On some finite difference schemes for solution of hyperbolic heat conduction problems (Q657409)

The authors consider hyperbolic heat conduction modeled by a time operator \(L_\epsilon u:=\epsilon u_{tt}+u_t\) and first discuss difference schemes of two publications by other authors. Based on a theorem for 3-level difference schemes (going back to Samarskii and Gulin (1973), cited here from the book of \textit{A. A. Samarskii, P. P. Matus} and \textit{P. N. Vabishchevich} [Difference schemes with operator factors. Mathematics and its Applications (Dordrecht). 546. Dordrecht: Kluwer Academic Publishers. (2002; Zbl 1018.65103)]) they point out that one of the former schemes is neglecting the initial condition for the time derivative \(u_t\) and unstable for \(\epsilon=0\) whereas the other (\(L_2\)-stable) scheme is, under a slightly more stringent condition, also stable in an energy norm, but not consistent if the time step goes to zero independently of \(\epsilon\). They also propose two schemes and show their stability. It should be remarked that these authors, for the proposed schemes, don't discuss the approximation of the initial condition of \(u_t\) either, and that they don't seem to know the powerful material connected with the approximation of \(L_\epsilon u\), not to speak of approximation and convergence uniformly in \(\epsilon\), (see, e.g., [\textit{H.-G. Roos, M. Stynes} and \textit{L. Tobiska}, Robust numerical methods for singularly perturbed differential equations. Springer Series in Computational Mathematics 24. Berlin: Springer. (2008; Zbl 1155.65087)]). Admittedly, the latter concerns boundary value problems whereas the authors consider initial value problems (for this time operator).