Chapter 1 (John H. Mathews) PDF
Chapter 1 (John H. Mathews) PDF
Chapter 1 (John H. Mathews) PDF
numliers
Overview
Get ready for a treat. You're about to begin studying some of the most beautiful
ideas in mathematics. They are ideas with surprises. They evolved over several
centuries, yet they greatly simplify extremely difficult computations, making
some as easy as sliding a hot knife through butter. They also have applications
in a variety of areas, ranging from fluid flow, to electric circuits, to the mysterious
quantum world. Generally, they belong to the area of mathematics known as
complex analysis, which is the subject of this book. This chapter focuses on the
development of entities we now call complex numbers.
Complex analysis can roughly be thought of as the subject that applies the theory
of calculus to imaginary numbers. But what exactly are imaginary numbers?
Usually, students learn about them in high school with introductory remarks
from their teachers along the following lines: "We can't take the square root of
a negative number. But let's pretend we can and begin by using the symbol
i =
Rules are then learned for doing arithmetic with these numbers. At
some level the rules make sense. If i = A, it stands to reason that i2 = -l.
However, it is not uncommon for students to wonder whether they are really
doing magic rather than mathematics.
v=I."
If you ever felt that way, congratulate yourself! You're in the company of
some of the great mathematicians from the sixteenth through the nineteenth
centuries. They, too, were perplexed by the notion of roots of negative numbers.
Our purpose in this section is to highlight some of the episodes in the very
colorful history of how thinking about imaginary numbers developed. We intend
to show you that, contrary to popular belief, there is really nothing imaginary
about "imaginary numbers." They are just as real as "real numbers."
Our story begins in 1545. In that year, the Italian mathematician Girolamo
Cardano published Ars Magna (The Great Art), a 40-chapter masterpiece in
which he gave for the first time a method for solving the general cubic equation
Cardano did not have at his disposal the power of today's algebraic notation,
and he tended to think of cubes or squares as geometric objects rather than algebraic quantities. Essentially, however, his solution began with the substitution
z = xThis move transformed Equation (1-1) into a cubic equation without
a squared term, which is called a depressed cubic. To illustrate, begin with
z3 + 9z2 + 24z + 20 = 0 and substitute z = x - a32 = X = x - 3. The equation
then becomes (x - 3)3 + 9 (x - 3)2 + 24 (x - 3) + 20 = 0, which simplifies to
x3 ~ 3x + 2 = O.
You need not worry about the computational details here, but in general the
substitution z = x - a32 transforms Equation (1-1) into
a;.
1 2
1
2 3
h
b- - al - 3a2,
were
an d c -_ -3a1a2
+ 27a2
+ ao.
If Cardano could get any value of x that solved a depressed cubic, he could
easily get a corresponding solution to Equation (1-1) from the identity z = x - a32 .
Happily, Cardano knew how to solve a depressed cubic. The technique had been
communicated to him by Niccolo Fontana who, unfortunately, came to be known
as Tartaglia (the stammerer) due to a speaking disorder. The procedure was also
independently discovered some 30 years earlier by Scipione del Ferro of Bologna.
Ferro and Tartaglia showed that one of the solutions to Equation (1-2) is
3 _ ~
c
4
b +
272
_ ~
c
4
b .
27
Although Cardano would not have reasoned in the following way, today we
can take this value for x and use it to factor the depressed cubic into a linear
and quadratic term. The remaining roots can then be found with the quadratic
formula. For example, to solve z3 + 9z2 + 24z + 20 = 0, use the substitution
z = x-3 to get x3 -3x+2 = 0, which is a depressed cubic in the form of Equation
(1-2). Next, apply the "Ferro-Tartaglia" formula with b = -3 and c = 2 to get
x
- -22 +-
22
4
(_3)3
(_3)3
+ -+ - -22 - 22-4 + -= R'
-1 + R
-1 = -2 . Smce
27
27
x + 2 must be a factor of x3 - 3x + 2. Dividing x + 2 into
X = -2
is a root,
x3 - 3x + 2 gives x2 - 2x + 1, which yields the remaining (duplicate) roots of
x = 1. The solutions to z3+9z2+24z+20
= 0 are obtained by recalling z = x-3,
which yields the three roots Zl = -2 - 3 = -5, and Z2 = Z3 = 1 - 3 = -2.
So, by using Tartaglia's work and a clever transformation technique, Cardano
was able to crack what had seemed to be the impossible task of solving the
u - vA
= \/2 -llA,
+ vA)3
= 2 + llj=I,
(u + v";=lf
= u3 + 3(u2)vv=1 + 3(u)(vH)2
+ (vv=1)3
3
2
= u + 3(u) (vv=1)2 + 3(u )vv=1 + (vH)3
= (u3 - 3uv2) + (3u2v - v3)v=1
=
=2+11H.
(J - j)
(1-;) )
By equating like parts of Equations (1-4) and 1-.)) Bombelli reasoned that
= 2 and v(3u2 - v2) = ll. Perhaps thinking even more wildly,
Bombelli then supposed that u and v were integers. ThE' only integer factors of
2 are 2 and 1, so the equation u( u2 - 3v2) = 2 led Bombelli to conclude that
u = 2 and u2 - 3v2 = 1. From this conclusion it follows that v2 = 1. or v = 1.
Amazingly, u = 2 and v = 1 solve the second equation v(3u2 - v2) = ll, so
Bombelli declared the values for u and v to be u = 2 and v = 1, respectively.
u(u2 - 3v2)
Since (2
+ l1R +
,/2
-l1R
= (2
+ R) + (2 -
R)
= 4,
Wallis imagined these solutions as displacements to the left and right from
the point -b. He saw each displacement, whose value is Jb2 - c2, as the length
of the sides of the right triangles shown in Figure 1.1.
The points PI and P2 represent the solutions to our equation, which is clearly
correct if b2 - c2 2': O. But how should we picture PI and P2 when negative roots
arise (i.e., when f)L. - cL. < UF -WaIhs reasoneatnat,
witn nega"ilV1::Ttn.h1;,
'u'wuuili
be less than c, so the lines of length b in Figure 1.1 would no longer be able to
reach all the way to the x-axis. Instead, they would stop somewhere above it, as
Figure 1.2 shows. Wallis argued that PI and P2 should represent the geometric
locations of the solutions x = -b - -/b2 - c2 and x = -b + Jb2 - c2 when
Two additional mathematicians deserve mention. The Frenchman AugustinLouis Cauchy (1789-1857) formulated many of the classic theorems that are now
part of the corpus of complex analysis. The German Carl Friedrich Gauss (17771855) reinforced the utility of complex numbers by using them in his several
proofs of the fundamental theorem of algebra (see Chapter 6). In an 1831 paper,
he produced a clear geometric representation of x + iy by identifying it with the
point (x, y) in the coordinate plane. He also described how to perform arithmetic
operations with these new numbers.
It would be a mistake, however, to conclude that in 1831 complex numbers were transformed into legitimacy. In that same year the prolific logician
Augustus De Morgan commented in his book, On the Study and Difficulties of
Mathematics, "We have shown the symbol
to be void of meaning, or rather
self-contradictory and absurd. Nevertheless, by means of such symbols, a part
of algebra is established which is of great utility."
v=a
There are, indeed, genuine logical problems associated with complex numbers. For example, with real numbers VOJj = y'liVb so long as both sides of
the equation are defined. Applying this identity to complex numbers leads to
1 = VI = vi ( -1) (-1) = A A = -1. Plausible answers to these problems
can be given, however, and you will learn how to resolve this apparent contradiction in Section 2.2. De Morgan's remark illustrates that many factors are needed
to persuade mathematicians to adopt new theories. In this case, as always, a
firm logical foundation was crucial, but so, too, was a willingness to modify some
ideas concerning certain well-established properties of numbers.
As time passed, mathematicians gradually refined their thinking, and by the
end of the nineteenth century complex numbers were firmly entrenched. Thus,
as it is with many new mathematical or scientific innovations, the .theory of
complex numbers evolved by way of a very intricate process. But what is the
theory that Tartaglia, Ferro, Cardano, Bombelli, Wallis, Euler, Cauchy, Gauss,
and so many others helped produce? That is, how do we now think of complex
numbers? We explore this question in the remainder of this chapter.
2. Explain why cubic equations, rather than quadratic equations, played a pivotal
role in helping to obtain the acceptance of complex numbers.
v=T.
-v=T
being represented
5. Use Bombelli's technique to get all solutions to the following depressed cubics.
(a) x3 - 30x - 36
O.
(c) x3-60x-32=0.
+ 18 =
(a) Z3 - 6z2
3z
+ 3z2
24z
(b) z3
+ 28
O.
= O.
consistent with the representation used today? To help you answer this question,
refer to the article by Alec Norton and Benjamin Lotto, "Complex Roots Made
Visible," The College Mathematics Journal, 15(3), June 1984, pp. 248-249.
8. Investigate library or web resources and write up a detailed description explaining
why the solution to the depressed cubic, Equation (1-3), is valid.
+ Z2 =
+ (X2, Y2)
= (Xl + iyd + (X2 + iY2)
= (Xl + X2) + i (YI + Y2)
(Xl,
= (Xl
Zl
+ Z2 =
Zl -
Z2
yd
+ X2,
YI
+ Y2)
+ (X2, Y2)
= (Xl + X2, YI + Y2).
(1-8)
(1-9)
(Xl,
Yl)
Zl
+ Z2
Zl -
Zl
Z2
Z2
+ (5, -6)
(8, 1)
+ Z2 =
Zl -
(3, 7)
+ 7i) + (5 = (3 + 7i) - (5 (3
6i)
13).
= 8+i
6i) = -2
and
and
+ 13i.
= (Xl,
Yl)(X2,
Y2)
+ iyd(X2 + iY2)
= X1X2 + iX1Y2 + iX2Yl + i2Y1Y2
= X1X2 - Y1Y2 + i(X1Y2 + X2Yl)
= (X1X2 - Y1Y2, X1Y2 + x2yd
ZlZ2
(Xl
(Xl,
= (X1X2
ZlZ2
yd(X2'
-
Y1Y2,
Y2)
X1Y2
+ X2Yl).
(1-10) I
ZlZ2
+ 7i)(5
(3
~ 6i)
15 - 42( -1)
+ (-18 + 35)i
+ 17i
= 57
= (57, 17).
Of course, it makes sense that the answer came out as we expected because
we used the notation x + iy as motivation for our definition in the first place.
To motivate our definition for division, we proceed along the same lines as
we did for multiplication, assuming that Z2 -I- 0:
Zl
Z2
(Xl,
(X2,
yd
Y2)
(Xl
+ iyd
(X2
+ iY2)
We need to figure out a way to write the preceding quantity in the form
To do so, we use a standard trick and multiply the numerator and
denominator by X2 - iY2, which gives
X
+ iy.
Zl
(Xl
+ iyd(X2
- iY2)
Z2
(X2
+ iY2)(X2
- iY2)
(xI,yd
(X2,Y2)
XIX2
X~
+ VIV2
+
y~
-XIV2
,
X~
+ X2VI)
+ V~
EXAMPLE 1.3
21
=~=
(5, -6)
22
If
21
= (3, 7) and
(15-4218+35)
25 + 36' 25 + 36
22
= (5,
-6), then
= (-27
for multiplication,
53)
61 '61
.
(3,7)
(5, -6)
21
22
3 + 7i
5 - 6i
3 + 7i 5 + 6i
---5 - 6i 5 + 6i
15 + 18i + 35i + 42i2
25 + 30i - 30i - 36i2
15 - 42 + (18 + 35) i
25 + 36
53.
51 + 61 t
-27
(-27 53).
61 '61
(PI) Commutative
(P2)
Associative
21
21
+ 22
(22
= 22
23)
21.
(21
22)
23
(P3) Additive identity: There is a complex number w such that z+w = z for
all complex numbers z. The number w is obviously the ordered pair (0, 0).
(P4) Additive inverses: For any complex number z, there is a unique complex
number 1] (depending on z) with the property that z+1] = (0, 0). Obviously,
if z = (x, y) = x + iy, the number 1] will be (-x, -y) = -x - iy = -z.
(P5)
Commutative
(P6)
Associative
ZlZ2
= Z2Z1'
= (ZlZ2)Z3.
Zl(Z2Z3)
(P7) Multiplicative
identity: There is a complex number (such that z( = z
for all complex numbers z. As you might expect, (1, 0) is the unique complex
number ( having this property. We ask you to verify this identity in the
exercises for this section.
(P8) Multiplicative
inverses:
For any complex number z = (x, y) other
than the number (0, 0), there is a complex number (depending on z), which
we denote Z-l, having the property that ZZ-l = (1, 0) = 1. Based on our
definition for division, it seems reasonable that the number z-l would be
z-l = (1,0) = 1 = _1_
= ~
=
+ i---=.1L- =
---=.1L-) We
z
x2+y2
x+iy
_x_
xY+y2
(_x_
x2+y2
x2+y2,
x2+y2
ask you to confirm this result in the exercises for this section.
None of these properties is difficult to prove. Most of the proofs make use
of corresponding facts in the real number system. To illustrate, we give a proof
of property (PI).
Proof
Z2 = (X2,
of the commutative
law for addition:
be arbitrary complex numbers. Then,
Let
Zl
Y2)
(Xl,
(Xl
yd + (X2,
+ X2,
(X2+X1,
(X2,
Z2
Y2)
Y1
Y2)
+ Y2)
Y2+Y1)
+ (Xl, yd
(by definition
of addition
of complex
numbers)
of addition
of complex
numbers)
+ zl
Actually, you can think of the real number system as a subset of the complex number system. To see why, let's agree that, as any complex number of the
form (t, 0) is on the x-axis, we can identify it with the real number t. With this
correspondence, we can easily verify that our definitions for addition, subtraction, multiplication, and division of complex numbers are consistent with the
corresponding operations on real numbers. For example, if Xl and X2 are real
numbers, then
(Xl, 0)(X2,
0)
(XIX2 - 0, 0 + 0)
(XIX2.
0)
correspondence)
of multiplication
the consistence
of complex
numbers)
of our correspondence).
It is now time to show specifically how the symbol i relates to the quantity
Note that
A.
(0, 1)(0, 1)
(0 - 1, 0 + 0)
(by definition
of multiplication
of complex
numbers)
(-1. 0)
-1
correspondence).
which means i = (0, 1) = R. SO, the next time you are having a discussion
with your friends and they scoff when you claim that A is not imaginary,
calmly put your pencil on the point (0. 1) of the coordinate plane and ask them
if there is anything imaginary about it. When they agree there isn't, you can tell
them that this point, in fact, represents the mysterious R in the same way
that (1, 0) represents 1.
We can also see more clearly now how the notation X + iy equates to (x. y).
Using the preceding conventions (i.e., X = (x, 0), etc.), we have
(x, 0)
(x, 0)
(x, y)
+ (0. l)(y, 0)
+ (0, y)
discussed
(by definition
of multiplication
(by definition
of addition
conventions)
of complex
of complex
numbers)
numbers).
Thus, we may move freely between the notations x + iy and (x, y), depending on which is more convenient for the context in which we are working.
Students sometimes wonder whether it matters where the "i" is located in writing a complex number. It does not. Generally, most texts place terms containing
an "i" at the end of an expression, and place the "i" before a variable but after
a constant. Thus. we write x + iy. u + iv, etc .. but 3 + 7i. 5 - 6i. and so forth.
Because letters lower in the alphabet generally denote constants, you will usually
(but not always) see the expression a + bi instead of a + ib. Many authors write
quantities like 1 + iv'3 instead of 1 + v'3i to make sure the "i" is not mistakenly
thought to be inside the square root symbol. Additionally, if there is concern
that the "i" might be missed, it is sometimes placed before a lengthy expression,
as in 2 cos (-~7r + 2m]") + i2 sin (-~7r + 2m]").
We close this section with three important definitions and a theorem involving them. We ask you for a proof of the theorem in the exercises.
x - iy .
EXAMPLE 1.4
a) Re (-3 + 7i) = -3 and Re[(9, 4)] = 9. b) Im( -3
7i) = 7 and Im[(9, 4)] = 4. c) -3 + 7i = -3 - 7i and (9, 4) = (9, -4) .
Re(z)=z+z.
2
Im(z)
z-z
=~.
Zl,
and
Z2
are arbitrary
complex numbers.
Because of what it erroneously connotes, it is a shame that the term imaginary is used in Definition (1.6). It was coined by the brilliant mathematician and
philosopher Rene Descartes (1596-1650) during an era when quantities such as
R were thought to be just that. Gauss, who was successful in getting mathematicians to adopt the phrase complex number rather than imaginary number,
also suggested that they use lateral part of z in place of imaginary part of z. Unfortunately, that suggestion never caught on, and it appears we are stuck with
what history has handed down to us.
(a) i275.
(b) ~.
(c) Re(i).
(d) Im(2).
(e) (i - 1)3.
+ 5).
(7 + 6i) + 1m (5 -
(f) (7 - 2i)(3i
(g) Re
(h) 1m ( ;~~~) .
(i)
(4-i)(1-3i)
-1+2i
(a) (1 + i)(2
+ i)(3 + i).
(b) (3+i)/(2+i).
(c) Re [(i - 1)3] .
(d) Im[(l
+ i)-2].
(e) ;~~: _
~-=-~i.
(f) (1 + i)-2
(g) Re [(x -
(h) 1m
iy)2] .
(X~iY) .
4i) .
(i) Re [(x
(j) 1m [(x
3. Show that
+ iy) (x
+ iy)3]
zz
4. Verify Identities
5. Let P
- iy)].
(1-12)-(1-19).
(z) = anzn
a polynomial
of degree n.
(a) Suppose that an, an-I, ... , aI, ao are all real. Show that if Zl is a root
of P, then Zl is also a root. In other words, the roots must be complex
conjugates, something you likely learned without proof in high school.
(b) Suppose not all of an, an-I, ... , aI, ao are real. Show that P has at
least one root whose complex conjugate is not a root. Hint: Prove the
contrapositive.
(c) Find an example of a polynomial that
complex conjugates, and some not.
6. Let Zl = (Xl, Yl) and Z2
disprove the following.
(a) Re(Zl
+ Z2)
= Re(zl)
(X2,
Y2) be arbitrary
complex numbers.
as
Prove or
+ Re(z2).
+ Z2)
= 1m (Zl)
+ 1m (Z2)'
= '\'
6
(n)
h
k Zk Wn-k , were
(n)
n!
k = k!(n-k)!'
k=O
9. Let's use the symbol * for a new type of multiplication of complex numbers defined
by Zl *Z2 = (XIX2, YlY2). This exercise shows why this is an unfortunate definition.
(a) Use the definition given in property (P7) and state what the multiplicative identity ( would have to be for this new multiplication.
(b) Show that if you use this new multiplication, nonzero complex numbers
of the form (0, a) have no inverse. That is, show that if Z = (0, a),
there is no complex number w with the property that z * w = (, where
( is the multiplicative identity you found in part (a).
10. Explain why the complex number (0, 0) (which, you recall, we identify with the
real number 0) has no multiplicative inverse.
11. Prove property
12. Verify that if z = (x, y), with x and y not both 0, then Z-l = (l,z 0) (i.e., Z-l = ~).
Hint: Let z = (x, y) and use the (ordered pair) definition for division to compute
Z-l
= ((1,0.
Then, with the result you obtained, use the (ordered pair) definition
x,y
for multiplication to confirm that zz-l = (1, 0) = 1.
13. From Exercise 12 and basic cancellation laws, it follows that z-l = ~ ~ ~. The
numerator here, Z, is trivial to calculate and, as the denominator zz is a real
number (Exercise 3), computing the quotient ~ should be rather straightforward.
Use this fact to compute Z-l if z = 2 + 3i and again if z = 7 - 5i.
14. Show, by equating the real numbers Xl and X2 with (Xl, 0) and (X2, 0), respectively, that the complex definition for division is consistent with the real definition
for division. Hint: Mimic the argument given in the text for multiplication.
Complex numbers are ordered pairs of real numbers, so they can be represented
by points in the plane. In this section we show the effect that algebraic operations
on complex numbers have on their geometric representations.
\lVecan represent the number z = x + iy = (x, y) by a position vector in the
xy plane whose tail is at the origin and whose head is at the point (x, y). When
the xy plane is used for displaying complex numbers, it is called the complex
plane, or more simply, the z plane. Recall that Re(z) = x and Im(z) = y.
Geometrically, Re(z) is the projection of z = (x, y) onto the x-axis, and Im(z)
is the projection of z onto the y-axis. It makes sense, then, to call the x-axis the
real axis and the y-axis the imaginary
axis, as Figure 1.3 illustrates.
Addition of complex numbers is analogous to addition of vectors in the
plane. As we saw in Section 1.2, the sum of Zl = Xl + iYl = (Xl, Yl) and
Z2 = X2 + iY2 = (X2,
Y2) is (Xl + X2, Yl + Y2). Hence Zl + Z2 can be obtained
vectorially by using the "parallelogram law," where the vector sum is the vector
represented by the diagonal of the parallelogram formed by the two original
vectors. Figure 1.4 illustrates this notion.
The difference Zl - Z2 can be represented by the displacement vector from
the point Z2 = (X2, Y2) to the point Zl = (Xl, Yl), as Figure 1.5 shows.
Imaginary axis
Y
n;,?,f'
I
X
Ci
Z
2. .
Copy of vector 21
(positioned
y
at the tail of vector 22)
.,..~.ZI+Z2
"
'z~
Copy of vector 22
(positioned at the tail of vector 2t)
Copy of vector
~
..l ..~
ZI-
4 + 3i
Z2
z:J
,. ZI
----~
r~
-z 2
Z1-
7.
1(positioned
/L
Figure 1.6
The real and imaginary
parts of a complex number.
+ iy
is a
The number Izi is the distance between the origin and the point z = (x, y).
The only complex number with modulus zero is the number o. The number
z = 4 + 3i has modulus 14+ 3il = J 42 + 33 = v'25 = 5 and is depicted in Figure
1.6. The numbers IRe(z)l, IIm(z)l, and Izi are the lengths of the sides of the right
triangle OPQ shown in Figure 1.7. The inequality IZ11< IZ21means that the
point Z1 is closer to the origin than the point Z2. Although obvious from Figure
1.7, it is still profitable to work out algebraically the standard results that
= (x, y)
=x+
~--+-----+-----<~-(-x, 0):
iy
x
:(x, 0)
- - - - - - - -' z= (x, -y)
(O,-y)
=x-iy
Figure 1.7
components,
Figure 1.8
The geometry
and conjugation.
of negation
Zl
and
Z2
then
IZ1 + z212
(by Identity
(1-22))
(by Identity
(1-13))
(by Identity
(1-22)
commutative
<
(by Identities
and the
law)
(1-12)
(by Identity
(1-10))
(by Identity
(1-21))
and (I-I.!))
EXAMPLE 1.5 To produce an example of which Figure 1.9 is a reasonable illustration, we let Zl = 7 + i and Z2 = 3 + 5i. Then IZll = )49 + 1 =
v'5O and IZ21 = )9 + 25 = V34. Clearly. Zl + Z2 = 10 + 6i: hence IZl + z21 =
)100 + 36 = J136. In this case. we can verify the triangle inequality without
recourse to computation of square roots because IZl + z21
)136 = 2V34 =
V55 +
v'34
IZll + hi
identities
Taking square roots of the terms on the left and right establishes
important identity:
EXAMPLE
1.6
If
IZ21 = )9 + 4 = /13.
V5/13 = IZ111z21 .
Zl
another
Also
Figure 1.10 illustrates the multiplication shown in Example 1.6. The length
of the ZlZ2 vector apparently equals the product of the lengths of Zl and Z2.
confirming Equation (1-2.]). but why is it located in the second quadrant when
both Zl and Z2 are in the first quadrant? The answer to this question will become
apparent to you in Section 1.4.
(a) 1(1+i)(2+i)l.
(b)
I~-=-~il.
(c) 1(1+i)50I.
(d) Izzl, where z = x + iy.
(e) Iz-112,wherez=x+iy.
(a)
Zl
= 2 + 3i and
(b)
Zl
-1 + 2i and
Z2
-2 + 3i.
(c)
Zl
1 + iV3 and
Z2
-1
(a) ~+i.
(b) V2+i(V2+1).
(c) 2+3i.
(d) -./
+ iV3.
(c) (1-26).
Z2
= 4 + i.
+ iV3.
(a) Iz + 1 - 2il = 2.
(b) Re(z+I)=O.
(c) Iz+2il::':;1.
2::
IRe(z)1 + IIm(z)l
z! ~Z2
is the midpoint
Z2.
and -iz
13. Show that the equation of the line through the points Zl and
in the form z = Zl + t(Z2 - Zl), where t is a real number.
14. Show that the nonzero vectors Zl and
Z2
Z2
can be expressed
O.
I k~l
inequality:
ZkU!"k
I ::.:;
::.:;IZl - z21.
to prove that
Zk
I ::.:;
k=l
IZk
numbers n.
k=l
24. Let Zl and Z2 be two distinct points in the complex plane, and let K be a positive
real constant that is less than the distance between Zl and Z2.
(a) Show that the set of points {z:
with foci Zl and Z2
Iz - zll-
of the hyperbola
Iz - z21 = K} is a hyperbola
25. Let
Zl and Z2 be two distinct points in the complex plane, and let K be a positive
real constant that is greater than the distance between Zl and Z2.
(a) Show that the set of points {z : [z with foci Zl and Z2.
zll + Iz - z21 = K}
is an ellipse
(b) Find the equation of the ellipse with foci 3i that goes through .the
point 8 - 3i.
(c) Find the equation of the ellipse with foci 2i that goes through the
point 3 + 2i.
OF COMPLEX
NUMBERS,
Definition
1. 9: Polar representation
z = (x,y) =x+ iy
-'1 =(rcos
e, rsin e) = r(cos
1
1
I
(x, 0)
e + i sin e)
(0, y)
and
z = (x, y) = x + iy
: =(r cos e, r sin e) = r(cos e+ i sin e)
....
r
e
(x, 0)
As Figure 1.11 (b) shows, B can be any value for which the identities cos B = ~
and sin B = ~ hold. For z =1= 0, the collection of all values of B for which
z = r(cosB+isinB)
is denoted arg z. Formally, we have the following definitions.
If z
=1=
0,
arg (1 + i)
7r
}
+ 2n7r : n is an inteO"er
{~
4
b'
177r
... -- 77r -7r -97r -...
4'4'4'4'
Let z
=1=
be a complex number.
Then
arg z C arctan -,
where arctan;
= {B: tan B = ;} . Note that, as with arg, arctan is a set (as
opposed to Arc tan. which is a number). We specifically identify arg z as a
proper subset of arctan ~ because tan B has period n, whereas cos B and sin B
have period 2n. In selecting the proper values for arg z, we must be careful in
specifying the choices of arctan ~ so that the point z associated with rand B lies
in the appropriate quadrant.
_
EXAMPLE 1.10
-g".
-J3 - i =
-5n
2cos --
+ i2sin--
-5n
6
= 2 cas (-5n
-6- + 2mr ) + i2 sin (-5n
-6- + 2nn ) .
Arg
(-J3 -
arg
(-v3 - i)
i)
6
-5n
= { -6-
+ 2nn : n is an integer
- i) is indeed a proper
EXAMPLE 1.11
Ii
Argz
Argz
="2
if Imz
n
-"2
> O. and
if Imz < O.
00
stituting
eiy
= L
00
tr (iy l = 2:=
trikyk
k=O
k=O
At this point. our argument loses rigor because we haye not talked about infinite
series of complex numbers. let alone whether such series converge. Nevertheless.
if we merely take the last series as a formal expression and split it into two series
according to whether the index k is even (k = 2n) or odd (k = 2n + 1), we get
1
L __
i2ny2n + L
co
n=O (2n)!
=
=
co
n=O (2n
t y2n
_1_ (i2
n=O (2n)'
i2n+1y2n+l
+ I)!
n=O (2n
+ 1)1
f_l_(-lty2n+if
n=O (2n)!
= cos y
gives
eX
n=O (2n
t iy2n+l
(i2
1
+ I)!
(_lty2n+1
+ i sin y.
Thus, it seems the only possible value for eZ is that given by Equation (1-:11).
We will use this result freely from now on and. as stated. supply a rigorous proof
in Chapter 5.
If we set x = 0 and let e take the role of y in Equation (1~31 . we get a
famous result known as Euler's formula:
e'~= (O,I)=i
iOrl
\+;e
~
a
=e "=(l,O)=1
-x
tr Li~e-if=(.fi _.fi)=.fi=
2'
'ffi
2
If (j is a real number, eie will be located somewhere on the circle with radius
1 centered at the origin. This assertion is easy to verify because
Figure 1.12 illustrates the location of the points eie for various values of (j.
Note that, when (j = 1T, we get ei7r = (COS1T, sinn) = (-1, 0) = -1, so
Together with the rules for exponentiation that we will verify in Chapter 5,
Equation (1-35) has interesting applications. If Zl = r1 ei&, and Z2 = r2ei&2, then
ZlZ2
r1 ei&l r2ei&2
r1 r2ei(&,
+&2)
(ZlZ2)
arg
Zl,
Before proceeding with the proof, we recall two important facts about sets.
First, to establish the equality of two sets, we must show that each is a subset of
the other. Second, the sum of two sets is the sum of all combinations of elements
Zl
+ arg
Z2
Proof
Let 8 E arg (ZlZ2). Because ZlZ2 = r1 r2ei(O, +02), it follows from
Formula (1-37) that 81 + 82 E arg (ZlZ2). By Equation (1-29) there is some
integer n such that 8 = 81 + 82 + 2mr. Further, as Zl = r1 eiO" 81 E arg Zl
Likewise, Z2 = r2ei02 gives 82 E arg Z2. But if 82 E arg Z2, then 82 + 2mr E
arg Z2. This result shows that 8 = 81 + (82 + 2mr) E arg Zl + arg Z2. Thus,
arg(zlz2) ~ arg Zl +arg Z2. The proof that arg Zl +arg Z2 ~ arg(zlz2) is left
as an exercise.
1
Z-
= -1 [cos(-8) +isin(-8)]
r
z=
Zl =
Z2
(cos8 - isin8)
r1
r2
[cos (81
10
= _e-
82)]
iO
re-
and
r1 ei(O,-02).
r2
If z is in the first quadrant. the positions of the numbers z, z, and Z-l are
as shown in Figure 1.14 when Izi < 1. Figure 1.15 depicts the situation when
Izl > 1.
y
t=
(0,1)
= (0,1)
Figure 1.14
Relative positions of z, Z,
and z-l when Iz[ < 1.
Figure 1.15
Relative positions of z, Z,
and Z-l when
> 1.
Izl
EXAMPLE 1.13
Therefore,
1
z-l
If z
= ~ [cos
= 1 + i, then r =
(~4"') + i sin (7)] =
Izi
V2
and
= Argz =
%.
~ [ ...;;- i ...;;]
and has modulus
v'2
2'
v'2 -
(7r 7r)
8 [
.. (7r 7r)]
+ 2i.
2)3
(a) 1 - i.
(b)
-/3 + i.
(c)
(-1-i/3)2
(d) (1 - i)3 .
(e) 1+~v'3'
(f)
i2
1.
(g)
(a)
(b) (1+i)3=-2+2i.
(c)
2i(v'3+i) (l+i/3)
(d)
1~i
=4-4i.
(a) -4.
(b) 6-6i.
(c) -7i.
= -8.
7r .. 7r)
(d)
-2V3-2i.
(e)
(1~t)2'
(f) i+6../3
(g) 3 + 4i.
(h) (5 + 5i)3 .
4. Show that arg
Zl
+ arg
Z2
<;;; arg
+ ib
thus completing
Zl Z2.
1.3.
form.
p,
(a) eT.
(b) 4e-i .
h
(c) 8e'3.
6
(d) - 2e is''
(e) 2ie-i-
(f) 6eie'''.
(g) e2ei7r
(h)
el~e-'lrr
iff Z2 =
Z2
+ iV3
+ Arg
Zl
that
the equation
Arg Zl S % and -2"
this criterion.
<
<
Arg(zlzz)
Arg Z2 S
the identity
and Z2
-V3 + i.
Show that the equation
Z2 does not hold for the specific choice of Zl and Z2
8. Show
-2"
where
CZl,
arg( ~)
Arg
Describe
%.
+ Arg Z2 is true if
the set of points that meets
Zl
arg Zl - arg
i=
-Arg(z).
Prove your
Z2
i= 0,
arg
Zl
arg
Z2.
then
(a) Arg(zz) = O.
(b) Arg(z
14. Let
that
Zl,
Z2,
0' E
+ z)
= 0 when Re(z)
>
O.
at the vertex
arg
(Z2 -
Zl)
arg
(Z3 -
Zl)
is an expression
Zl.
+ i sin)
can be
in Figure 1.17.
NUMBERS,
The real numbers are deficient in the sense that not all algebraic operations on
them produce real numbers. Thus, for V=T to make sense, we must consider the
domain of complex numbers. Do complex numbers have this same deficiency?
That is, if we are to make sense of expressions such as ~.
must we appeal
to yet another new number system? The answer to this question is no. In other
words, any reasonable algebraic operation performed on complex numbers gives
complex numbers. Later we show how to evaluate intriguing expressions such as
ii. For now we only look at integral powers and roots of complex numbers.
The important players in this regard are the exponential and polar forms of
a nonzero complex number z = reie = r( cos B + i sin B). By the laws of exponents
(which, you recall, we have promised to prove in Chapter 5) we have
1.15
Show that
(-V3 -
( -v3;;:;- i )3 = ( 2e .(_5r.))3
t
-8i.
-6-
(3
(-1511
-1511)
= 2 et'(-15r.))
-6= 8 cos -6- + isin -6-
EXAMPLE 1.16
Solution
(-V3 -
Evaluate
i)30
(-V3 - i)30.
(2ei(-r))30
230e-i251r
_230.
the
e.
The real part of this expression is cos5 8 - 10 cos3 sin 2 + 5 cos sin 4
Equating
this to the real part of cos 5B + i sin 58 on the right side of Equation (1--10)
establishes the desired result.
Theorem
theorem of algebra)
Refer to Chapter
6.
EXAMPLE 1.18 Let P (z) = z3 + (2 - 2i) z2 + (-1 - 4i) z - 2. This polynomial of degree 3 can be written as P (z) = (z - i)2 (z + 2). Hence the equation
P (z) = 0 has solutions Zl = i, Z2 = i, and Z3 = -2. Thus, in accordance with
Theorem 1.4, we have three solutions, with Zl and Z2 being repeated roots.
Theorem 1.4 implies that if we can find n distinct solutions to the equation
zn = c (or zn - C = 0), we will have found all the solutions. We begin our
search for these solutions by looking at the simpler equation zn = 1. Solving
this equation will enable us to handle the more general one quite easily.
To solve zn = 1 we first note that, from Identities (1-29) and (1-37), we
can deduce an important condition that determines when two nonzero complex
numbers are equal. If we let Zl = Tl eiB, and Z2 = T2eiB2, then
where k is an integer. That is, two nonzero complex numbers are equal iff their
moduli agree and an argument of one equals an argument of the other to within
an integral multiple of 21f.
We now find all solutions to zn = 1 in two stages, with each stage corresponding to one direction in the iff part of Relation (1-41)' First, we show that
if we have a solution to zn = 1, then the solution must have a certain form.
Second, we show that any quantity with that form is indeed a solution.
For the first stage, suppose that z = TeiB is a solution to zn = 1. Putting
the latter equation in exponential form gives TneinB = 1 ei'O, so Relation (1-,n)
implies that Tn = 1 and
= + 21fk. In other words,
ne
e = 27rk ,
n
where k is an integer.
So, if z = TeiO is a solution to zn = 1, then Relation (1-42) must be true.
This observation completes the first stage of our solution strategy. For the second
2 k
'0'
2~k
stage, we note that if T = 1, and
= ~,
then z = Te' = e'--n is indeed a
solution to
zn
= 1 because
and k = 3, then
z = e
i
zn
(ei 2';,k) n
is a solution to
Z7
(e )
i
= ei6rr = 1.
i2k~
n
21fk.
=cos--+zsm--,
n
. 21fk
n
fork=0,1,2,
... ,n-1.
Wn
e'n
2Jr
= cos -
Wn
given by
2Jr
+ i sin -
Geometrically, the nth roots of unity are equally spaced points that lie on
the unit circle C1 (0) = {z : Izi = I} and form the vertices of a regular polygon
with n sides .
EXAMPLE
1
vaueszk-e
i 2"k
these solutions are 1, i. V2tV2) and V22iV2. The primitive 8th root of
V2
unity is W8 -- ei- -- ei~ -- cos2':
4 +isin2': 4- - V2
2 +i 2'
rn
p,
nO
--n + i-n
2
and
-OJ3
22
~-i =OJ~
i-n _,.,7
-~8
i4>+2nk
= pne
Zk
for k
= pn
solutions
cos
given by
+27rk
.. +27rk)
---+
2sm --n
= 0, 1, 2, ... , n - 1.
n sides.
((w~r =
by Wn = ei 2; increases an argument
l--Hi) contain n distinct values .
= 2
Formula
( cas
~ + 27r k
~ 3
of that number by
2:, so that
+ i sin
Expressions
~).
(1 .j I gives
+ i sin
~
2
+ 27rk)
3
for k = O. 1. 2.
'
Zo
Z2
= -2i.
1.5 (Quadratic
formula) If az2
+ bz + c =
},
EXAMPLE
Solution
1.21
The quadratic
z2
+ (1 + i) z + 5i = O.
formula gives
1
-(l+i)+
z -
-------------
[(1+i)2-4(1)(5i)r
2(1)
-(1+i)+(-18i)!
2
--------
.
'(_1L).
.
. 1.
1(-~+2k1r)
As -18~ = 18et 2, EquatlOns (1-45) gIve (-18~)2 = 182et
2
, for k = 0
and 1. In Cartesian form, this expression reduces to 3 - 3i and -3 + 3i. Thus,
our solution set is {-(Hi)+(3-3i)
-(Hi)+(-3+3i)}
or {I - 2i -2 + i}
2
'
'
(a)
(b)
(l+i):
(l-i) .
(1-40).
4. Let z be any nonzero complex number and let n be an integer. Show that zn
is a real number.
(a) (-2+2i)3.
+ (z)n
(b) (-1)5.
1
(c) (-64)4.
1
(d) (8)6.
+ 1)3
Z4 -
4z3
+ 6z2
4z
+5=a
if Zl
= z3.
4Y2 + i4Y2.
= r ""
n ( cos m(8+2.".k)
n
= i is a root.
+.~ Sln
. m(8+27I'k))
n
f or k = 0, 1,..
, n - 1.
(b) Use part (a) and De Moivre's formula to derive Lagrange's identity:
1 5in[(n+1)0]
1 + cas e + cas 2e + ... + cas ne = 2" + 2'~
, where 0 < e < 211.
SIn
12. If 1 =
(z -
13. Let
Zl)
Zk
(z -
Z2)'"
i= 1 be an
, Zn-1
(z -
2"
= 1 + Z + Z2 + ... + zn-l
+ z~ + ... + Z~-1 = O.
14. Equation
= 2.
(b) Use induction to verify the identity for all positive integers.
(c) How would you verify this identity for all negative integers?
Z4 + 4 = 0, and use them to demonstrate that
factored into two quadratics with real coefficients.
Z4
+ 4 can be
1- 1 is valid.
In this section we investigate some basic ideas concerning sets of points in the
plane. The first concept is that of a curve. Intuitively, we think of a curve as
a piece of string placed on a flat surface in some type of meandering pattern.
More formally, we define a curve to be the range of a continuous complex-valued
function z (t) defined on the interval [a. b]. That is, a curve C is the range of
a function given by z (t) = (x (t). y (t)) = x (t) + iy (t). for a ::::::
t ::::::
b, where
both x (t) and y (t) are continuous real-valued functions. If both x (t) and y (t)
are differentiable, we say that the curve is smooth. A curve for which x (t) and
y (t) are differentiable except for a finite number of points is called piecewise
smooth. We specify a curve C as
and say that z (t) is a parametrization for the curve C. Note that, with this
parametrization, we are specifying a direction for the curve C, saying that C is
a curve that goes from the initial point z (a) = (x (a), y (a)) = x (a) + iy (a)
to the terminal point z (b) = (x (b). y (b)) = x (b) + iy (b). If we had another
function whose range was the same set of points as z (t) but whose initial and
final points were reversed, we would indicate the curve that this function defines
by -C.
.
EXAMPLE
straight-line
Solution
Refer to Figure 1.21. The vector form of a line shows that the
direction of C is Zl - Zo0 As Zo is a point on C, its vector equation is
+ (Zl
= [xo + (Xl
C : z (t) = Zo
C : z (t)
- zo)
-
t,
xo) t]
for 0
:s; t :s;
+ i [YO + (Yl
1,
or
- YO)
t],
for 0
:s; t :s;
1.
f igurf' 1.22
The curve x (t) = sin 2t cas t. y (t) = sin 2t sin t for 0
forms a four-leaved rose.
:s; t :s;
21r. which
3;
3;, it
Note further that, at (0,0) , the curve has crossed over itself (at points other
than those corresponding with t = 0 and t = 27l"); we want to be able to distinguish when a curve does not cross over itself in this way. The curve C is called
simple if it does not cross over itself, except possibly at its initial and terminal
points. In other words, the curve C : z (t), for a ::::t ::::b, is simple provided that
z (td -I- z (t2) whenever t1 -I- t2, except possibly when t1 = a and t2 = b .
EXAMPLE 1.23
R can be parametrized
Solution
0::::
= (xo + Rcost)
= xo+iyo
+ i (Yo + Rsint) =
and radius
Zo
+ Reit,
for
Figure 1.23 shows that, as t varies from 0 to 27l", the circle is traversed
counterclockwise. If you were traveling around the circle in this manner, its
interior would be on your left. When a simple closed curve is parametrized in
this fashion, we say that the curve has a positive orientation.
We will have
more to say about this idea shortly.
We need to develop some vocabulary that will help describe sets of points in
the plane. One fundamental idea is that of an E neighborhood
of the point zoo
It is the open disk of radius E > 0 about Zo shown in Figure 1.24. Formally, it
is the set of all points satisfying the inequality {z : I z - Zo I < E} and is denoted
Do (zo). That is,
l.......
\
\
EXAMPLE 1.24 The solution sets of the inequalities Izl < 1, Iz - il < 2,
and Iz + 1 + 2il < 3 are neighborhoods of the points 0, i, and -1 - 2i, with
radii 1, 2, and 3, respectively. They can also be expressed as D1 (0), D2 (i), and
D3 (-1 - 2i).
We also define Dc (zo), the closed disk of radius c centered at zo, and
D; (zo) , the punctured disk of radius c centered at zo, as
Do(zo)={z:lz-zol:::;c},
D; (zo)
and
(1-50 )
(1-51)
Let S = D1 (0) = {z :
Solution
We show that every point of S is an interior point of S. Let zo be a
point of S. Then Izol < 1, and we can choose E = 1 ~ Izol > O. We claim that
Dc (zo) ~ S. If z E Dc (zo), then
EXAMPLE
domain.
1.26
{z : Re (z)
> O} is a
Solution
First we show that H is connected.
Let Zo and Zl be any two
points in H. We claim the obvious, that the straight-line segment G given by
Equation (1-48) lies entirely within H. To prove this claim, we let z (t*) =
Zo + (Zl - zo) t*, for some t* E [0, 1]' be an arbitrary point on G. We must show
that Re (z (t*)) > O. Now,
Re (z (t*)) = Re (zo
+ (Zl -
Re (zo (1 - t*))
(1 - t*) Re (zo)
zo) t*)
+ Re (zlt*)
+ t*Re(zl)'
If t* = 0, the last expression becomes Re (zo), which is greater than zero because
Zo E H. Likewise, if t* = 1, then Equation (1-53) becomes Re (Zl), which also is
positive. Finally, if 0 < t* < 1, then each term in Equation (1-53) is positive, so
in this case we also have Re (z (t*)) > O.
To show that H is open, we suppose without loss of generality that the
inequality Re (zo) :::;Re (zd holds. We claim that De (zo) <;;; H, where E =
Re (zo). We leave the proof of this claim as an exercise.
A domain, together with some, none, or all its boundary points, is called
a region.
For example, the horizontal strip {z: 1 < 1m (z) :::; 2} is a region.
A set formed by taking the union of a domain and its boundary is called a
closed region; thus, {z: 1 :s; 1m (z) :s; 2} is a closed region. A set S is said to
be a bounded
set if it can be completely contained in some closed disk, that
is, if there exists an R > 0 such that for each z in S we have Izl :s; R. The
rectangle given by {z : Ixl :s; 4 and Iyl :s; 3} is bounded because it is contained
inside the disk D5(0). A set that cannot be enclosed by any closed disk is called
an unbounded
set.
We mentioned earlier that a simple closed curve is positively oriented if its
interior is on the left when the curve is traversed. How do we know, though,
that any given simple closed curve will have an interior and exterior? Theorem
1.6 guarantees that this is indeed the case. It is due in part to the work of the
French mathematician Camille Jordan (1838-1922) .
Theorem 1.6 (The Jordan curve theorem) The complement of any simple
closed curve C can be partitioned into two mutually exclusive domains, I and
E, in such a way that I is bounded, E is unbounded, and C is the boundary
for both I and E. In addition, I U E U G is the entire complex plane. The
domain I is called the interior of G, and the domain E is called the exterior
01G.
+ 2t + i (t + 1)
Hint: Use x = t2
+ 2t, Y = t + 1 and
3. Find a parametrization
t.
y = x2 that
that
Re (zo)
>
O. Show that
Izi =
1 that joins
is counterclockwise
is clockwise.
Izi >
is not a domain.
= {z : Izl :::: I}
of the set S.
Iz
(v) {reiB : 0
(vi) {rei8
(vii) {z:
<r <
1 and -
r > 1 and
I}.
Z2
lie in
Dl
11. Let
{Z],
Z2,
...
of the neighborhood
D(zo)
z such
that
Iz + 21 < 1 or Iz - 21 < 1.
points,
(a) prove that if a set is closed, then it contains all its accumulations
(b) prove that if a set contains all its accumulation
16. Prove that
D]
Di
(0) .
points.
points of