Chapter 1.8

Some Special Functions

[31/0%]
Show
LaTeX
Problem 1.8.1

Define

f(x)={e−1/x2(x≠0)0(x=0) f(x) = \begin{cases} e^{-1 / x^{2}} & (x \neq 0) \\ 0 & (x = 0)\end{cases}

Prove that ff has derivatives of all orders at x=0x = 0, and that f(n)(0)=0f^{(n)}(0) = 0 for n=1,2,3,…n = 1,2,3, \ldots.

?
Problem 1.8.2

Let aija_{i j} be the number in the ii th row and jj th column of the array so that

aij={0(i<j)−1(i=j)2j−i(i>j) a_{i j} = \begin{cases} 0 & (i < j) \\ -1 & (i = j) \\ 2^{j-i} & (i > j)\end{cases}

Prove that

∑i∑jaij=−2,∑j∑iaij=0 \sum _{i} \sum _{j} a_{i j} = -2, \quad \sum _{j} \sum _{i} a_{i j} = 0
?
Problem 1.8.3

Prove that

∑i∑jaij=∑j∑iaij \sum _{i} \sum _{j} a_{i j} = \sum _{j} \sum _{i} a_{i j}

if aij≥0a_{i j} \geq 0 for all ii and jj (the case +∞=+∞+\infty = +\infty may occur).

?
Problem 1.8.4

Prove the following limit relations:

?
(a)

lim⁡x→0bx−1x=log⁡b(b>0)\lim_{x \rightarrow 0} \frac{b^{x}-1}{x} = \log b \quad (b > 0).

(b)

lim⁡x→0log⁡(1+x)x=1\lim_{x \rightarrow 0} \frac{\log (1+x)}{x} = 1.

(c)

lim⁡x→0(1+x)1/x=e\lim_{x \rightarrow 0}(1+x)^{1 / x} = e.

(d)

lim⁡n→∞(1+xn)n=ex\lim_{n \rightarrow \infty }\left(1+\frac{x}{n}\right)^{n} = e^{x}.

Problem 1.8.5

Find the following limits

?
(a)

lim⁡x→0e−(1+x)1/xx\lim_{x \rightarrow 0} \frac{e-(1+x)^{1 / x}}{x}.

(b)

lim⁡n→∞nlog⁡n[n1/n−1]\lim_{n \rightarrow \infty } \frac{n}{\log n}\left[n^{1 / n}-1\right].

(c)

lim⁡x→0tan⁡x−xx(1−cos⁡x)\lim_{x \rightarrow 0} \frac{\tan x-x}{x(1-\cos x)}.

(d)

lim⁡x→0x−sin⁡xtan⁡x−x\lim_{x \rightarrow 0} \frac{x-\sin x}{\tan x-x}.

Problem 1.8.6

Suppose f(x)f(y)=f(x+y)f(x) f(y) = f(x+y) for all real xx and yy.

?
(a)

Assuming that ff is differentiable and not zero, prove that

f(x)=ecx f(x) = e^{c x}

where cc is a constant.

(b)

Prove the same thing, assuming only that ff is continuous.

Problem 1.8.7

If 0<x<π20 < x < \frac{\pi }{2}, prove that

2π<sin⁡xx<1 \frac{2}{\pi } < \frac{\sin x}{x} < 1
?
Problem 1.8.8

For n=0,1,2,…n = 0,1,2, \ldots, and xx real, prove that

∣sin⁡nx∣≤n∣sin⁡x∣ \left|\sin n x\right| \leq n\left|\sin x\right|

Note that this inequality may be false for other values of nn. For instance,

∣sin⁡12π∣>12∣sin⁡π∣ \left|\sin \frac{1}{2} \pi \right| > \frac{1}{2}\left|\sin \pi \right|
?
Problem 1.8.9
?
(a)

Put sN=1+(12)+⋯+(1/N)s_{N} = 1+\left(\frac{1}{2}\right)+\cdots +(1 / N). Prove that

lim⁡N→∞(sN−log⁡N) \lim _{N \rightarrow \infty }\left(s_{N}-\log N\right)

exists. (The limit, often denoted by γ\gamma, is called Euler's constant. Its numerical value is 0.5772…0.5772 \ldots It is not known whether γ\gamma is rational or not).

(b)

Roughly how large must mm be so that N=10mN = 10^{m} satisfies sN>100s_{N} > 100?

Problem 1.8.10

Prove that ∑1/p\sum 1 / p diverges; the sum extends over all primes. (This shows that the primes form a fairly substantial subset of the positive integers).

Hint: Given NN, let p1,…,pkp_{1}, \ldots , p_{k} be those primes that divide at least one integer ≤N\leq N. Then

∑n=1N1n≤∏j=1k(1+1pj+1pj2+⋯ )=∏j=1k(1−1pj)−1≤exp⁡∑j=1k2pj \begin{aligned} \sum _{n = 1}^{N} \frac{1}{n} & \leq \prod _{j = 1}^{k}\left(1+\frac{1}{p_{j}}+\frac{1}{p_{j}^{2}}+\cdots \right) \\ & = \prod _{j = 1}^{k}\left(1-\frac{1}{p_{j}}\right)^{-1} \\ & \leq \exp \sum _{j = 1}^{k} \frac{2}{p_{j}} \end{aligned}

The last inequality holds bocause

(1−x)−1≤e2x (1-x)^{-1} \leq e^{2 x}

if 0≤x≤120 \leq x \leq \frac{1}{2}. (There are many proofs of this result. See, for instance, the article by I. Niven in Amer. Math. Monthly, vol. 78, 1971, pp. 272-273, and the one by R. Bellman in Amer. Math. Monthly, vol. 50, 1943, pp. 318-319).

?
Problem 1.8.11

Suppose f∈Rf \in \mathscr {R} on [0,A][0, A] for all A<∞A < \infty, and f(x)→1f(x) \rightarrow 1 as x→+∞x \rightarrow +\infty. Prove that

lim⁡t→0t∫0∞e−txf(x)dx=1(t>0) \lim _{t \rightarrow 0} t \int _{0}^{\infty } e^{-t x} f(x) d x = 1 \quad (t > 0)
?
Problem 1.8.12

Suppose 0<δ<π,f(x)=10 < \delta < \pi , f(x) = 1 if ∣x∣≤δ,f(x)=0\left|x\right| \leq \delta , f(x) = 0 if δ<∣x∣≤π\delta < \left|x\right| \leq \pi, and f(x+2π)=f(x)f(x+2 \pi ) = f(x) for all xx.

?
(a)

Compute the Fourier coefficients of ff.

(b)

Conclude that

∑n=1∞sin⁡(nδ)n=π−δ2(0<δ<π) \sum _{n = 1}^{\infty } \frac{\sin (n \delta )}{n} = \frac{\pi -\delta }{2} \quad (0 < \delta < \pi )
(c)

Deduce from Parseval's theorem that

∑n=1∞sin⁡2(nδ)n2δ=π−δ2 \sum _{n = 1}^{\infty } \frac{\sin ^{2}(n \delta )}{n^{2} \delta } = \frac{\pi -\delta }{2}
(d)

Let δ→0\delta \rightarrow 0 and prove that

∫0∞(sin⁡xx)2dx=π2 \int _{0}^{\infty }\left(\frac{\sin x}{x}\right)^{2} d x = \frac{\pi }{2}
(e)

Put δ=π/2\delta = \pi / 2 in (c). What do you get?

Problem 1.8.13

Put f(x)=xf(x) = x if 0≤x<2π0 \leq x < 2 \pi, and apply Parseval's theorem to conclude that

∑n=1∞1n2=π26 \sum _{n = 1}^{\infty } \frac{1}{n^{2}} = \frac{\pi ^{2}}{6}
?
Problem 1.8.14

If f(x)=(π−∣x∣)2f(x) = (\pi -\left|x\right|)^{2} on [−π,π][-\pi , \pi ], prove that

f(x)=π23+∑n=1∞4n2cos⁡nx f(x) = \frac{\pi ^{2}}{3}+\sum _{n = 1}^{\infty } \frac{4}{n^{2}} \cos n x

and deduce that

∑n=1∞1n2=π26,∑n=1∞1n4=π490 \sum _{n = 1}^{\infty } \frac{1}{n^{2}} = \frac{\pi ^{2}}{6}, \quad \sum _{n = 1}^{\infty } \frac{1}{n^{4}} = \frac{\pi ^{4}}{90}

(A recent article by E. L. Stark contains many references to series of the form ∑n−s\sum n^{-s}, where ss is a positive integer. See Math. Mag., vol. 47, 1974, pp. 197-202).

?
Problem 1.8.15

With DnD_{n} as defined in (77), put

KN(x)=1N+1∑n=0NDn(x) K_{N}(x) = \frac{1}{N+1} \sum _{n = 0}^{N} D_{n}(x)

Prove that

KN(x)=1N+1⋅1−cos⁡(N+1)x1−cos⁡x K_{N}(x) = \frac{1}{N+1} \cdot \frac{1-\cos (N+1) x}{1-\cos x}

and that

?
(a)

KN≥0K_{N} \geq 0

(b)

12π∫−ππKN(x)dx=1\frac{1}{2 \pi } \int_{-\pi }^{\pi } K_{N}(x) d x = 1,

(c)

KN(x)≤1N+1⋅21−cos⁡δK_{N}(x) \leq \frac{1}{N+1} \cdot \frac{2}{1-\cos \delta } if 0<δ≤∣x∣≤π0 < \delta \leq \left|x\right| \leq \pi

If sN=sN(f;x)s_{N} = s_{N}(f ; x) is the NN th partial sum of the Fourier series of ff, consider the arithmetic means

σN=s0+s1+⋯+sNN+1 \sigma _{N} = \frac{s_{0}+s_{1}+\cdots +s_{N}}{N+1}

Prove that

σN(f;x)=12π∫−ππf(x−t)KN(t)dt \sigma _{N}(f ; x) = \frac{1}{2 \pi } \int _{-\pi }^{\pi } f(x-t) K_{N}(t) d t

and hence prove Fejér's theorem: If ff is continuous, with period 2π2 \pi, then σN(f;x)→f(x)\sigma_{N}(f ; x) \rightarrow f(x) uniformly on [−π,π][-\pi , \pi ]. Hint: Use properties (a),(b),(c)(a),(b),(c) to proceed as in Theorem 7.26.

Problem 1.8.16

Prove a pointwise version of Fejér's theorem:

If f∈Rf \in \mathscr {R} and f(x+),f(x−)f(x+), f(x-) exist for some xx, then

lim⁡N→∞σN(f;x)=12[f(x+)+f(x−)] \lim _{N \rightarrow \infty } \sigma _{N}(f ; x) = \frac{1}{2}[f(x+)+f(x-)]
?
Problem 1.8.17

Assume ff is bounded and monotonic on [−π,π)[-\pi , \pi ), with Fourier coefficients cnc_{n}, as given by (62).

?
(a)

Use Exercise 17 of Chap. 6 to prove that {ncn}\left\{ n c_{n}\right\} is a bounded sequence.

(b)

Combine (a) with Exercise 16 and with Exercise 14(e) of Chap. 3, to conclude that

lim⁡N→∞sN(f;x)=12[f(x+)+f(x−)] \lim _{N \rightarrow \infty } s_{N}(f ; x) = \frac{1}{2}[f(x+)+f(x-)]

for every xx.

(c)

Assume only that f∈Rf \in \mathscr {R} on [−π,π][-\pi , \pi ] and that ff is monotonic in some segment (α,β)⊂[−π,π](\alpha , \beta ) \subset [-\pi , \pi ]. Prove that the conclusion of (b)(b) holds for every x∈(α,β)x \in (\alpha , \beta ). (This is an application of the localization theorem).

Problem 1.8.18

Define

f(x)=x3−sin⁡2xtan⁡xg(x)=2x2−sin⁡2x−xtan⁡x \begin{aligned} & f(x) = x^{3}-\sin ^{2} x \tan x \\ & g(x) = 2 x^{2}-\sin ^{2} x-x \tan x \end{aligned}

Find out, for each of these two functions, whether it is positive or negative for all x∈(0,π/2)x \in (0, \pi / 2), or whether it changes sign. Prove your answer.

?
Problem 1.8.19

Suppose ff is a continuous function on R1,f(x+2π)=f(x)\mathbb {R}^{1}, f(x+2 \pi ) = f(x), and α/π\alpha / \pi is irrational. Prove that

lim⁡N→∞1N∑n=1Nf(x+nα)=12π∫−ππf(t)dt \lim _{N \rightarrow \infty } \frac{1}{N} \sum _{n = 1}^{N} f(x+n \alpha ) = \frac{1}{2 \pi } \int _{-\pi }^{\pi } f(t) d t

for every xx. Hint: Do it first for f(x)=eikxf(x) = e^{i k x}.

?
Problem 1.8.20

The following simple computation yields a good approximation to Stirling's formula.

For m=1,2,3,…m = 1,2,3, \ldots, define

f(x)=(m+1−x)log⁡m+(x−m)log⁡(m+1) f(x) = (m+1-x) \log m+(x-m) \log (m+1)

if m≤x≤m+1m \leq x \leq m+1, and define

g(x)=xm−1+log⁡m g(x) = \frac{x}{m}-1+\log m

if m−12≤x<m+12m-\frac{1}{2} \leq x < m+\frac{1}{2}. Draw the graphs of ff and gg. Note that f(x)≤log⁡x≤g(x)f(x) \leq \log x \leq g(x) if x≥1x \geq 1 and that

∫1nf(x)dx=log⁡(n!)−12log⁡n>−18+∫1ng(x)dx \int _{1}^{n} f(x) d x = \log (n!)-\frac{1}{2} \log n > -\frac{1}{8}+\int _{1}^{n} g(x) d x

Integrate log⁡x\log x over [1,n][1, n]. Conclude that

78<log⁡(n!)−(n+12)log⁡n+n<1 \frac{7}{8} < \log (n!)-\left(n+\frac{1}{2}\right) \log n+n < 1

for n=2,3,4,…n = 2,3,4, \ldots (Note: log⁡2π∼0.918…\log \sqrt{2 \pi } \sim 0.918 \ldots) Thus

e7/8<n!(n/e)nn<e e^{7 / 8} < \frac{n!}{(n / e)^{n} \sqrt{n}} < e
?
Problem 1.8.21

Let

Ln=12π∫−ππ∣Dn(t)∣dt(n=1,2,3,…) L_{n} = \frac{1}{2 \pi } \int _{-\pi }^{\pi }\left|D_{n}(t)\right| d t \quad (n = 1,2,3, \ldots )

Prove that there exists a constant C>0C > 0 such that

Ln>Clog⁡n(n=1,2,3,…) L_{n} > C \log n \quad (n = 1,2,3, \ldots )

or, more precisely, that the sequence

{Ln−4π2log⁡n} \left\{ L_{n}-\frac{4}{\pi ^{2}} \log n\right\}

is bounded.

?
Problem 1.8.22

If α\alpha is real and −1<x<1-1 < x < 1, prove Newton's binomial theorem

(1+x)α=1+∑n=1∞α(α−1)⋯(α−n+1)n!xn (1+x)^{\alpha } = 1+\sum _{n = 1}^{\infty } \frac{\alpha (\alpha -1) \cdots (\alpha -n+1)}{n!} x^{n}

Hint: Denote the right side by f(x)f(x). Prove that the series converges. Prove that

(1+x)f′(x)=αf(x) (1+x) f^{\prime }(x) = \alpha f(x)

and solve this differential equation. Show also that

(1−x)−α=∑n=0∞Γ(n+α)n!Γ(α)xn (1-x)^{-\alpha } = \sum _{n = 0}^{\infty } \frac{\Gamma (n+\alpha )}{n!\Gamma (\alpha )} x^{n}

if −1<x<1-1 < x < 1 and α>0\alpha > 0.

?
Problem 1.8.23

Let γ\gamma be a continuously differentiable closed curve in the complex plane, with parameter interval [a,b][a, b], and assume that γ(t)≠0\gamma (t) \neq 0 for every t∈[a,b]t \in [a, b]. Define the index of γ\gamma to be

Ind⁡(γ)=12πi∫abγ′(t)γ(t)dt \operatorname {Ind}(\gamma ) = \frac{1}{2 \pi i} \int _{a}^{b} \frac{\gamma ^{\prime }(t)}{\gamma (t)} d t

Prove that Ind (γ)(\gamma ) is always an integer. Hint: There exists φ\varphi on [a,b][a, b] with φ′=γ′/γ,φ(a)=0\varphi^{\prime } = \gamma^{\prime } / \gamma , \varphi (a) = 0. Hence γexp⁡(−φ)\gamma \exp (-\varphi ) is constant. Since γ(a)=γ(b)\gamma (a) = \gamma (b) it follows that exp⁡φ(b)=exp⁡φ(a)=1\exp \varphi (b) = \exp \varphi (a) = 1. Note that φ(b)=2πi\varphi (b) = 2 \pi i Ind (γ)(\gamma ).

Compute Ind (γ)(\gamma ) when γ(t)=eint,a=0,b=2π\gamma (t) = e^{i n t}, a = 0, b = 2 \pi. Explain why Ind (γ)(\gamma ) is often called the winding number of γ\gamma around 0.

?
Problem 1.8.24

Let γ\gamma be as in Exercise 23, and assume in addition that the range of γ\gamma does not intersect the negative real axis. Prove that Ind (γ)=0(\gamma ) = 0. Hint: For 0≤c<∞0 \leq c < \infty, Ind (γ+c)(\gamma +c) is a continuous integer-valued function of cc. Also, Ind (γ+c)→0(\gamma +c) \rightarrow 0 as c→∞c \rightarrow \infty.

?
Problem 1.8.25

Suppose γ1\gamma_{1} and γ2\gamma_{2} are curves as in Exercise 23, and

∣γ1(t)−γ2(t)∣<∣γ1(t)∣(a≤t≤b) \left|\gamma _{1}(t)-\gamma _{2}(t)\right| < \left|\gamma _{1}(t)\right| \quad (a \leq t \leq b)

Prove that Ind (γ1)=\left(\gamma_{1}\right) = Ind (γ2)\left(\gamma_{2}\right). Hint: Put γ=γ2/γ1\gamma = \gamma_{2} / \gamma_{1}. Then ∣1−γ∣<1\left|1-\gamma \right| < 1, hence Ind (γ)=0(\gamma ) = 0, by Exercise 24. Also,

γ′γ=γ2′γ2−γ1′γ1 \frac{\gamma ^{\prime }}{\gamma } = \frac{\gamma _{2}^{\prime }}{\gamma _{2}}-\frac{\gamma _{1}^{\prime }}{\gamma _{1}}
?
Problem 1.8.26

Let γ\gamma be a closed curve in the complex plane (not necessarily diffierentiable) with parameter interval [0,2π][0,2 \pi ], such that γ(t)≠0\gamma (t) \neq 0 for every t∈[0,2π]t \in [0,2 \pi ].

Choose δ>0\delta > 0 so that ∣γ(t)∣>δ\left|\gamma (t)\right| > \delta for all t∈[0,2π]t \in [0,2 \pi ]. If P1P_{1} and P2P_{2} are trigonometric polynomials such that ∣Pȷ(t)−γ(t)∣<δ/4\left|P_{\jmath }(t)-\gamma (t)\right| < \delta / 4 for all t∈[0,2π]t \in [0,2 \pi ] (their existence is assured by Theorem 8.15), prove that

Ind⁡(P1)=Ind⁡(P2) \operatorname {Ind}\left(P_{1}\right) = \operatorname {Ind}\left(P_{2}\right)

by applying Exercise 25. Define this common value to be Ind (γ)(\gamma ). Prove that the statements of Exercises 24 and 25 hold without any differentiability assumption.

?
Problem 1.8.27

Let ff be a continuous complex function defined in the complex plane. Suppose there is a positive integer nn and a complex number c≠0c \neq 0 such that

lim⁡∣z∣→∞z−nf(z)=c \lim _{\left|z\right| \rightarrow \infty } z^{-n} f(z) = c

Prove that f(z)=0f(z) = 0 for at least one complex number zz. Note that this is a generalization of Theorem 8.8. Hint: Assume f(z)≠0f(z) \neq 0 for all zz, define

γr(t)=f(rett) \gamma _{r}(t) = f\left(r e^{t t}\right)

for 0≤r<∞,0≤t≤2π0 \leq r < \infty , 0 \leq t \leq 2 \pi, and prove the following statements about the curves γr\gamma_{r} :

?
(a)

Ind (γ0)=0\left(\gamma_{0}\right) = 0

(b)

Ind (γr)=n\left(\gamma_{r}\right) = n for all sufficiently large rr.

(c)

Ind (γr)\left(\gamma_{r}\right) is a continuous function of rr, on [0, ∞\infty).

Show that (a),(b)(a),(b), and (c)(c) are contradictory, since n>0n > 0.

Problem 1.8.28

Let DD be the closed unit disc in the complex plane. (Thus z∈Dz \in D if and only if ∣z∣≤1\left|z\right| \leq 1). Let gg be a continuous mapping of DD into the unit circle TT. (Thus, ∣g(z)∣=1\left|g(z)\right| = 1 for every z∈Dz \in D).

Prove that g(z)=−zg(z) = -z for at least one z∈Tz \in T. Hint: For 0≤r≤1,0≤t≤2π0 \leq r \leq 1,0 \leq t \leq 2 \pi, put

γr(t)=g(rett) \gamma _{r}(t) = g\left(r e^{t t}\right)

and put ψ(t)=e−ttγ1(t)\psi (t) = e^{-t t} \gamma_{1}(t). If g(z)≠−zg(z) \neq -z for every z∈Tz \in T, then ψ(t)≠−1\psi (t) \neq -1 for every t∈[0,2π]t \in [0,2 \pi ]. Hence Ind (ψ)=0(\psi ) = 0, by Exercises 24 and 26. It follows that Ind (γ1)=1\left(\gamma_{1}\right) = 1. But Ind (γ0)=0\left(\gamma_{0}\right) = 0. Derive a contradiction, as in Exercise 27.

?
Problem 1.8.29

Prove that every continuous mapping ff of DD into DD has a fixed point in DD. (This is the 2-dimensional case of Brouwer's fixed-point theorem). Hint: Assume f(z)≠zf(z) \neq z for every z∈Dz \in D. Associate to each z∈Dz \in D the point g(z)∈Tg(z) \in T which lies on the ray that starts at f(z)f(z) and passes through zz. Then gg maps DD into T,g(z)=zT, g(z) = z if z∈Tz \in T, and gg is continuous, because

g(z)=z−s(z)[f(z)−z] g(z) = z-s(z)[f(z)-z]

where s(z)s(z) is the unique nonnegative root of a certain quadratic equation whose coefficients are continuous functions of ff and zz. Apply Exercise 28.

?
Problem 1.8.30

Use Stirling's formula to prove that

lim⁡x→∞Γ(x+c)xcΓ(x)=1 \lim _{x \rightarrow \infty } \frac{\Gamma (x+c)}{x^{c} \Gamma (x)} = 1

for every real constant cc.

?
Problem 1.8.31

In the proof of Theorem 7.26 it was shown that

∫−11(1−x2)ndx≥43n \int _{-1}^{1}\left(1-x^{2}\right)^{n} d x \geq \frac{4}{3 \sqrt{n}}

for n=1,2,3,…n = 1,2,3, \ldots. Use Theorem 8.20 and Exercise 30 to show the more precise result

lim⁡n→∞n∫−11(1−x2)ndx=π \lim _{n \rightarrow \infty } \sqrt{n} \int _{-1}^{1}\left(1-x^{2}\right)^{n} d x = \sqrt{\pi }
?