No label defined (Q4952187)

The author considers the approximation of the scalar linear parabolic problem \[ u_t- \text{div}(a\text{ grad }u)+ bu= f\quad\text{in }\Omega\times (0,T], \] \[ u= 0\quad\text{on }\partial\Omega\times (0,T],\quad u(\cdot, 0)= u_0\quad\text{in }\Omega \] in a two-dimensional polygon \(\Omega\), where \(a= a(x,t)\geq a_0> 0\), \(b= b(x,t)\geq b_0> 0\), \(f= f(x,t)\), and \(u_0\) are sufficiently smooth functions. The discretization in space is a conforming piecewise linear finite element method, whereas in time an \(A\)-stable single step \(\theta\)-scheme is employed. Starting with results known for the corresponding elliptic problem [cf. \textit{R. Duran}, \textit{M. A. Muschietti} and \textit{R. Rodriguez}, Numer. Math. 59, No. 2, 107-127 (1991; Zbl 0716.65098)] superconvergence results and, by using recovery operators, a posteriori error estimates are derived. The estimators are proven to be asymptotically exact. A lower bound for the error has not been given, so that the efficiency of the estimator suggested seems to be an open question. However, the analysis requires quite regular solutions \(u,u_t\in L^2(0, T; H^2(\Omega)^2)\), \(u_{tt}\in L^2(0,T;\) \(L^2(\Omega)^2)\) and thus \(u\in {\mathcal C}([0,T]\); \(H^2(\Omega)^2\), \(u_t\in{\mathcal C}([0, T]\); \(H^1(\Omega)^2)\), as well as higher regularity for the Crank-Nicolson time discretization). At the end of the paper, the author remarks that method and results ``can be generalized to nonlinear problems in a straightforward fashion''. Unfortunately, the reader is not told what kind of nonlinearity (and which assumptions on it) the author has in mind. Regarding nonlinear problems, immediately the question of solvability and regularity (and hence compatibility of the problem's data) comes up, the Aubin-Nitsche trick, which has been used in the course of the analysis, is no longer at hand, and the applied discrete Gronwall lemma could end up in a constant that depends exponentially on some characteristics of the nonlinearity. So far, the remark rests, more or less, void. Many misprints make reading the paper unpleasant.