Existence of solutions for a nonlinear system of parabolic equations with gradient flow structure (Q403832): Difference between revisions

Given \(d\geq 3\), \(m>2(d-1)/d\), \(K_\rho>0\), \(\chi\in\mathbb{R}\), \(K_c>0\), \(\alpha\in\mathbb{R}\), \(\beta\geq 0\), and \(V\in W^{2,\infty}(\mathbb{R}^d)\), the existence of a global weak solution to the parabolic system \[ \partial_t \rho = \mathrm{div}\left[ K_\rho \nabla \rho^m + \rho \nabla \left( V + \chi c \right) \right]\;, \qquad \partial_ t c = K_c \Delta c - \alpha \rho - \beta c \] in \((0,\infty)\times \mathbb{R}^d\) is proved when \(\alpha \chi>0\) by exploiting its underlying gradient flow structure. More precisely, introducing the energy \[ \mathcal{H}(\rho,c) = \int_{\mathbb{R}^d} \left( \frac{K_\rho}{m-1} \rho^m + \rho V + \frac{K_c \chi}{2 \alpha} |\nabla c|^2 + \frac{\chi \beta}{2 \alpha} c^2 + \chi \rho c \right) dx, \] the above system has a gradient flow structure associated to \(\mathcal{H}\) in \(\mathcal{P}_2(\mathbb{R}^d)\times L^2(\mathbb{R}^d)\), the metric being the \(2\)-Wasserstein distance in the space \(\mathcal{P}_2(\mathbb{R}^d)\) of probability measures in \(\mathbb{R}^d\) with finite second moment for the first component and the usual distance associated to the \(L^2(\mathbb{R}^d)\)-norm for the second component. Note that the constraint \(\alpha\chi>0\) seems to be necessary for the gradient flow structure to be available while the constraint \(m>2(d-1)/d\) excludes the possibility of finite time blowup when \(\chi<0\) (recall that, when \(\chi<0\) and \(V\equiv 0\), the above system is the parabolic-parabolic Keller-Segel system with nonlinear diffusion which has been intensively studied recently). Taking advantage of the gradient flow structure a variational scheme is set up to construct a time-discrete solution which is shown to converge to a weak solution of the original problem as the time step decreases to zero. An important step of the proof is the derivation of additional regularity properties of the minimizers of the variational scheme which allows one to identify the Euler-Lagrange equations solved by the minimizers. It is performed by adapting techniques from [\textit{D. Matthes} et al., Commun. Partial Differ. Equations 34, No. 11, 1352--1397 (2009; Zbl 1187.35131)] and [\textit{A. Blanchet} and \textit{P. Laurençot}, Commun. Partial Differ. Equations 38, No. 4--6, 658--686 (2013; Zbl 1282.35202)].