When the equivalence rule for improper integrals fails
Statement
1. (a) For , study the convergence of .
(b) Study the convergence of .
2. (a) Show that diverges, so is not absolutely convergent.
(b) Let and . Show that at but that converges while diverges.
(c) Explain why the equivalence rule for improper integrals fails here.
Hint
For 1(a), integrate by parts with , . For 1(b), linearize . For 2(a), use . For 2(c), recall that the equivalence rule requires functions of constant sign.
Show solutionHide solution
Solution
(a) Convergence of for . Integrate by parts on :
Since and , the integral on the right converges absolutely. The boundary term . Hence converges.
(b) Divergence of . Linearization:
By (a) with , converges. But diverges (Riemann with ). A sum of a convergent and a divergent integral diverges, so diverges.
(2a) Non-absolute convergence. Since ,
By comparison with (1b), diverges. Hence converges (by (a) with ) but not absolutely: it is conditionally convergent.
(2b) An equivalence that misleads. Let and . Then
so at . However, by (a) with , converges. But diverges (sum of a convergent and a divergent integral by (1b)). So the equivalence of integrands does not preserve convergence.
(2c) Why the equivalence rule fails. The rule " and have the same nature" requires that and have a constant sign near . Here, the oscillations of change sign infinitely often, so the comparison fails: even though , the small "correction" accumulates a divergent contribution, since is not sign-changing.