On the nonlinear stabilization of the wave equation (Q4389925)

This paper considers the problem \(Lu+f(u_t)=0\) on \(\Omega\times{\mathbb{R}}^n\), where \(\Omega\subset{\mathbb{R}}^n\) is open, bounded, and smooth, and \(L\) is the wave operator. The boundary conditions are: \[ u=0 \quad\text{on }\Gamma_0\subset\partial\Omega;\quad u_\nu+(m\cdot\nu)(au+g(u_t))=0\quad\text{on }\Gamma_1=\partial\Omega\setminus\Gamma_0. \] Here, \(m(x)=x-x_0\) where \(x_0\) is given, and \(\Gamma_0:=\{m\cdot\nu\leq 0\}\). Exponential energy decay is proved if (1) \(f\) and \(g\) are non-increasing; (2) there are constants \(c_1\) and \(c_2\) such that \(| g(x)/x| \) and \(| f(x)/bx| \) both lie between \(c_1\) and \(c_2\), with \(b=(3aR^2-n)/2\) and \(R=\sup_\Omega | m| \); (3) \(\max (n-2,n/3)\leq aR^2 \leq n\); (4) \(\Gamma_0\) is non-empty or \(a>0\) and \(\inf_{\Gamma_1}(m\cdot\nu)>0\), and (5) \(\bar{\Gamma}_0\bigcap\bar{\Gamma}_1\) is empty if \(n>3\). The decay rate is given explicitly in terms of \(c_1\), \(c_2\), \(a\), \(R\) and \(n\), and this improves, if \(n=2\), a result due to \textit{V. Komornik} [SIAM J. Control Optimization 29, 197-208 (1991; Zbl 0749.35018)]. The method consists in using the multiplier \(2m\cdot\nabla u+2aR^2u\).