Chapter 1.3

Numerical Sequences and Series

[25/0%]
Show
LaTeX
Problem 1.3.1

Prove that convergence of {sn}\left\{ s_{n}\right\} implies convergence of {∣sn∣}\left\{ \left|s_{n}\right|\right\}. Is the converse true?

?
Problem 1.3.2

Calculate lim⁡(n2+n−n)\lim \left(\sqrt{n^{2}+n}-n\right).

?
Problem 1.3.3

If s1=2s_{1} = \sqrt{2}, and

sn+1=2+sn(n=1,2,3,…) s_{n+1} = \sqrt{2+\sqrt{s_{n}}} \quad (n = 1,2,3, \ldots )

prove that {sn}\left\{ s_{n}\right\} converges, and that sn<2s_{n} < 2 for n=1,2,3,…n = 1,2,3, \ldots.

?
Problem 1.3.4

Find the upper and lower limits of the sequence {sn}\left\{ s_{n}\right\} defined by

s1=0;s2m=s2m−12;s2m+1=12+s2m s_{1} = 0 ; \quad s_{2 m} = \frac{s_{2 m-1}}{2} ; \quad s_{2 m+1} = \frac{1}{2}+s_{2 m}
?
Problem 1.3.5

For any two real sequences {an},{bn}\left\{ a_{n}\right\} ,\left\{ b_{n}\right\}, prove that

lim sup⁡n→∞(an+bn)≤lim sup⁡n→∞an+lim sup⁡n→∞bn, \limsup _{n \rightarrow \infty }\left(a_{n}+b_{n}\right) \leq \limsup _{n \rightarrow \infty } a_{n}+\limsup _{n \rightarrow \infty } b_{n},

provided the sum on the right is not of the form ∞−∞\infty -\infty.

?
Problem 1.3.6

Investigate the behavior (convergence or divergence) of Σan\Sigma a_{n} if

?
(a)

an=n+1−na_{n} = \sqrt{n+1}-\sqrt{n};

(b)

an=n+1−nna_{n} = \frac{\sqrt{n+1}-\sqrt{n}}{n};

(c)

an=(nn−1)na_{n} = (\sqrt[n]{n}-1)^{n};

(d)

an=11+zna_{n} = \frac{1}{1+z^{n}}, for complex values of zz.

Problem 1.3.7

Prove that the convergence of Σan\Sigma a_{n} implies the convergence of

Σann \Sigma \frac{\sqrt{a_{n}}}{n}

if an≥0a_{n} \geq 0.

?
Problem 1.3.8

If Σan\Sigma a_{n} converges, and if {bn}\left\{ b_{n}\right\} is monotonic and bounded, prove that Σanbn\Sigma a_{n} b_{n} converges.

?
Problem 1.3.9

Find the radius of convergence of each of the following power series:

?
(a)

∑n3zn\sum n^{3} z^{n},

(b)

∑2nn!zn\sum \frac{2^{n}}{n!} z^{n},

(c)

∑2nn2zn\sum \frac{2^{n}}{n^2} z^{n},

(d)

∑n33nzn\sum \frac{n^{3}}{3^{n}} z^{n}.

Problem 1.3.10

Suppose that the coefficients of the power series ∑anzn\sum a_{n} z^{n} are integers, infinitely many of which are distinct from zero. Prove that the radius of convergence is at most 1.

?
Problem 1.3.11

Suppose an>0,sn=a1+⋯+ana_{n} > 0, s_{n} = a_{1}+\cdots +a_{n}, and Σan\Sigma a_{n} diverges.

?
(a)

Prove that ∑an1+an\sum \frac{a_{n}}{1+a_{n}} diverges.

(b)

Prove that

aN+1sN+1+⋯+aN+ksN+k≥1−sNsN+k \frac{a_{N+1}}{s_{N+1}}+\cdots +\frac{a_{N+k}}{s_{N+k}} \geq 1-\frac{s_{N}}{s_{N+k}}

and deduce that ∑ansn\sum \frac{a_{n}}{s_{n}} diverges.

(c)

Prove that

ansn2≤1sn−1−1sn \frac{a_{n}}{s_{n}^{2}} \leq \frac{1}{s_{n-1}}-\frac{1}{s_{n}}

and deduce that ∑ansn2\sum \frac{a_{n}}{s_{n}^{2}} converges.

(d)

What can be said about

Σan1+nan and Σan1+n2an ?  \Sigma \frac{a_{n}}{1+n a_{n}} \text{ and } \Sigma \frac{a_{n}}{1+n^{2} a_{n}} \text{ ? }
Problem 1.3.12

Suppose an>0a_{n} > 0 and Σan\Sigma a_{n} converges. Put

rn=∑m=n∞am r_{n} = \sum _{m = n}^{\infty } a_{m}
?
(a)

Prove that

amrm+⋯+anrn>1−rnrm \frac{a_{m}}{r_{m}}+\cdots +\frac{a_{n}}{r_{n}} > 1-\frac{r_{n}}{r_{m}}

if m<nm < n, and deduce that ∑anrn\sum \frac{a_{n}}{r_{n}} diverges.

(b)

Prove that

anrn<2(rn−rn+1) \frac{a_{n}}{\sqrt{r_{n}}} < 2\left(\sqrt{r_{n}}-\sqrt{r_{n+1}}\right)

and deduce that ∑anrn\sum \frac{a_{n}}{\sqrt{r_{n}}} converges.

Problem 1.3.13

Prove that the Cauchy product of two absolutely convergent series converges absolutely.

?
Problem 1.3.14

If {sn}\left\{ s_{n}\right\} is a complex sequence, define its arithmetic means σn\sigma_{n} by

σn=s0+s1+⋯+snn+1(n=0,1,2,…) \sigma _{n} = \frac{s_{0}+s_{1}+\cdots +s_{n}}{n+1} \quad (n = 0,1,2, \ldots )
?
(a)

If lim⁡sn=s\lim s_{n} = s, prove that lim⁡σn=s\lim \sigma_{n} = s.

(b)

Construct a sequence {sn}\left\{ s_{n}\right\} which does not converge, although lim⁡σn=0\lim \sigma_{n} = 0.

(c)

Can it happen that sn>0s_{n} > 0 for all nn and that lim⁡sup⁡sn=∞\lim \sup s_{n} = \infty, although lim⁡σn=0\lim \sigma_{n} = 0?

(d)

Put an=sn−sn−1a_{n} = s_{n}-s_{n-1}, for n≥1n \geq 1. Show that

sn−σn=1n+1∑k=1nkak s_{n}-\sigma _{n} = \frac{1}{n+1} \sum _{k = 1}^{n} k a_{k}

Assume that lim⁡(nan)=0\lim \left(n a_{n}\right) = 0 and that {σn}\left\{ \sigma_{n}\right\} converges. Prove that {sn}\left\{ s_{n}\right\} converges.

(e)

Derive the last conclusion from a weaker hypothesis: Assume M<∞M < \infty, ∣nan∣≤M\left|n a_{n}\right| \leq M for all nn, and lim⁡σn=σ\lim \sigma_{n} = \sigma. Prove that lim⁡sn=σ\lim s_{n} = \sigma, by completing the following outline:

If m<nm < n, then

sn−σn=m+1n−m(σn−σm)+1n−m∑i=m+1n(sn−si) s_{n}-\sigma _{n} = \frac{m+1}{n-m}\left(\sigma _{n}-\sigma _{m}\right)+\frac{1}{n-m} \sum _{i = m+1}^{n}\left(s_{n}-s_{i}\right)

For these ii,

∣sn−si∣≤(n−i)Mi+1≤(n−m−1)Mm+2 \left|s_{n}-s_{i}\right| \leq \frac{(n-i) M}{i+1} \leq \frac{(n-m-1) M}{m+2}

Fix ε>0\varepsilon > 0 and associate with each nn the integer mm that satisfies

m≤n−ε1+ε<m+1 m \leq \frac{n-\varepsilon }{1+\varepsilon } < m+1

Then (m+1)/(n−m)≤1/ε(m+1) /(n-m) \leq 1 / \varepsilon and ∣sn−si∣<Mε\left|s_{n}-s_{i}\right| < M \varepsilon. Hence

lim sup⁡n→∞∣sn−σ∣≤Mε \limsup _{n \rightarrow \infty }\left|s_{n}-\sigma \right| \leq M \varepsilon

Since ε\varepsilon was arbitrary, lim⁡sn=σ\lim s_{n} = \sigma.

Remark.
?

Part (d) gives a converse of part (a), but under the additional assumption that nan→0n a_{n} \rightarrow 0.

Problem 1.3.15

Definition 3.21 can be extended to the case in which the ana_{n} lie in some fixed Rk\mathbb {R}^{k}. Absolute convergence is defined as convergence of Σ∣an∣\Sigma \left|\mathbf{a}_{n}\right|. Show that Theorems 3.22,3.23,3.25(a),3.33,3.34,3.42,3.45,3.473.22,3.23,3.25(a), 3.33,3.34,3.42,3.45,3.47, and 3.55 are true in this more general setting. (Only slight modifications are required in any of the proofs).

Definition 3.21: given a sequence {an}\left\{ a_{n}\right\}, with {an}\left\{ a_{n}\right\} one associates the sequence of partial sums sn=∑k=1naks_{n} = \sum_{k=1}^{n} a_{k}; the series ∑n=1∞an\sum_{n=1}^{\infty } a_{n} (also written Σan\Sigma a_{n}) converges to ss, written ∑n=1∞an=s\sum_{n=1}^{\infty } a_{n} = s, if {sn}\left\{ s_{n}\right\} converges to ss; if {sn}\left\{ s_{n}\right\} diverges, the series is said to diverge.

The cited results, as stated in the book for (real or complex) numbers.

Theorem 3.22: Σan\Sigma a_{n} converges if and only if for every ε>0\varepsilon > 0 there is an integer NN such that ∣∑k=nmak∣≤ε\left|\sum_{k=n}^{m} a_{k}\right| \leq \varepsilon if m≥n≥Nm \geq n \geq N.

Theorem 3.23: If Σan\Sigma a_{n} converges, then lim⁡n→∞an=0\lim_{n \rightarrow \infty } a_{n} = 0.

Theorem 3.25(a): If ∣an∣≤cn\left|a_{n}\right| \leq c_{n} for n≥N0n \geq N_{0}, where N0N_{0} is some fixed integer, and if Σcn\Sigma c_{n} converges, then Σan\Sigma a_{n} converges.

Theorem 3.33 (Root Test): Given Σan\Sigma a_{n}, put α=lim sup⁡n→∞∣an∣n\alpha = \limsup_{n \rightarrow \infty } \sqrt[n]{\left|a_{n}\right|}. Then (a) if α<1\alpha < 1, Σan\Sigma a_{n} converges; (b) if α>1\alpha > 1, Σan\Sigma a_{n} diverges; (c) if α=1\alpha = 1, the test gives no information.

Theorem 3.34 (Ratio Test): The series Σan\Sigma a_{n} (a) converges if lim sup⁡n→∞∣an+1/an∣<1\limsup_{n \rightarrow \infty } \left|a_{n+1} / a_{n}\right| < 1, (b) diverges if ∣an+1/an∣≥1\left|a_{n+1} / a_{n}\right| \geq 1 for all n≥n0n \geq n_{0}, where n0n_{0} is some fixed integer.

Theorem 3.42: Suppose (a) the partial sums AnA_{n} of Σan\Sigma a_{n} form a bounded sequence; (b) b0≥b1≥b2≥⋯b_{0} \geq b_{1} \geq b_{2} \geq \cdots; (c) lim⁡n→∞bn=0\lim_{n \rightarrow \infty } b_{n} = 0. Then Σanbn\Sigma a_{n} b_{n} converges.

Theorem 3.45: If Σan\Sigma a_{n} converges absolutely, then Σan\Sigma a_{n} converges.

Theorem 3.47: If Σan=A\Sigma a_{n} = A, and Σbn=B\Sigma b_{n} = B, then Σ(an+bn)=A+B\Sigma \left(a_{n}+b_{n}\right) = A+B, and Σcan=cA\Sigma c a_{n} = c A, for any fixed cc.

Theorem 3.55: If Σan\Sigma a_{n} is a series of complex numbers which converges absolutely, then every rearrangement of Σan\Sigma a_{n} converges, and they all converge to the same sum. (A rearrangement is Σan′\Sigma a_{n}^{\prime } with an′=akna_{n}^{\prime } = a_{k_{n}}, where {kn}\left\{ k_{n}\right\} is a sequence in which every positive integer appears once and only once.)

?
Problem 1.3.16

Fix a positive number α\alpha. Choose x1>αx_{1} > \sqrt{\alpha }, and define x2,x3,x4,…x_{2}, x_{3}, x_{4}, \ldots, by the recursion formula

xn+1=12(xn+αxn) x_{n+1} = \frac{1}{2}\left(x_{n}+\frac{\alpha }{x_{n}}\right)
?
(a)

Prove that {xn}\left\{ x_{n}\right\} decreases monotonically and that lim⁡xn=α\lim x_{n} = \sqrt{\alpha }.

(b)

Put εn=xn−α\varepsilon_{n} = x_{n}-\sqrt{\alpha }, and show that

εn+1=εn22xn<εn22α \varepsilon _{n+1} = \frac{\varepsilon _{n}^{2}}{2 x_{n}} < \frac{\varepsilon _{n}^{2}}{2 \sqrt{\alpha }}

so that, setting β=2α\beta = 2 \sqrt{\alpha },

εn+1<β(ε1β)2n(n=1,2,3,…) \varepsilon _{n+1} < \beta \left(\frac{\varepsilon _{1}}{\beta }\right)^{2^{n}} \quad (n = 1,2,3, \ldots )
(c)

This is a good algorithm for computing square roots, since the recursion formula is simple and the convergence is extremely rapid. For example, if α=3\alpha = 3 and x1=2x_{1} = 2, show that ε1/β<110\varepsilon_{1} / \beta < \frac{1}{10} and that therefore

ε5<4⋅10−16,ε6<4⋅10−32 \varepsilon _{5} < 4 \cdot 10^{-16}, \quad \varepsilon _{6} < 4 \cdot 10^{-32}
Problem 1.3.17

Fix α>1\alpha > 1. Take x1>αˉx_{1} > \sqrt{ } \bar{\alpha }, and define

xn+1=α+xn1+xn=xn+α−xn21+xn x_{n+1} = \frac{\alpha +x_{n}}{1+x_{n}} = x_{n}+\frac{\alpha -x_{n}^{2}}{1+x_{n}}
?
(a)

Prove that x1>x3>x5>⋯x_{1} > x_{3} > x_{5} > \cdots.

(b)

Prove that x2<x4<x6<⋯x_{2} < x_{4} < x_{6} < \cdots.

(c)

Prove that lim⁡xn=α\lim x_{n} = \sqrt{\alpha }.

(d)

Compare the rapidity of convergence of this process with the one described in Exercise 16.

Problem 1.3.18

Replace the recursion formula of Exercise 16 by

xn+1=p−1pxn+αpxn−p+1 x_{n+1} = \frac{p-1}{p} x_{n}+\frac{\alpha }{p} x_{n}^{-p+1}

where pp is a fixed positive integer, and describe the behavior of the resulting sequences {xn}\left\{ x_{n}\right\}.

?
Problem 1.3.19

Associate to each sequence a={αn}a = \left\{ \alpha_{n}\right\}, in which αn\alpha_{n} is 0 or 2, the real number

x(a)=∑n=1∞αn3n x(a) = \sum _{n = 1}^{\infty } \frac{\alpha _{n}}{3^{n}}

Prove that the set of all x(a)x(a) is precisely the Cantor set described in Sec. 2.44.

?
Problem 1.3.20

Suppose {pn}\left\{ p_{n}\right\} is a Cauchy sequence in a metric space XX, and some subsequence {pnt}\left\{ p_{n t}\right\} converges to a point p∈Xp \in X. Prove that the full sequence {pn}\left\{ p_{n}\right\} converges to pp.

?
Problem 1.3.21

Prove the following analogue of Theorem 3.10(b)3.10(b) : If {En}\left\{ E_{n}\right\} is a sequence of closed nonempty and bounded sets in a complete metric space XX, if En⊃En+1E_{n} \supset E_{n+1}, and if

lim⁡n→∞diam⁡En=0 \lim _{n \rightarrow \infty } \operatorname {diam} E_{n} = 0

then ⋂1∞En\bigcap_{1}^{\infty } E_{n} consists of exactly one point.

?
Problem 1.3.22

Suppose XX is a nonempty complete metric space, and {Gn}\left\{ G_{n}\right\} is a sequence of dense open subsets of XX. Prove Baire's theorem, namely, that ⋂1∞Gn\bigcap_{1}^{\infty } G_{n} is not empty. (In fact, it is dense in XX). Hint: Find a shrinking sequence of neighborhoods EnE_{n} such that En⊂GnE_{n} \subset G_{n}, and apply Exercise 21.

?
Problem 1.3.23

Suppose {pn}\left\{ p_{n}\right\} and {qn}\left\{ q_{n}\right\} are Cauchy sequences in a metric space XX. Show that the sequence {d(pn,qn)}\left\{ d\left(p_{n}, q_{n}\right)\right\} converges. Hint: For any m,nm, n,

d(pn,qn)≤d(pn,pm)+d(pm,qm)+d(qm,qn) d\left(p_{n}, q_{n}\right) \leq d\left(p_{n}, p_{m}\right)+d\left(p_{m}, q_{m}\right)+d\left(q_{m}, q_{n}\right)

it follows that

∣d(pn,qn)−d(pm,qm)∣ \left|d\left(p_{n}, q_{n}\right)-d\left(p_{m}, q_{m}\right)\right|

is small if mm and nn are large.

?
Problem 1.3.24

Let XX be a metric space.

?
(a)

Call two Cauchy sequences {pn},{qn}\left\{ p_{n}\right\} ,\left\{ q_{n}\right\} in XX equivalent if

lim⁡n→∞d(pn,qn)=0 \lim _{n \rightarrow \infty } d\left(p_{n}, q_{n}\right) = 0

Prove that this is an equivalence relation.

(b)

Let X∗X^{*} be the set of all equivalence classes so obtained. If P∈X∗,Q∈X∗P \in X^{*}, Q \in X^{*}, {pn}∈P,{qn}∈Q\left\{ p_{n}\right\} \in P,\left\{ q_{n}\right\} \in Q, define

Δ(P,Q)=lim⁡n→∞d(pn,qn) \Delta (P, Q) = \lim _{n \rightarrow \infty } d\left(p_{n}, q_{n}\right)

by Exercise 23, this limit exists. Show that the number Δ(P,Q)\Delta (P, Q) is unchanged if {pn}\left\{ p_{n}\right\} and {qn}\left\{ q_{n}\right\} are replaced by equivalent sequences, and hence that Δ\Delta is a distance function in X∗X^{*}.

(c)

Prove that the resulting metric space X∗X^{*} is complete.

(d)

For each p∈Xp \in X, there is a Cauchy sequence all of whose terms are pp; let PpP_{p} be the element of X∗X^{*} which contains this sequence. Prove that

Δ(Pp,Pq)=d(p,q) \Delta \left(P_{p}, P_{q}\right) = d(p, q)

for all p,q∈Xp, q \in X. In other words, the mapping φ\varphi defined by φ(p)=Pp\varphi (p) = P_{p} is an isometry (i.e., a distance-preserving mapping) of XX into X∗X^{*}.

(e)

Prove that φ(X)\varphi (X) is dense in X∗X^{*}, and that φ(X)=X∗\varphi (X) = X^{*} if XX is complete. By (d), we may identify XX and φ(X)\varphi (X) and thus regard XX as embedded in the complete metric space X∗X^{*}. We call X∗X^{*} the completion of XX.

Problem 1.3.25

Let XX be the metric space whose points are the rational numbers, with the metric d(x,y)=∣x−y∣d(x, y) = \left|x-y\right|. What is the completion of this space? (Compare Exercise 24).

?