On a Keller-Segel system with logarithmic sensitivity and non-diffusive chemical (Q476603): Difference between revisions

This paper is concerned with the system of partial differential equations \[ u_t=\Delta u-\chi\nabla\cdot(u\nabla(\log w)),\qquad w_t=uw^\lambda, \] posed on \(\Omega\times (0,\infty)\) with \(\Omega\subset\mathbb{R}^n\) \((n\geq 1)\) an open domain with smooth boundary. This model is of Keller-Segel type: The equation for \(u\) describes chemotaxis of microorganisms with logarithmic chemotactic sensitivity. The signalling substance described by \(w\) is assumed to be non-diffusive, but of growing concentration. The system above is endowed with homogeneous Neumann boundary conditions and nonnegative initial data with the further restriction that \(w_0\) is uniformly positive. The main results of this work deal with local- and global-in-time existence as well as with the blow-up of solutions depending on the values of \(\chi>0\) and \(\lambda>0\). A related study has been made in [the second author et al., Commun. Partial Differ. Equations 35, No. 2, 245--274 (2010; Zbl 1193.35013)] for a similar model in one spatial dimension. After the change of variables \(v:=(1-\lambda)^{-1} w^{1-\lambda}\) (\(\lambda\neq 1\)), resp. \(v:=\log w\) (\(\lambda=1\)), the authors obtain the following local existence result by classical fixed-point arguments (cf. Theorem 1.1):\newline If \(u_0,v_0\in W^{k,\infty}(\Omega)\cap W^{k+1,p}(\Omega)\) for some \(p>n\), then there exists \(T_{\max}\in (0,\infty]\) and a solution \(u,v\in L^\infty([0,T_{\max});W^{k,\infty}(\Omega))\cap L^p([0,T_{\max});W^{k+1,p}(\Omega))\) blowing up in the respective norm as \(t\) approaches \(T_{\max}\). Furthermore, \(u(\cdot,t)\) is nonnegative and \(v(x,\cdot)\) is increasing for fixed \(x\in\Omega\). For bounded \(\Omega\), the question of global existence versus blow-up in finite time is adressed as follows (cf. Theorem 1.2):\newline If \(\lambda>1\), then \(T_{\max}<\infty\) unless \(u_0\equiv 0\). On the other hand, for \(\lambda\in (0,1)\) with \(\chi\leq 1-\lambda\), one has \(T_{\max}=\infty\). For the remaining case \(\lambda\in (0,1)\), \(\chi>1-\lambda\), on the whole space \(\Omega=\mathbb{R}^n\) (cf. Theorem 1.3), the authors exclude the existence of nontrivial self-similar blow-up profiles \(\Phi,\Psi\) of the form \[ u(x,t)=\frac1{(T_{\max}-t)^{\frac{\alpha}{2}+1}}\Phi\left(\frac{x}{\sqrt{T_{\max}-t}}\right),\quad v(x,t)=\frac1{(T_{\max}-t)^{\frac{\alpha}{2}}}\Psi\left(\frac{x}{\sqrt{T_{\max}-t}}\right), \] in the case \(\alpha>-2\), assuming \(T_{\max}<\infty\). There, \(\Phi\) and \(\Psi\) are supposed to solve a corresponding stationary system in the sense of distributions.