The quadrature formulas we shall discuss in this paper belong to the same type as the well known formula given by Gauss.They can be characterized as follows.Let p(x) denote a function which does not assume negative values in a given interval (a, b) and is subject to the condition that the integrals Cm = I xmp(x)dx, or moments of the distribution determined by p(x), exist.Under such circumstances it is always possible to find » real numbers Xl < X2 < Xi < • • • < X» belonging to the interval (a, b) together with n corresponding constants Ai, At, • • • , An such that the formula (1) f P(x)f(x)dx = Aif(Xi) + Atf(x2) + ■■■+ Anf(Xn) holds true whenever f(x) is a polynomial whose degree does not exceed 2»-1, and this property completely determines the numbers xi} x2, • • • , xK as well as the corresponding coefficients A\, A2, • • • , An.It is important to notice that these coefficients are all positive.For any function f(x) which is not a polynomial of degree g 2«-1 the right member of (1) ceases to represent exactly the integral in the left member, and we must add the remainder 2?» in order to have an exact equality:(2) f P(x)f(x)dx = Aif(xi) + A2f(Xi) + ■■■+ Anf(Xn) + Rn. JaFor any function f(x) possessing a derivative of order 2» the following expression for the remainder Rn can be obtained:(3) Rn = p*kq 2»!I P(x)u2(x)dx, Ja
No takes yet. Share an insight, caveat, or question.
J. V. Uspensky (1928) studied this question.