2 1. Preliminary Notions One can also obtain a more accurate asymptotic expression in the case where the function f is decreasing. In fact, one can prove that if f : [1, ∞) → R+ is continuous, decreasing and such that limx→∞ f(x) = 0, then there exists a constant A such that (1.2) 1≤n≤x f(n) = x 1 f(t) dt + A + O(f(x)). To prove this result, compare the areas provided by expressions ∑ 1≤n≤x f(n) and x 1 f(t) dt. We would like to show that the expression D(N) := N 1 f(t) dt − 2≤n≤N f(n), where N is a positive integer, tends to a positive constant as N → ∞. First, it is clear that D(N) ≥ 0 for each integer N ≥ 2. So let R(N) = ∞ n=N+1 n n−1 f(t) dt − f(n) . In order to prove (1.2), it is suﬃcient to show that R(N) = O(f(N)). But for each pair of positive integers M and N with M ≥ N + 3, we have M−1 n=N+1 f(n) + f(M) M N f(t) dt f(N) + M−1 n=N+1 f(n), so that 0 M n=N+1 n n−1 f(t) dt − f(n) f(N) − f(M) f(N). It follows from this that 0 ≤ ∞ n=N+1 n n−1 f(t) dt − f(n) ≤ f(N), which implies that R(N) = O(f(N)), thereby establishing formula (1.2). 1.2. The Euler-MacLaurin formula We saw in the previous section that one could approximate the sum of a function by an integral, provided this function was monotonic. Here, we will see that if the function has a continuous derivative, then a more precise approximation can be obtained.

Purchased from American Mathematical Society for the exclusive use of nofirst nolast (email unknown) Copyright 2012 American Mathematical Society. Duplication prohibited. Please report unauthorized use to cust-serv@ams.org. Thank You! Your purchase supports the AMS' mission, programs, and services for the mathematical community.