An explicit inversion formula for tridiagonal matrices

From MaRDI portal





Consider a real tridiagonal matrix \(\overline A=(\overline a_{ij}), \overline a_{ii}=a_i+b_i+a_{i+1}\), \(\overline a_{i,i-1}=-a_i\), \(\overline a_{i,i+1}=-a_{i+1}\). An explicit inversion formula for the matrix \(A\) is derived by using the inversion formula given by \textit{T. Yamamoto} and \textit{Y. Ikeba} [Linear Algebra Appl. 24, 105-111 (1979; Zbl 0408.15004)]. The matrix \(A\) is associated with with the discretization of a two-point boundary value problem \(-(p(x)u'(x))'+r(x)u(x)=f(x)\) in \((a,b)\), \(u(a)=u(b)\), where \((a, b)\subset R\). \(p, q, r\) are functions, \(p\in C^2[a, b]\), \(p(x)\geq\delta >0\), \(\delta\) is a constant, and \(q, r\in L^{\infty}(a, b)\). Numerical experiments present that the usage of the derived inversion formula enables a systematic error analysis when discretizing this two-point boundary value problem by using the finite difference method of \textit{G. H. Shortley} and \textit{R. Weller} [J. Appl. Phys. 9, 334-344 (1938; Zbl 0019.03801)]. For the entire collection see [Zbl 0973.00063].











This page was built for publication: An explicit inversion formula for tridiagonal matrices

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q2765624)