Richardson extrapolation of iterated discrete projection methods for eigenvalue approximation (Q953369): Difference between revisions

The eigenvalue approximation of a compact linear integral operator with a smooth kernel is discussed. At first, asymptotic error expansions of the iterated discrete Galerkin and iterated discrete collocation methods are proposed. The Gauss quadrature rule and a composite quadrature formula are used to calculate integrals numerically. To obtain an asymptotic error expansion of the operator the Euler-Maclaurin (not: Maclacrin) summation is applied and gives a representation with Bernoulli polynomials. Then, the Richardson extrapolation can be used and gives error estimations of the eigenvalue approximation. The results are demonstrated by means of the very simple integral operator: \((Kx)(s)= \int^1_0e^{st}x(t)dt\), \(0\leq t\leq 1\).