\(L^{p}-L^{q}\) decay estimates for dissipative linear hyperbolic systems in 1D (Q2404254)

Given \(A,B \in \mathbb{R}^{n\times n}\) the authors are interested in the Cauchy problem for partially dissipative hyperbolic systems \[ \partial_t u + A \partial_x u+ Bu=0,\,\,\,u(0,x)=u_0(x). \] Here the following assumptions are supposed: {\parindent=0.7cm\begin{itemize}\item[1.] Hyperbolicity: The matrix \(A\) is diagonalizable with real eigenvalues. \item[2.] Partial dissipativity: The eigenvalue \(0\) of \(B\) is semi-simple with algebraic multiplicity \(m \geq 1\). The spectrum \(\sigma(B)\) of \(B\) can be decomposed as \(\sigma(B)= \{0\} \cup \sigma_0\), where \(\sigma_0 \subset \{ \lambda \in \mathbb{C}: \Re \lambda >0\}\). \item[3.] Reduced hyperbolicity: The matrix \(C\) is diagonalizable with real eigenvalues. The matrix \(C\) appears in the reduced system \(\partial_t w + C \partial_x w \thickapprox 0\), where \(w:=P_0^{(0)}u\) and \(C:=P_0^{(0)} A P_0^{(0)}\). Here \(P_0^{(0)}\) is the unique eigenprojection associated with the eigenvalue \(0\). \item[4.] Uniform dissipativity: All eigenvalues \(\lambda(i\xi)\) of the matrix operator \(E(i\xi):=-(B+i\xi A)\) satisfy the Shizuta-Kawashima condition \[ \Re (\lambda(i\xi))\leq - \frac{\theta|\xi|^2}{1+|\xi|^2},\text{ for }\xi \neq 0, \] where \(\theta\) is a suitable positive constant. \end{itemize}} Under these assumptions the authors are able to prove the following result about \(L^q-L^p\) decay estimates: If \(u=u(t,x)\) is a solution to the above Cauchy problem with \(u_0 \in L^q(\mathbb{R})\), then the \(L^q-L^p\) estimates \[ \|(u-U-V)(t,\cdot)\|_{L^p} \leq Ct^{-\frac{1}{2}(\frac{1}{q}-\frac{1}{p})-\frac{1}{2}}\|u_0\|_{L^q} \] hold for \(t \geq 1\), \(1\leq q \leq p \leq \infty\) with a positive constant \(C=C(p,q)\). Here \(U=U(t,x)\) is the diffusion wave and \(V=V(t,x)\) is the exponentially decaying wave. Both waves satisfy the \(L^q-L^p\) estimates (\(\delta >0\)) \[ \|U(t,\cdot)\|_{L^p} \leq C t^{-\frac{1}{2}(\frac{1}{q}-\frac{1}{p})}\|u_0\|_{L^q},\,\,\,\|V(t,\cdot)\|_{L^2}\leq C e^{-\delta t}\|u_0\|_{L^2}. \] The proof bases on a complex interpolation argument once \(L^r\) estimates for the fundamental solution to the system in the frequency space and Fourier multiplier estimates are established. The authors prove a better \(L^q-L^p\) decay estimate if a symmetry property and a reduced strictly hyperbolicity condition are assumed.