Let f be piecewise continuous on [a,b) (b∈R or +∞). The integral converges when limx→b−∫axf exists; one then writes ∫abf for the limit. (Similarly on (a,b], and on (a,b) by splitting at an interior point — the choice does not matter, by Chasles.) The integral converges absolutely when ∫ab∣f∣ converges; absolute convergence implies convergence, by the Cauchy criterion:
∫xyf≤∫xy∣f∣
and completeness of R (the primitive has the Cauchy property). In detail: let F(x)=∫axf and G(x)=∫ax∣f∣. If ∫b∣f∣ converges, G has a limit at b−, so for every ε>0 there is c<b with G(y)−G(x)≤ε whenever c≤x≤y<b; the display transfers this Cauchy property to F. For any sequence xn→b− the values F(xn) then form a Cauchy sequence of reals, convergent by completeness, and interlacing two such sequences shows the limit is the same for all of them: F has a limit at b−.
Examples
Example 9.3(Two warm-ups, worked to the end)
(a)∫01lntdt: the integrand blows up at 0+, but ∣lnt∣=o(t−1/2) there (logarithms lose to powers), and ∫0t−1/2converges: absolute convergence. The value, by parts on [ε,1]:
∫ε1lntdt=[tlnt−t]ε1=−1−εlnε+εε→0+−1.
(b)∫0∞1+t2lntdt: trouble at both ends, so split at 1. Near 0: ∣lnt∣ integrable as in (a); near ∞: 1+t2lnt=o(t−3/2): absolutely convergent. The substitution t=u1 maps (0,1) onto (1,∞) and
the two halves cancel, and the integral is 0. Closing insight: symmetry under t↦t1 is worth a page of computation — the same trick already powered Exercise 9.3.
Example 9.4(One value, three integrals)
Study I=∫0∞t21−costdt. At 0: 1−cost∼2t2, so the integrand extends continuously by the value 21 — no singularity at all. At ∞: 0≤t21−cost≤t22: absolute convergence (Theorem 9.2). Value: integrate by parts on [ε,M] with u=1−cost, v′=t−2:
∫εMt21−costdt=[−t1−cost]εM+∫εMtsintdt.
The bracket vanishes at both ends (ε1−cosε∼2ε; bounded numerator at M), and the integral tends to the Dirichlet value 2π (Exercise 9.10): I=2π. Closing insight: with 1−cost=2sin22t and u=2t,
I=∫0∞(2u)22sin2u2du=∫0∞(usinu)2du:
the three classics ∫0∞tsintdt, ∫0∞(tsint)2dt (Exercise 9.11) and I all share the value 2π, passed around by parts and substitution — and only the first is semi-convergent: integration by parts traded away the absolute convergence for a simpler integrand.
where the last integral converges by the same integration by parts as above (with sin2t in the bracket): a divergent piece minus a convergent one diverges. So ∫1∞tsintdtconverges without converging absolutely — the integral analogue of the alternating series, with integration by parts playing the role of the alternating test.