Weak stabilization in degenerate parabolic equations in divergence form: application to degenerate Keller-Segel systems (Q2127342)

The large time behaviour of non-negative global and bounded weak solutions to the initial-value problem \[ \partial_t u = \operatorname{div}\big( f(u) \nabla u + g(u,x,t) \big), \qquad x\in \Omega, \ t>0, \tag{\(E_g\)} \] supplemented with no-flux boundary conditions and initial condition \(u_0\in L^\infty(\Omega)\), \(u_0\ge 0\), is investigated when \(\Omega\) is a smooth bounded domain of \(\mathbb{R}^N\), \(N\ge 1\), the function \(f\in C([0,\infty))\times C^2((0,\infty))\) satisfies \[ f(\sigma)\ge k_0 \sigma^{m-1}, \quad \sigma\ge 0, \qquad\qquad \limsup_{\sigma\to 0} \sigma f'(\sigma) < \infty \] for some \(m\ge 1\) and \(k_0>0\), and the vector field \(g\) can be approximated in a suitable sense by a family of vector fields \(g_\varepsilon\in C([0,\infty)\times\bar{\Omega}\times [0,\infty),\mathbb{R}^N)\) such that, for any \(w\in L^\infty(\Omega\times (0,\infty))\) with \(\|w\|_{L^\infty(\Omega\times (0,\infty))} \le c\), the vector-valued function \(g_\varepsilon(w)\) belongs to \(L^2(\Omega\times(0,\infty),\mathbb{R}^N)\) with \[ \sup_{\varepsilon\in (0,1)} \Big\{ \|g_\varepsilon(w)\|_{L^2(\Omega\times(0,\infty),\mathbb{R}^N)} \Big\}\le M(c). \] It is then shown that, if \((E_{g_\varepsilon})\) has a non-negative global and bounded classical solution \(u_\varepsilon\) which satisfies in addition \(u_\varepsilon\le u_{\mathrm{max}}\) in \(\Omega\times (0,\infty)\) for all \(\varepsilon\in (0,1)\), then \((E_g)\) has a non-negative global and bounded weak solution \(u\) such that \(u\le u_{\max}\) in \(\Omega\times (0,\infty)\) and \[ u(t) \stackrel{*}{\rightharpoonup} \frac{1}{|\Omega|} \int_\Omega u_0(x)\ \mathrm{d}x \quad\text{ in }\quad L^\infty(\Omega) \] as \(t\to\infty\). The assumed \(L^2\)-uniform boundedness on \(g_\varepsilon\) is instrumental in the proof of the last property. Applications to quasilinear Keller-Segel systems are next given.