quad (t,x)\in(0,\infty)\times{\mathbb R},\tag{1} \] \[ u(0,x)=\varepsilon f(x),\quad u_t(0,x)=\varepsilon g(x)\quad \text{for}\quad x\in{\mathbb R}, \tag{2} \] where \(\square=(\partial^2/\partial t^2)-(\partial^2/\partial x^2)\), \(u_t=(\partial/\partial t)u\), \(u_x=(\partial/\partial x)u\), etc., \(\varepsilon\) is a small positive parameter, \(f\), \(g\in C_0^\infty({\mathbb R})\) and the nonlinear function \(F(\dots)=F(\lambda)\) with \(\lambda=(u,u_t,u_x,u_{tx},u_{xx})\) can be written in the form \(F(\lambda)=\sum_{i=1}^{10}c_iG_i(\lambda) +N(\lambda)+H(\lambda)\) in a neighborhood of \(\lambda=0\); here \(\{c_i\}_{i=1}^{10}\subset{\mathbb R}\), \(\{G_i\}_{i=1}^{10}\) are linearly independent polynomials of \(\lambda\) defined by \[ \begin{aligned} G_1(\lambda)&=u(-u^2+3u_t^2-3u_x^2),\\ G_2(\lambda)&=u_t(-3u^2+u_t^2-u_x^2)+2u(u_tu_{xx}-u_xu_{tx}),\\ G_3(\lambda)&=u_x(-u^2+u_t^2-u_x^2)+2u(u_tu_{tx}-u_xu_{tx}),\\ G_4(\lambda)&=u^3-2u^2u_{xx}-3uu_t^2+2u_t^2u_{xx}-2u_tu_xu_{tx} -u(u_{tx}^2-u_{xx}^2),\\ G_5(\lambda)&=(-u^2+u_t^2-u_x^2)u_{tx}-2uu_tu_x,\\ G_6(\lambda)&=-uu_x^2+2u_x(u_tu_{tx}-u_xu_{xx})+u(u_{tx}^2-u_{xx}^2),\\ G_7(\lambda) &=3u^2u_t-6uu_tu_{xx}-u_t^3-3u_t(u_{tx}^2-u_{xx}^2),\\ G_8(\lambda)&=u^2u_x-2uu_tu_{tx}-2uu_xu_{xx}-u_t^2u_x-u_x(u_{tx}^2-u_{xx}^2),\\ G_9(\lambda)&=-2uu_xu_{tx}-u_tu_x^2+u_t(u_{tx}^2-u_{xx}^2),\\ G_{10}(\lambda)&=-u_x^3+3u_x(u_{tx}^2-u_{xx}^2),\end{aligned} \] \(N\) is of the form \(N(\lambda)=P_1(\lambda)(u_tu_{tx}-u_xu_{xx}+uu_x)+ P_2(\lambda)(u_tu_{xx}-u_xu_{tx})+P_3(\lambda)(u_{tx}^2-u_{xx}^2+uu_{xx})\), \(\{P_i(\lambda)\}_{i=1}^3\) are homogeneous polynomials of degree~\(1\), and \(H(\lambda)\) is a smooth function of \(\lambda\) of degree~\(4\), i.e.\ \(H(\lambda)=O(|\lambda|^4)\) near \(\lambda=0\). The main result of the paper is the following: Theorem 1.1. For any integer \(k\geq 15\) there exists a positive constant \(\varepsilon_0\) such that for any \(\varepsilon\in(0,\varepsilon_0]\) the Cauchy problem \text{(1)}, \text{(2)} has a unique classical solution \(u=u(t,x)\in C^\infty([0,\infty)\times{\mathbb R})\). Moreover, the solution \(u\) has a free profile in the sense that there exist \(u_0\in H^{k+1}({\mathbb R})\) and \(u_1\in H^k({\mathbb R})\) (\(H^k({\mathbb R})\) is the usual Sobolev space) such that \[ \|(u-U)(t,\cdot)\|_{H^{k+1}({\mathbb R})}+\|\partial_t(u-U)(t,\cdot)\|_{H^k({\mathbb R})}\to 0\quad\text{as}\quad t\to+\infty, \] where \(U\) is the solution to the Cauchy problem for the linear Klein-Gordon equation \((\square+1)U=0\) in \((0,\infty)\times{\mathbb R}\) with initial data \(U(0,x)=u_0(x)\) and \(U_t(0,x)=u_1(x)\), \(x\in{\mathbb R}\).