Global attractors for semilinear damped wave equations with supercritical exponent (Q1346232): Difference between revisions

The author investigates a nonlinear damped wave equation of the form: \[ u_{tt}+ g(u_ t)- \Delta u+ f(u)=0,\;\;u=0 \quad \text{on} \;\partial\Omega, \quad u(0)= u_ 0,\;\;u_ t (0) =u_ 1.\tag{P} \] Here, \(\Omega\) is a smooth and bounded domain, and \(f\), \(g\) are functions subject to various constraints: (1) \(g(0)=0\), \(\liminf g'(z) >0\) as \(| z|\to 0\), (2) \(c_ 1(| z|^ r- c_ 2)\leq | g(z)|\leq c_ 3 (1+| z|^ q)\), (3) \(| f'' (z)|\leq c_ 4 (1+| z|^ p)\). Here \(1\leq r\leq q<5\) and \(p< 2((2r-1)/ (r+1))\). Moreover: (4) \(\liminf f(z) z^{-1}> -\lambda_ 1\) as \(| z|\to \infty\), with \(\lambda_ 1\) the first positive Dirichlet eigenvalue of \(-\Delta\) on \(\Omega\). The basic space underlying (P) is \(H_ 0^ 1 (\Omega)\times L^ 2 (\Omega)\), i.e. \(u\in H^ 1_ 0 (\Omega)\), \(u_ t\in L^ 2 (\Omega)\). A familiar definition of global attractor is given. The main result then states: \((*)\) under the above assumptions, (P) has a unique global attractor. One of the main difficulties with the proof of \((*)\) is its interpretation: conditions (2), (3) admit functions \(f\), \(g\) which do not induce smooth mappings from \(H^ 1_ 0 (\Omega)\) and \(L^ 2 (\Omega)\) into \(L^ 2 (\Omega)\) respectively. This fact forces the author to introduce the concept of weak solution of (P), for whose definition we refer to the paper. The main feature of these weak solutions is that it is unknown whether they are unique or not. Statement \((*)\) now more explicitly states that there is a unique global attractor with respect to weak solutions. The proof of this statement is based on some preparatory lemmas and propositions, whose aim is to secure the existence of an absorbing set. The proofs of these propositions use rather delicate estimates. The paper contains further information on the structure of the trajectories in the absorbing set, and on their decay properties.