Abstract
Let P and A be symmetric linear operators defined on a dense domain $D \subset H$, a (real) Hilbert space. Let $(x,Av) \geqq \lambda (x,x)$ for all $x \in D$ and some $\lambda > 0$ and $(x,Px) > 0$ for all $x \in D$, $x \ne 0$. Let D have a Hilbert space structure and let the embedding be continuous. Let $\mathcal{F}:D \to H$ be a nonlinear gradient operator and let $\mathcal{G}$ be a potential associated with $\mathcal{F}$. Suppose that $2(2\alpha + 1)\mathcal{G}(x) \leqq (x,\mathcal{F}(x))$ for all $x \in D$ and some $\alpha > 0$. If $u:[0,T] \to D$ is a solution to $Pu_{tt} = - Au + \mathcal{F}(u)$ with $u(0) = u_0 ,v_t (0) = v_0 $, and if $\mathcal{G}(u_0 ) > \frac{1}{2}[(u_0 ,Au_0 ) + (v_0 ,Pv_0 )] - (\lambda \alpha /(2\alpha + 1))(u_0 ,Pu_0 )$, then, for some $T < \infty $, $\lim _{t \to T^ - } (u,Pu) = + \infty $. An analogous result holds for weak solutions to this equation and for the damped equation $Pu_{tt} + \hat Au_t + Au = \mathcal{F}(u)$, where $\hat A$ is a positive semidefinite linear operator.

This publication has 2 references indexed in Scilit: