Full discretization of some reaction diffusion equation with blow up (Q2501227)

The authors consider the nonlinear reaction-diffusion problem \(u_t-\Delta u^{1+ \delta}+\gamma\vec V\cdot\nabla u^{1+\delta}=\alpha u^p\) in \(\Omega \times (0, \infty)\), \(u=0\) on \(\partial\Omega\times(0, \infty)\), \(u(x,0)=u_0(x)\) in \(\Omega\). Here \(\Omega\) denotes a bounded domain in \(\mathbb{R}^d\), \(\delta>0\), \(p>1\). For \(v:=u^{1+ \delta}\) they construct the time semi-discretization to the above problem and prove the existence of the unique solution \(v_{n+1}\in H_0^1 (\Omega)\) given \(v_n\in H_0^1(\Omega)\) of this semi-discretization, where \(v_n\) denotes an approximation of \(v\) at the time \(t_n\), \(t_n= t_{n-1}+\Delta t\). A method of finding the sequence \((v_{n,j})j\geq 0\), which converges uniformly to \(v_n\), is proposed. (It leads to a full discretization of the original problem). The algorithm to obtain \((v_{n,j}) j\geq 0\) is written out and in the one-dimensional spatial case (using the finite difference approximation) it is considered in details. Numerical simulations for \(\Omega=(-1,1)\), homogeneous boundary conditions and \(u_0(x)=5(1-x^2)\) are done and compared with numerical results obtained by a scheme based on the Runge-Kutta method.