周次: 14 日期: May 25, 2022 节次: 1

In this lecture, we continue the discussion on the root test (for the series with non-negative terms). And we discuss alternating series, there’s a test called Leibniz’s or alternating series test. Alternating series converges in a manner that positive and negative terms cancelling each other by a large amount. With this fact, we define the absolute convergence and conditional convergence, and find them totally different from the rearrangement perspective.

The Root Test ( For the Series with Non-negative Terms)

The ratio test is convenient for the case when an+1/an is relatively simple. For more complicated series, we may need other test method. For example

Guiding Example

Let an={n/2n,n odd, 1/2n,n even, does nan converge?

This is not a geometric series, and the nth term approaches 0. So we do not know the series diverges. The Integral Test does not applicable as the sequence an is not decreasing. The Ratio Test produces

an+1an={12n,n odd n+12,n even

As n, the ratio has no limit.

A test that will answer the question is the Root Test.

Theorem 13. The Root Test

Let an be a series with an0 for nN, and suppose that

limnann=ρ.

Then

(a) the series converges if ρ<1

(b) the series diverges if ρ>1

(c) the test is inconclusive if ρ=1.

Proof.

Example 3. Applying the Root Test

(a) n=1n22n

( Converges )

(b) n=13n(1+1n)n2

( Diverges )

(c) n=1(11+n)n

( Converges )

Guiding Example (Revisited)

Let an={n/2n,n odd, 1/2n,n even, does nan converge?

( Converges )

Alternating Series

A series in which the terms are alternately positive and negative is an alternating series.

Examples (Alternating Series)

  • 112+1314++(1)n+1n+ (alternating harmonic series, converges)
  • 2+112+1418++(1)n42n+ (converges)
  • 12+34+56++(1)n+1n+ (diverges)

These are alternating series.

Theorem 14. (The Alternating Series Test , Leibniz’s Theorem)

If un,0 (starting from nN), then the alternating series

n=1(1)n+1un=u1u2+u3u4+

converges.

Proof.

Example 1. The alternating harmonic series

n=1(1)n+11n=112+1314+converges

Graphical Interpretation on Convergence

Why for un0, the series

u1u2+u3u4+u5

converges ?

Untitled

From the graphical interpretation, we also learned that if the sum for alternating series is L, then the distance of the partial sum sn and the total sum L satisfies

|Lsn|<un+1,,nN

Theorem 15. The Alternating Series Estimation Theorem

If the alternating series n(1)n+1un satisfies un0, then for nN the partial sum

sn=u1u2+u3+(1)n+1un

approximates the sum L with an error whose absolute value is less than un+1, the numerical value of the first unused term. Furthermore, the remainder Lsn has the same sign as the first unused term.

Example 2. Try Theorem 15 on the series whose sum we know

n(1)n12n=112+1418+116132+1641128|+1256

The sum of the series is 2/3. And the partial sum indicating by the | sign is 23(1(12)6).

The remainder, 23(12)8 , is positive and less than 128.

Absolute and Conditional Convergence

Definition. Absolutely Convergent

A series nan converges absolutely (is absolutely convergent) if the corresponding series of absolute values, |an|,converges.

Definition. Conditionally Convergent

A series that converges but does not converge absolutely converges conditionally.

Examples.

  • All convergent series with non-negative terms converge absolutely
  • 1+sin(1)12+sin(2)22+sin(3)32+ converges absolutely
  • 112+1314+ converges, but not absolutely. (Conditionally convergent)

Why Absolutely Convergent?

  1. It is easier to test. (We have a number of tests for the series with non-negative terms)
  2. Converges absolutely converges!

Theorem 16. The Absolute Convergence Test

If n|an| converges, then nan converges.

Proof.

Remark.

Absolutely convergent is a stronger than conditionally convergent. So a conditionally convergent series may fail to be an absolutely convergent series.

Example 3. Applying the Absolute Convergence Test

  1. n=1(1)nlnnn2 converges absolutely
  2. n=1(1)nlnnn converges conditionally

Example 4. Being Alternating is Good for Convergence

  1. If p is a positive constant, then

n=11np converges when p>1, but n=1(1)nnp converges when p>0.

n(1)nnp 0<p1 p>1
Converge ✔️ ✔️
Absolutely ✔️
Conditionally ✔️

The difference between absolute and conditional convergences is best shown from the following phenomenon.

Absolute Convergent Series: Rearrange the terms does not change the sum Conditional Convergent Series: Rearrange the terms DOES change the sum!

Rearranging Series

Theorem 17. The Rearrangement Theorem for Absolutely Convergent Series

If nan converges absolutely, and b1,b2,b3,,bn, is any rearrangement of the sequence an, then nbn converges absolutely and

nbn=nan.

Example 5. Rearrangement on Absolutely Convergent Series

114+19116+

Can be rearranged to

114116+19+125+14913616411001144+

And since the series absolutely convergent, it really doesn’t matter how we rearrange it. It keeps absolutely converging to the same value.

However, a rearrangement on a conditionally convergent series is more interesting!

Example 6. Rearranging the Alternating Harmonic Series

The alternating harmonic series is conditionally convergent series! Thus

112+1314+1516+

can be rearranged to reach different sums.

  • Rearrange to 1/2 of the original sum. Consider the following rearrangement

    (11214)+(131618)+(15110112)++(12n+114n+214n+4)+

    Note that the grouping partial sum has the following form

    n=0(12n+114n+214n+4)=n=0(14n+214n+4)=12n=0(12n+112n+2)

  • Rearrange to a divergent series. Since series of terms 12n1 diverges to +, and the series of terms 12n diverges to , we construct the following rearrangement

    $$ \scriptsize\underbrace{\overbrace{\underbrace{1+\frac13 + \frac15 + \cdots + \frac1{2m_1+1}}{\text{Larger than } 2} - \frac1{2}-\frac14 - \cdots - \frac{1}{2n_1 +1}}^{\text{smaller than }-3}+\frac{1}{2m_1+3}+\frac1{2m_1+5} + \cdots + \frac1{2m_2+1}}{\text{larger than }4}+\cdots $$

    such that the partial sum swings large in either direction

  • Rearrange to converge to 1. Making the partial sum swings about 1

    • Adds 1 (Making the partial sum just 1)
    • Subtracts 12 (Making the partial sum just down to a value < 1)
    • Adds 13, 15 (Making the partial sum just >1 again)
    • Subtracts 14 (Making the partial sum just < 1 again)
    • Adds 17, 19 (Making the partial sum just > 1 again)

    The new arrangement looks like

    112+13+1514+17+1916+111+11318+115+117110+119+121 112+123+125114+127116+

    Since the amount by which our partial sums exceed 1 or fall below it approaches zero, the new series converges to 1.

Summary

Untitled