The numerical solution from measurement data of linear integral equations of the first kind

The problem considered in this paper is the solution of the Volterra integral equation of the first kind for the input function where we are given the impulse response or kernel and discrete, noise contaminated, observed values of the output function. The present method, introduced here, is based on assuming a multipolynomial form for the input function. Using the integral equation and the least squares method, the coefficients of all the terms in these polynomials are found simultaneously, in order to avoid bias toward any subset of the data. The solution is obtained in such a way that successive polynomials are continuous in function value and derivative at the points of juncture. Thus a smoothed estimate of the input function is generated. The development of the algorithm is presented and applied to two illustrative problems. Possible improvements and extensions of the method are also discussed.