Loading [MathJax]/jax/output/HTML-CSS/jax.js

Saturday, 21 July 2018



Convergence to zero of a function whose improper integral converges.


It is well known that if a series n=1an converges, then an0 as n. The reason is that since the sequence of partial sums SN=Nn=1an  is a convergent sequence of real numbers, it must be Cauchy, so in particular |Sn|=|Sn+1Sn| can be made smaller than any given ε>0, provided n is sufficiently large.
A similar argument does not hold for convergent integrals. Contrary to this situation, if f is a function for which the improper Riemann integral af(x)dx converges, one cannot deduce that limxf(x)=0. Although one still has a Cauchy condition, namely, that the integrals βαf(x)dx are small if α,β are sufficiently large, there is enough freedom to form spikes on the graph of f, which do not contribute much to the integral. Such "spikes" will necessarily cause the function not to be uniformly continuous. If, however, the function is uniformly continuous, then limxf(x)=0, for if this were not the case, then for some ε>0 we could find an infinite sequence {xn}n=1 such that |xnxn+1|>1, say, for which |f(xn)|>ε for all n. Since f is uniformly continuous, there exists 0<δ<1, such that for all n, if |xxn|<δ, then |f(xn)f(x)|<ε/2. In particular,
|xn+δxnδf(x)dx|δεfor all n,
contradicting the Cauchy criterion.

No comments:

Post a Comment