The critical exponent for semilinear \(\sigma \)-evolution equations with a strong non-effective damping (Q2057929)

In the present paper, the authors study the semilinear Cauchy problem \[ u_{tt}+ (-\Delta)^\sigma u+(-\Delta)^\theta u_t=f(u,u_t),\quad u(0,x)=0,\quad u_t(0,x)=u_1(x) \] for \(\sigma\)-evolution equations with a structural damping \((-\Delta)^\theta u_t\), where \(\theta \in (\frac{\sigma}{2},\sigma]\). So, the damping is non-effective, a classification of damping terms was introduced by the authors in an earlier paper. The nonlinearity is of power type \(|u|^\alpha\) or of power derivative type \(|u_t|^\alpha\) with \(\alpha>1\). The main goal is to verify the critical exponents in both cases \(|u|^\alpha\) and \(|u_t|^\alpha\). If \(f(u,u_t)=|u|^\alpha\), then the critical exponent \(\alpha_{\text{crit}}=\alpha_{\text{crit}}(n,\sigma)=1+\frac{2\sigma}{n-\sigma}\) for \(n >\sigma\). If \(f(u,u_t)=|u_t|^\alpha\), then the critical exponent \(\alpha_{\text{crit}}=\alpha_{\text{crit}}(n,\sigma)=1+\frac{\sigma}{n}\) for low space dimensions. It should be mentioned, that both critical exponents are independent of \(\theta \in (\frac{\sigma}{2},\sigma]\) in opposite to the effective case \(\theta \in [0,\frac{\sigma}{2}]\). But \(\theta \in (\frac{\sigma}{2},\sigma]\) appears in the derived energy estimates for the global (in time) small data Sobolev solutions. One of the most important questions is that for the regularity of the data \(u_1=u_1(x)\). In the first case \(f(u,u_t)=|u|^\alpha\) the data is taken from \(L^1(\mathbb{R}^n)\cap L^\eta(\mathbb{R}^n)\) with a suitable \(\eta\geq 2\) depending on the sign of \(n-\theta\). In the second case \(f(u,u_t)=|u_t|^\alpha\) the data is taken from \(L^1(\mathbb{R}^n)\cap L^\alpha(\mathbb{R}^n)\). First, the authors recall suitable \(L^p-L^q\) estimates of solutions to the linear Cauchy problem \[ u_{tt}+ (-\Delta)^\sigma u+(-\Delta)^\theta u_t=0,\quad u(0,x)=0,\quad u_t(0,x)=u_1(x). \] Then, by Duhamel's principle and tricky estimates of the nonlinear term some well-posedness can be proved by using contraction argument. Finally, the authors discuss if \(L^1\) property of the data is replaced by \(L^p\) for \(p>1\). Then they propose a family of critical exponents and show by test function method that these exponents are really critical.