High-order numerical solution of second-order one-dimensional hyperbolic telegraph equation using a shifted Gegenbauer pseudospectral method (Q2804377)

Subject of this paper is application of the so-called shifted Gegenbauer pseudospectral method to the one-dimensional telegraph equation of the form NEWLINE\[NEWLINEu_{tt}+{\beta_1}u_t+{\beta_2}u=u_{xx}+f(x,t),\;x\in[0,l],\;t\in [0,\tau],\leqno(1)NEWLINE\]NEWLINE with initial and Dirichlet boundary conditions. The author gives a very complete introduction to the theory of the shifted Gegenbauer polynomials: NEWLINE\[NEWLINEC_n^{\alpha}(x)={{n!\Gamma(\alpha+{1\over2}}) \over{\Gamma(n+\alpha+{1\over2}})}P_n^{(\alpha-{1\over2},\alpha-{1\over2})},\leqno(2)NEWLINE\]NEWLINE where \(P_n^{(\alpha,\beta)}\) is the Jacobi polynomial, containing also Gegenbauer-Gauss quadrature. Also, the bivariate shifted Gegenbauer polynomials are introduced: NEWLINE\[NEWLINE_{l,\tau}C_{n,m}^{\alpha}(x,t)=C_{l,n}^{(\alpha)}(x)C_{\tau,m}^{(\alpha)}(t),\leqno(3)NEWLINE\]NEWLINE for \((x,\tau)\in[0,l]\times[0,\tau]\).NEWLINENEWLINEThe mixed initial-boundary value problem (1) is reformulated as integral equation, denoted by NEWLINE\[NEWLINE\Theta\phi=\Omega\phi-\psi. \leqno(4)NEWLINE\]NEWLINE The integrals involved in (4) are approximated via Gegenbauer-Gauss qua\-dra\-tu\-res using the uni- and bivariate Gegenbauer shifted polynomials.NEWLINENEWLINEIn order to get a system of linear algebraic equations approximating Problem (4), the finite set of nodes on the 2D domain of definition of Problem (4) is chosen. For optimality of the method, the optimal choice of quadratures for any node, via solution of a certain minimalization problem is provided.NEWLINENEWLINEError estimates are given. The author joins four numerical examples.NEWLINENEWLINEReviewer's remark: The way to create the algorithm seems to be rather complicated. Anyway, it should be interesting to compare the proposed algorithm with the standard finite difference method (with or without the `time-splitting'), on this one-dimensional telegraph equation.