Series

Arun Ram
Department of Mathematics and Statistics
University of Melbourne
Parkville, VIC 3010 Australia
aram@unimelb.edu.au

Last updates: 2 June 2012

Series

Let X be a topological group with operation addition and let (a1, a2, a3, ) be a sequence in X.

The following result establishes what is necessary to see that radius of convergence is a good notion for power series.

Let (a1, a2,a3, ) be a sequence in or . Let r,s.

  1. (a) Let r,s. Assume n=0 ansn converges. If |r| < |s| then n=0 an |r|n converges.
  2. (b) If n=1 |an| converges then n=1 an converges.

Proof of (a).

Proof of (b).
Let An =|a1| +|a2| ++ |an| and sn =a1+a2 ++an.
Since n=1 |an| =(A1, A2,A3, ) converges, the sequence (A1, A2,A3, ) is Cauchy.
Since | sm-sn| =|an+1 ++am| |am+1 | ++ |an| = | Am-An |, the sequence (s1, s2,s3, ) is Cauchy.
Since Cauchy sequences converge in and (in any complete metric space), the sequence (s1,s2, ) = n=1 an converges.

Let (an) be a sequence in 0.

  1. Assume limn (an+1 an) =a exists and a<1. Then n=1 an converges.
  2. Assume limn (an+1 an) =a exists and a>1. Then n=1 an diverges.

Proof.
(a) Assume limn an+1 an=a exists and a<1.
Let ε >0 so that a+ε<1.
Since limn an+1 an=a, there exists N >0 such that if n >0 and n>N then an+1 an<a+ε.
Then n=1 an = a1+a2 ++aN+ aN+1+ aN+2+ = a1++ aN+ aN+1+ aN+1 ( aN+2 aN+1 ) +aN+1 ( aN+2 aN+1 ) ( aN+3 aN+2 ) + < a1++ aN+ aN+1+ aN+1 (a+ε) +aN+1 (a+ε)2 + = a1++ aN+ aN+1 (1+ (a+ε) + (a+ε)2 +(a+ε)3 +) = a1++ aN+ aN+1 ( 11-(a+ε) ). So n=1 an converges.

(b) Assume limn an+1 an=a exists and a>1.
Let ε >0 so that a-ε<1.
Since limn an+1 an there exists N >0 such that if n >0 and n>N then an+1 an>a-ε.
Then n=1 an = a1+a2 ++aN+ aN+1+ aN+2+ = a1++ aN+ aN+1+ aN+1 ( aN+2 aN+1) +aN+1 (aN+2 aN+1) (aN+3 aN+2) + > a1++ aN+ aN+1+ aN+1 (a-ε) +aN+1 (a-ε)2 + > a1++ aN+ aN+1 (1+1+1+). So n+1 an diverges.

Leibniz's theorem. If (a1, a2,a3,) is a sequence in such that

  1. an 0,
  2. If n>0 then an an+1,
  3. If limn an=0,
then n=1 (-1) n-1an converges.

Proof.

  1. (a) If (a1, a2,a3, ) is a sequence in , n=1 an=a, and n=1 |an| converges absolutely
    then every rearrangement of n=1 an converges to a.
  2. (b) Assume that (a1, a2,a3, ) is a sequence in . Let l. If n=1 |an| converges conditionally (and not absolutely) then there exists a rearrangement of (an) which converges to l.

Harmonic series and the Riemann zeta function

If k=1, n=1 1n=1+ 12+( 13+14) +(15+ 16+17 +18)+ > 1+12+ 12+12 + So n=1 1n diverges.

Example. n=1 1n2 = 1+( 122+ 132) +( 142+ 152+ 162+ 172) +(182 +) < 1+222 +442+ 882 + = 1+12+ 14+18 + = 1+12+ (12)2 +(12)3 + = 11-12 =2. In fact, according to Wolfram Alpha, n=1 1n2 =π26. If k>1 then n=1 1nk = 1+(12k +13k) +(14k+15k+ 16k+ 17k) +(18k +) < 1+22k +44k+ 88k+ = 1+1 2k-1 +14k-1 +18k-1 + = 1+ 12k-1 +( 12k-1 )2 +( 12k-1 )3 + = 11- 12k-1 = 2k-1 2k-1-1 . So n=1 1nk converges.

If k<1 then n=1 1nk = 1+12k +13k +14k + > 1+12 +13+14 +. So n=1 1nk diverges.

Let k >0. n=1 1nk converges if k>1, and n=1 1nk diverges if k1.

Let s.

Example. ζ(2)= π26 .

Notes and References

This section proves the popular series theorems for a undergraduate calculus course.

PERHAPS WE SHOULD STILL PUT IN INTEGRAL TESTS??? or are these too obvious, and should really just be done on examples????

A traditional reference for this material is [Ru1, Ch. 3 and 4] with Theorem (rearr) being [Ru1, Theorem 3.54 and 3.55], the Ratio test is [Ru1, Theorem 3.34], the Leibniz rule is (somewhat hidden) in [Ru1, Theorem 3.42 and Theorem 3.43]. The discussion of harmonic series is in [Ru1, Theorem 3.28].

Another traditional reference for this material is [TF, Chapter 16, particularly 16-5]. An advantage of this reference is that it has a wealth of examples.

PUT IN REFERENCES TO BOURBAKI.

WHERE WILL WE PUT SERIES EXAMPLES???

PUT IN A PROPER DEFINITION OF RADIUS OF CONVERGENCE.

References

[Bou] N. Bourbaki, Algèbre, Chapitre ?: ??????????? MR?????.

[Ru1] G.B. Thomas and R.L. Finney, Calculus and Analytic Geometry, Fifth edition, Addison Wesley, 1980. MR?????.

[Ru1] W. Rudin, Principles of mathematical analysis, McGraw-Hill, 1976. MR0385023

[Ru2] W. Rudin, Real and complex analysis, Third edition, McGraw-Hill, 1987. MR0924157.

page history