Dirichlet's test#Improper integrals

{{Short description|Test for series convergence}}

{{Calculus |Series}}

In mathematics, Dirichlet's test is a method of testing for the convergence of a series that is especially useful for proving conditional convergence. It is named after its author Peter Gustav Lejeune Dirichlet, and was published posthumously in the Journal de Mathématiques Pures et Appliquées in 1862.Démonstration d’un théorème d’Abel. Journal de mathématiques pures et appliquées 2nd series, tome 7 (1862), [http://portail.mathdoc.fr/JMPA/afficher_notice.php?id=JMPA_1862_2_7_A43_0 pp. 253–255] {{Webarchive|url=https://web.archive.org/web/20110721011902/http://portail.mathdoc.fr/JMPA/afficher_notice.php?id=JMPA_1862_2_7_A43_0|date=2011-07-21}}. See also [http://www.numdam.org/item/JMPA_1862_2_7__253_0/].

Statement

The test states that if (a_n) is a monotonic sequence of real numbers with \lim_{n \to \infty} a_n = 0 and (b_n) is a sequence of real numbers or complex numbers with bounded partial sums, then the series

:\sum_{n=1}^{\infty} a_n b_n

converges.{{harvnb|Apostol|1967|pp=407–409}}{{harvnb|Spivak|2008|p=495}}{{Harvnb|Rudin|1976|p=70}}

Proof

Let S_n = \sum_{k=1}^n a_k b_k and B_n = \sum_{k=1}^n b_k.

From summation by parts, we have that S_n = a_{n} B_n + \sum_{k=1}^{n-1} B_k (a_k - a_{k+1}). Since the magnitudes of the partial sums B_n are bounded by some M and a_n \to 0 as n\to\infty, the first of these terms approaches zero: |a_{n} B_n| \leq |a_{n} M| \to 0 as n\to\infty.

Furthermore, for each k, |B_k (a_k - a_{k+1})| \leq M|a_k - a_{k+1}|.

Since (a_n) is monotone, it is either decreasing or increasing:

  • If (a_n) is decreasing,

    \sum_{k=1}^n M|a_k - a_{k+1}| = \sum_{k=1}^n M(a_k - a_{k+1}) = M\sum_{k=1}^n (a_k - a_{k+1}),

    which is a telescoping sum that equals M(a_1 - a_{n+1}) and therefore approaches Ma_1 as n \to \infty. Thus, \sum_{k=1}^\infty M(a_k - a_{k+1}) converges.

  • If (a_n) is increasing,

    \sum_{k=1}^n M|a_k - a_{k+1}| = -\sum_{k=1}^n M(a_k - a_{k+1}) = -M\sum_{k=1}^n (a_k - a_{k+1}),

    which is again a telescoping sum that equals -M(a_1 - a_{n+1}) and therefore approaches -Ma_1 as n\to\infty. Thus, again, \sum_{k=1}^\infty M(a_k - a_{k+1}) converges.

So, the series \sum_{k=1}^\infty B_k(a_k - a_{k+1}) converges by the direct comparison test to \sum_{k=1}^\infty M(a_k - a_{k+1}). Hence S_n converges.

Applications

A particular case of Dirichlet's test is the more commonly used alternating series test for the case{{Harvnb|Rudin|1976|p=71}}

b_n = (-1)^n \Longrightarrow\left|\sum_{n=1}^N b_n\right| \leq 1.

Another corollary is that \sum_{n=1}^\infty a_n \sin n converges whenever (a_n) is a decreasing sequence that tends to zero. To see that

\sum_{n=1}^N \sin n

is bounded, we can use the summation formula{{cite web | url=https://math.stackexchange.com/q/1119043 | title=Where does the sum of $\sin(n)$ formula come from? }}

\sum_{n=1}^N\sin n=\sum_{n=1}^N\frac{e^{in}-e^{-in}}{2i}=\frac{\sum_{n=1}^N (e^{i})^n-\sum_{n=1}^N (e^{-i})^n}{2i}=\frac{\sin 1 +\sin N-\sin (N+1)}{2- 2\cos 1}.

Improper integrals

An analogous statement for convergence of improper integrals is proven using integration by parts. If the integral of a function f is uniformly bounded over all intervals, and g is a bounded non-negative monotonically decreasing function, then the integral of fg is a convergent improper integral.

Notes

References

  • {{Cite book |last=Apostol |first=Tom M. |author-link=Tom M. Apostol |title=Calculus |publisher=John Wiley & Sons |year=1967 |isbn=0-471-00005-1 |edition=2nd |volume=1 |orig-year=1961}}
  • Hardy, G. H., A Course of Pure Mathematics, Ninth edition, Cambridge University Press, 1946. (pp. 379–380).
  • {{Cite book |last=Rudin |first=Walter |author-link=Walter Rudin |title=Principles of mathematical analysis |publisher=McGraw-Hill |year=1976 |isbn=0-07-054235-X |edition=3rd |location=New York |oclc=1502474 |orig-date=1953}}
  • {{Cite book |last=Spivak |first=Michael |author-link=Michael Spivak |title=Calculus |publisher=Publish or Perish |year=2008 |isbn=978-0-914098-91-1 |edition=4th |location=Houston, TX |orig-year=1967}}
  • Voxman, William L., Advanced Calculus: An Introduction to Modern Analysis, Marcel Dekker, Inc., New York, 1981. (§8.B.13–15) {{ISBN|0-8247-6949-X}}.