Computation of the inverse Laplace transform based on a collocation method which uses only real values (Q2433784)

The authors develop a numerical algorithm for inverting a Laplace transform, based on Laguerre series expansion of the inverse function \[ f(x)= e^{\sigma x}\sum^\infty_{k= 0} c_k e^{-bx} L_k(2bx) \] under the assumption that the Laplace transform is known on the real axis only. The method belongs to the class of collocation methods, and is applicable when the Laplace transform function is regular at infinity. The difficulties associated with this inverse problem are due to its intrinsic ill-posedness and the derivation of error estimates is the first step towards the development of a relative inversion algorithm. The main contribution of this paper is to provide computable estimates of truncation, discretization, conditioning and roundoff errors introduced by numerical computations. Moreover, pseudoaccuracy is introduced which is used by the numerical algorithm in order to provide uniform scaled accuracy of the computed approximation for any \(x\) w.r.t. \(e^{\sigma x}\). These estimates are then employed to dynamically truncate the series expansion. With the aim to validate the reliability and usability of the algorithm experiments were carried out on several test functions.