Bạn đang xem bản rút gọn của tài liệu. Xem và tải ngay bản đầy đủ của tài liệu tại đây (1.8 MB, 341 trang )
Exercises
41
Proof
Set α = r cis θ = r eiθ , r > 0. Since e z = e x eiy , we will have e z = α if x = log r
and eiy = eiθ . Hence e z = α for all points z = x + i y with x = log r , y = Arg α =
θ ± 2kπ, k = 0, 1, 2 . . ..
v. (e z ) = e z .
Recall that (e z ) = (e z )x = e z .
To define sin z and cos z, note that for real y
eiy = cos y + i sin y
e−iy = cos y − i sin y
so that
sin y =
1 iy
(e − e−iy )
2i
and
1 iy
(e + e−iy ).
2
Thus we can define entire extensions of sin x and cos x by setting
cos y =
1 iz
(e − e−iz )
2i
1
cos z = (eiz + e−iz ).
2
sin z =
Many of the familiar properties of the sin and cos functions remain valid in the larger
setting of the complex plane. For example,
sin 2z = 2 sin z cos z
sin2 z + cos 2 z = 1
(sin z) = cos z.
These identities are easily verified and are left as an exercise. Moreover, in Section 6.3,
we will see that, in general, functional equations of the above form, known to be true
on the real axis, remain valid throughout the complex plane.
On the other hand, unlike sin x, sin z is not bounded in modulus by 1. For example,
| sin 10i | = 12 (e10 − e−10 ) > 10, 000.
Exercises
1. Show that
f x = lim
h→0
h real
provided the limits exist.
f (z + h) − f (z)
f (z + ih) − f (z)
; f y = lim
,
h
h
h→0
h real
42
3 Analytic Functions
2. a. Show that f (z) = x 2 + iy 2 is differentiable at all points on the line y = x.
b. Show that it is nowhere analytic.
3. Prove that the composition of differentiable functions is differentiable. That is, if f is differentiable
at z, and if g is differentiable at f (z), then g ◦ f is differentiable at z. [Hint: Begin by noting
g( f (z + h)) − g( f (z)) = [g ( f (z)) + ][ f (z + h) − f (z)]
where
→ 0 as h → 0.]
√
4. Suppose that g is a continuous “ z” (i.e., g 2 (z) = z) in some neighborhood of z. Verify that
√
g (z) = 1/2 z. [Hint: Use
g 2 (z) − g 2 (z 0 )
1=
z − z0
to evaluate
lim
z→z 0
g(z) − g(z 0 )
.]
z − z0
5. Suppose f is analytic in a region and f ≡ 0 there. Show that f is constant.
6. Assume that f is analytic in a region and that at every point, either f = 0 or f = 0. Show that f is
constant. [Hint: Consider f 2 .]
7. Show that a nonconstant analytic function cannot map a region into a straight line or into a circular
arc.
8. Find all analytic functions f = u + iv with u(x, y) = x 2 − y 2 .
9. Show that there are no analytic functions f = u + iv with u(x, y) = x 2 + y 2 .
10. Suppose f is an entire function of the form
f (x, y) = u(x) + iv(y).
Show that f is a linear polynomial.
11. a. Show that ez is entire by verifying the Cauchy-Riemann equations for its real and imaginary
parts.
b. Prove:
ez 1 +z 2 = ez 1 ez 2 .
12. Show: |ez | = e x .
13. Discuss the behavior of ez as z → ∞ along the various rays from the origin.
14. Find all solutions of
a. ez = 1,
c. ez = −3,
b. ez = i,
d. ez = 1 + i.
15. Verify the identities
a. sin 2z = 2 sin z cos z,
b. sin2 z + cos2 z = 1,
c. (sin z) = cos z.
16.* Show that
a. sin( π2 + iy) = 12 (e y + e−y ) = cosh y
b. | sin z| ≥ 1 at all points on the square with vertices ±(N + 12 )π ± (N + 12 )π i, for any positive
integer N.
c. | sin z| → ∞, as Imz = y → ±∞.
Exercises
43
17. Find (cos z) .
18. Find sin−1 (2)– that is, find the solutions of sin z = 2. [Hint: First set w = eiz and solve for ω.]
19.* Find all solutions of the equation:
z
ee = 1.
20. Show that sin(x + iy) = sin x cosh y + i cos x sinh y.
21. Show that the power series
f (z) = 1 + z +
z2
+ ... =
2!
∞
n=0
zn
n!
is equal to ez . [Hint: First show that f (z) f (w) = f (z + w), then show
f (x) = e x
f (iy) = cos y + i sin y
using the power series representations for the real functions
e x , cos x, sin x.]
22. Show:
g(z) = z −
z5
z3
+
− +...
3!
5!
is equal to sin z. [Hint: Use the power series representation for ez given in (21) to show that
g(z) =
23. Find a power series representation for cos z.
1 iz
(e − e−iz ).]
2i
Chapter 4
Line Integrals and Entire Functions
Introduction
Recall that, according to Theorem 2.9, an everywhere convergent power series
represents an entire function. Our main goal in the next two chapters is the somewhat
surprising converse of that result: namely, that every entire function can be expanded
as an everywhere convergent power series. As an immediate corollary, we will be
able to prove that every entire function is infinitely differentiable. To arrive at these
results, however, we must begin by discussing integrals rather than derivatives.
4.1 Properties of the Line Integral
4.1 Definition
Let f (t) = u(t) + i v(t) be any continuous complex-valued function of the real
variable t, a ≤ t ≤ b.
b
a
b
f (t)dt =
a
b
u(t)dt + i
v(t)dt.
a
4.2 Definition
a. Let z(t) = x(t) + i y(t), a ≤ t ≤ b. The curve determined by z(t) is called
piecewise differentiable and we set
z˙ (t) = x (t) + i y (t)
if x and y are continuous on [a, b] and continuously differentiable on each subinterval [a, x 1 ], [x 1 , x 2 ], . . . , [x n−1 , b] of some partition of [a, b].
b. The curve is said to be smooth if, in addition, z˙ (t) = 0 (i.e., x (t) and y (t) do not
both vanish) except at a finite number of points.
45
46
4 Line Integrals and Entire Functions
Throughout the remainder of the text, all curves will be assumed to be smooth
unless otherwise stated.
Finally, we define the important concept of a line integral.
4.3 Definition
Let C be a smooth curve given by z(t), a ≤ t ≤ b, and suppose f is continuous at
all the points z(t). Then, the integral of f along C is
b
f (z)dz =
f (z(t))˙z (t)dt.
a
C
Note that the integral along the curve C depends not only on the points of C but on
the direction as well. However, we will show that it is independent of the particular
parametrization. Intuitively, if z(t), a ≤ t ≤ b, and ω(t), c ≤ t ≤ d, trace the same
curve in the same direction, then λ = z −1 ◦ ω will be a 1-1 mapping of [c, d] onto
[a, b] such that
ω(t) = z(λ(t)).
(1)
However, if z is not 1-1, it is difficult to define z −1 . Instead, we take the existence of
some λ that satisfies (1) as the definition for equivalent curves.
C
z
a
ω
b
c
d
4.4 Definition
The two curves
C1 : z(t), a ≤ t ≤ b
and
C2 : ω(t), c ≤ t ≤ d
are smoothly equivalent if there exists a 1-1 C 1 mapping λ(t) : [c, d] → [a, b] such
that λ(c) = a, λ(d) = b, λ (t) > 0 for all t, and
ω(t) = z(λ(t)).
(It is easy to verify that the above is an equivalence relation. See Exercise 1.)
4.5 Proposition
If C1 and C2 are smoothly equivalent, then
f =
C1
f.
C2
4.1 Properties of the Line Integral
47
Proof
Suppose f (z) = u(z) + i v(z), C1 and C2 as above. Then, by definition
b
f =
a
C1
b
u(z(t))x (t)dt −
v(z(t))y (t)dt
a
b
+i
b
u(z(t))y (t)dt + i
a
v(z(t))x (t)dt
(1)
a
while
d
f =
C2
[u(z(λ(t))) + i v(z(λ(t)))][x (λ(t)) + i y (λ(t))]λ (t)dt.
(2)
c
Expanding the integrand in (2) and analyzing the four terms separately, we find that
they are exactly equal to the four corresponding terms in (1).
For example
d
b
u(z(λ(t)))x (λ(t))λ (t)dt =
c
u(z(t))x (t)dt,
a
by the change-of-variable theorem for ordinary real integrals, and the proof is
complete.
The following proposition points out the dependence of the line integral on the
direction of the curve.
4.6 Definition
Suppose C is given by z(t), a ≤ t ≤ b. Then −C is defined by z(b + a − t),
a ≤ t ≤ b. (Intuitively, −C is the point set of C traced in the opposite direction.)
4.7 Proposition
−C
f =
f.
C
Proof
−C
b
f =−
f (z(b + a − t))˙z (b + a − t)dt.
a
Again, expanding the integral into real and imaginary parts and applying the changeof-variable theorem to each (real) integral, we find
−C
a
f =
b
f (z(t))˙z (t)dt = −
f.
C
48
4 Line Integrals and Entire Functions
E XAMPLE 1
Suppose f (z) = x 2 + i y 2 (where x and y denote the real and imaginary parts of z,
respectively), and consider
C : z(t) = t + i t, 0 ≤ t ≤ 1.
Then
z˙ (t) = 1 + i
and
1
f (z)dz =
0
C
1
(t 2 + i t 2 )(1 + i )dt = (1 + i )2
t 2 dt = 2i /3.
0
♦
E XAMPLE 2
Let
f (z) =
x
1
y
= 2
−i 2
,
z
x + y2
x + y2
and set
C : z(t) = R cos t + i R sin t, 0 ≤ t ≤ 2π,
R = 0.
Then
2π
f (z)dz =
0
C
2π
=
sin t
cos t
−i
R
R
(−R sin t + i R cos t)dt
i dt = 2πi (See Exercise 8.)
0
That is, the integral of 1/z around any circle centered at the origin (traversed counterclockwise) is 2πi .
♦
E XAMPLE 3
Suppose f (z) ≡ 1, and let C be any smooth curve. Then
b
f (z)dz =
C
z˙ (t)dt = z(b) − z(a).
a
♦
The integrals defined above are natural generalizations of the definite integral and,
not too surprisingly, they share many of the same properties.
4.8 Proposition
Let C be a smooth curve; let f and g be continuous functions on C; and let α be any
complex number. Then
4.1 Properties of the Line Integral
I.
C
[ f (z) + g(z)]dz =
II.
C
α f (z)dz = α
C
49
f (z)dz +
C
C
g(z)dz
f (z)dz.
Proof
Exercise 4.
Notation: If α and β are complex numbers, the symbol α
the inequality |α| ≤ |β|.
β will be used to denote
4.9 Lemma
Suppose G(t) is a continuous complex-valued function of t. Then
b
b
G(t)dt
a
|G(t)|dt.
a
Proof
Suppose
b
G(t)dt = Reiθ , R ≥ 0.
(1)
a
By Proposition 4.8, then
b
e−iθ G(t)dt = R.
(2)
a
Suppose further that e−iθ G(t) = A(t) + i B(t), with A and B real-valued. Then,
according to (2),
b
R=
b
A(t)dt =
a
Re (e−iθ G(t))dt.
a
But Re z ≤ |Re z| ≤ |z|, hence
b
R≤
|G(t)|dt.
(3)
a
A comparison of (1) and (3) then gives the desired result.
4.10 M-L Formula
Suppose that C is a (smooth) curve of length L, that f is continuous on C, and that
f
M throughout C. Then
f (z)dz
C
M L.
50
4 Line Integrals and Entire Functions
Proof
Suppose C is given by z(t) = x(t)+i y(t), a ≤ t ≤ b. Then, by the previous lemma,
b
f (z)dz =
b
f (z(t))˙z dt
a
C
| f (z(t))˙z (t)|dt.
a
According to the Mean-Value Theorem for Integrals applied to the positive functions
| f (z(t))| and |˙z (t)|
f (z)dz)
b
max | f (z)|
z∈C
C
|˙z (t)|dt.
(4)
a
Finally, recall that for any curve given parametrically by (x(t), y(t)), a ≤ t ≤ b,
the arc length L is given by
b
L=
a
so that according to (4)
C
b
(x (t))2 + (y (t))2 dt =
|˙z (t)|dt,
a
f (z)dz
M L.
E XAMPLE
Let C be the unit circle and suppose f
1 on C. Then M = 1, L = 2π, and
f (z)dz
2π.
C
To see that the upper bound of 2π can actually be achieved, consider Example 2
above.
♦
4.11 Proposition
Suppose { f n } is a sequence of continuous functions and f n → f uniformly on the
smooth curve C. Then
f (z)dz = lim
n→∞ C
C
f n (z)dz.
Proof
f (z)dz −
C
f n (z)dz =
C
[ f (z) − f n (z)]dz
C
by Proposition 4.8. Taking n large enough so that | f (z) − f n (z)| <
and applying Proposition 4.10, shows that
f (z) −
C
fn (z)dz
C
· (length of C)
for all z ∈ C,
4.1 Properties of the Line Integral
51
for any pre-assigned > 0, and hence that
lim
n→∞ C
f n (z)dz =
f (z)dz.
C
The following generalization of the Fundamental Theorem of Calculus will be
crucial in the development of this chapter.
4.12 Proposition
Suppose f is the derivative of an analytic function F–that is, f (z) = F (z), where
F is analytic on the smooth curve C. Then
f (z)dz = F(z(b)) − F(z(a)).
C
Proof
The proof depends on a complex analogue of the chain-rule for differentiation. Letting
γ (t) = F(z(t)), a ≤ t ≤ b,
we wish to show that
γ˙ (t) = f (z(t))˙z (t)
at the all-but-finite number of points where z˙ (t) exists and is nonzero.
Note first that for any smooth curve λ(t), by considering the real and imaginary
parts of λ separately, it is easily seen that
λ(t + h) − λ(t)
˙
λ(t)
= lim
.
h
h→0
h real
Hence
F(z(t + h)) − F(z(t))
h→0
h
F(z(t + h)) − F(z(t)) z(t + h) − z(t)
·
.
= lim
h→0
z(t + h) − z(t)
h
γ˙ (t) = lim
[Since z˙ (t) = 0, we can find δ > 0 so that |h| < δ implies z(t + h) − z(t) = 0.]
Thus
γ˙ (t) = f (z(t))˙z (t).
Proposition 4.12 follows then by noting that
b
f (z)dz =
C
a
b
f (z(t))˙z (t)dt =
γ˙ (t)dt
a
= γ (b) − γ (a) = F(z(b)) − F(z(a)).