Chapter 2

Complex Numbers

2.1Introduction to Complex Numbers

Video Lecture 4, Part 1: Complex Numbers (0:00)

Motivation

In algebra, the quadratic formula

z = − b ± b 2 − 4 a c 2 a (2.1)

provides the general solution of a quadratic equation

a z 2 + b z + c = 0. (2.2)

When the discriminant

d = b 2 − 4 a c (2.3)

is negative, the square root involves a negative number. To extend the number system so that such roots exist, we introduce a new quantity:

i = − 1 , i 2 = − 1. (2.4)
DefinitionComplex Numbers

A complex number is any number of the form

z = x + i y , (2.5)

where x , y ∈ ℝ and i is the imaginary unit defined by i 2 = − 1 .

  • x is the real part of z : Re ( z ) = x .

  • y is the imaginary part of z : Im ( z ) = y .

Both x and y may be positive, negative, or zero.

Special Cases

Example
  1. z = 4 + 3 i : Re ( z ) = 4 , Im ( z ) = 3 .

  2. z = − 5 i : Re ( z ) = 0 , Im ( z ) = − 5 (pure imaginary).

  3. z = 7 : Re ( z ) = 7 , Im ( z ) = 0 (pure real).

  4. z = − 2 + i 5 : Re ( z ) = − 2 , Im ( z ) = 5 .

Example

Express − 50 in the form x + i y .

Solution.

− 50 = − 1 ⋅ 25 ⋅ 2 = − 1 ⋅ 25 ⋅ 2 = i ⋅ 5 2 = 5 2 i . (2.6)

Motivation from Algebra

Consider

z 2 − 2 z + 2 = 0. (2.7)

Applying the quadratic formula:

z = 2 ± 4 − 8 2 = 2 ± − 4 2 = 2 ± 2 i 2 = 1 ± i . (2.8)

The solutions 1 ± i are complex numbers with both real and imaginary parts nonzero.

Example

Solve z 2 + z + 1 = 0 .

Solution.

z = − 1 ± 1 − 4 2 = − 1 ± − 3 2 = − 1 ± i 3 2 . (2.9)

Thus, the solutions are

z = − 1 2 + i 3 2 and z = − 1 2 − i 3 2 . (2.10)

2.1.4The Complex Plane

Video Lecture 4, Part 2: The Complex Plane and Polar Form (2:48)

DefinitionGeometric Interpretation of Complex Numbers

A complex number z = x + i y can be represented as a point ( x , y ) in a two-dimensional coordinate plane, called the complex plane or Argand diagram.

  • The horizontal axis represents the real part x = Re ( z ) .

  • The vertical axis represents the imaginary part y = Im ( z ) .

RemarkRectangular Form

When we write z = x + i y , we say z is in rectangular form, because ( x , y ) are the rectangular coordinates of the point representing z .

RemarkPolar Form

We can also locate a point using polar coordinates ( r , θ ) :

x = r cos ⁡ ( θ ) , y = r sin ⁡ ( θ ) . (2.11)

Then

z = x + i y = r ( cos ⁡ ( θ ) + i sin ⁡ ( θ ) ) . (2.12)

Later we will see that cos ⁡ ( θ ) + i sin ⁡ ( θ ) = e i θ , giving the compact form

z = r e i θ . (2.13)

Diagram of the Complex Plane

A geometric representation of the complex number z = x + iy in the complex plane, showing its real and imaginary components, magnitude r, and argument .
Figure 2.1. A geometric representation of the complex number z = x + i y in the complex plane, showing its real and imaginary components, magnitude r , and argument θ .

Radians vs Degrees

In calculus and advanced mathematics, radians are the standard measure for angles.

Worked Examples

Example

Represent 1 + i 3 in polar form.

Solution.

r = 1 2 + ( 3 ) 2 = 1 + 3 = 2 , (2.14)
θ = arctan ⁡ ( 3 1 ) = π 3 ( first quadrant, so the calculator’s angle is right ) . (2.15)

Thus:

1 + i 3 = 2 ( cos ⁡ ( π 3 ) + i sin ⁡ ( π 3 ) ) = 2 e i π / 3 . (2.16)
Example

Convert 3 e − i π / 2 to rectangular form.

Solution.

cos ⁡ ( − π 2 ) = 0 , sin ⁡ ( − π 2 ) = − 1. (2.17)

So:

3 e − i π / 2 = 3 ( 0 − i ) = − 3 i . (2.18)

Terminology and Notation

Definition 2.1.3Standard Notation for Complex Numbers

For a complex number

z = x + i y = r ( cos ⁡ ( θ ) + i sin ⁡ ( θ ) ) = r e i θ , (2.19)

we write:

Re ( z ) = x , Im ( z ) = y , | z | = r = x 2 + y 2 , arg ⁡ ( z ) = θ . (2.20)

The angle is only fixed up to whole turns, since θ and θ + 2 π name the same point. So arg ⁡ ( z ) is multivalued: it stands for the whole family θ + 2 n π , n ∈ ℤ . The principal argument, written with a capital A , is the one member of that family lying in a chosen half-open interval of length 2 π :

Arg ( z ) = Θ ∈ ( − π , π ] , so that z = | z | e i Θ . (2.21)
RemarkA notation worth keeping straight

Throughout this chapter, lower case means all values and upper case means the principal one:

arg ⁡ ( z ) = Arg ( z ) + 2 n π , ln ⁡ ( z ) = Ln ( z ) + 2 n π i , n ∈ ℤ . (2.22)

The distinction looks pedantic now, but it is exactly what makes ln ⁡ ( − 1 ) and i i have infinitely many values later in the chapter.

RemarkGetting the quadrant right

A calculator's arctan ⁡ ( y / x ) always returns an angle in ( − π / 2 , π / 2 ) , so it cannot tell z from − z : the points 1 + i and − 1 − i give the same ratio y / x = 1 . Always sketch z first. For x > 0 the calculator's angle is right; for x < 0 turn it by π , adding π when y ≥ 0 and subtracting π when y < 0 , which keeps the answer in ( − π , π ] :

Arg ( − 1 − i ) = arctan ⁡ ( 1 ) − π = π 4 − π = − 3 π 4 . (2.23)

For x = 0 the ratio is undefined; then θ = π / 2 if y > 0 and θ = − π / 2 if y < 0 .

2.1.9Complex Conjugate

Video Lecture 4, Part 3: Conjugates and Complex Algebra (7:14)

DefinitionComplex Conjugate

The complex conjugate of a complex number z = x + i y is defined by

z ― = x − i y . (2.24)

Geometrically, z ― is the reflection of the point z across the real axis in the complex plane.

Complex number z = x + iy and its conjugate z = x - iy shown as reflections across the real axis in the complex plane.
Figure 2.2. Complex number z = x + i y and its conjugate z ― = x − i y shown as reflections across the real axis in the complex plane.

Basic Identities

For all z , w ∈ ℂ :

z z ― = | z | 2 , z + w ― = z ― + w ― , z w ― = z ― w ― , ( z w ) ― = z ― w ―   ( w ≠ 0 ) . (2.25)

Useful reconstructions:

Re ( z ) = z + z ― 2 , Im ( z ) = z − z ― 2 i . (2.26)

Worked Examples

ExamplePolar data from rectangular

Write z = − 1 − i in polar form.

Solution.

| z | = ( − 1 ) 2 + ( − 1 ) 2 = 2 , θ = Arg ( z ) = − 3 π 4     ( III quadrant ) . (2.27)

Hence

z = 2 e − 3 π i / 4 = 2 ( cos ⁡ ( − 3 π / 4 ) + i sin ⁡ ( − 3 π / 4 ) ) . (2.28)
ExampleExtracting real/imag via conjugates

Let z = 2 − 3 i 1 + i . Compute | z | , Re ( z ) , and Im ( z ) .

Solution.

z = ( 2 − 3 i ) ( 1 − i ) ( 1 + i ) ( 1 − i ) = 2 − 2 i − 3 i + 3 i 2 1 − i 2 = 2 − 5 i − 3 2 = − 1 − 5 i 2 . (2.29)

Thus Re ( z ) = − 1 2 ,   Im ( z ) = − 5 2 , and

| z | = ( − 1 2 ) 2 + ( − 5 2 ) 2 = 1 2 1 + 25 = 26 2 . (2.30)

(Alternatively, | z | = z z ― .)

Quick Facts

2.2Complex Algebra

Video Lecture 4, Part 3: Conjugates and Complex Algebra (7:14)

In this section, we develop the algebraic techniques needed to manipulate complex numbers in both rectangular form x + i y and polar form r e i θ . The key operations include:

These tools allow us to treat complex numbers as a natural extension of real numbers while benefiting from their geometric interpretation.

2.2.1Simplifying to x + i y or r e i θ

For z 1 = x 1 + i y 1 and z 2 = x 2 + i y 2 :

z 1 ± z 2 = ( x 1 ± x 2 ) + i ( y 1 ± y 2 ) , z 1 z 2 = ( x 1 x 2 − y 1 y 2 ) + i ( x 1 y 2 + x 2 y 1 ) . (2.32)

Addition is vector addition: z 1 + z 2 is the fourth corner of the parallelogram with sides z 1 and z 2 . Multiplication is easier to picture in polar form, where it multiplies the lengths and adds the angles (see the Quick Facts above). To divide, multiply numerator and denominator by the conjugate of the denominator:

z 1 z 2 = z 1 z 2 ― | z 2 | 2 , z 2 ≠ 0. (2.33)
ExamplePower and product
( 1 + i ) 2 = 1 + 2 i + i 2 = 2 i , ( 2 + i ) ( 3 − i ) = 6 − 2 i + 3 i − i 2 = 7 + i . (2.34)
ExampleDivision to x + i y

2 + i 3 − i = ( 2 + i ) ( 3 + i ) ( 3 − i ) ( 3 + i ) = 6 + 5 i + i 2 3 2 − ( − 1 ) = 5 + 5 i 10 = 1 2 + 1 2 i .

2.2.2Conjugates of Expressions

The conjugate distributes over sums/products/quotients:

z 1 ± z 2 ― = z 1 ― ± z 2 ― , z 1 z 2 ― = z 1 ― z 2 ― , ( z 1 z 2 ) ― = z 1 ― z 2 ―   ( z 2 ≠ 0 ) . (2.35)

2.2.3Absolute Value (Modulus)

| z | = x 2 + y 2 = z z ― , | z w | = | z | | w | , | z w | = | z | | w | . (2.36)
Example

| 5 + 3 i 1 − i | = ( 5 ) 2 + 3 2 1 2 + ( − 1 ) 2 = 14 2 = 7 .

2.2.4Complex Equations

Video Lecture 4, Part 4: Complex Equations (11:25)

Two complex numbers are equal iff their real parts are equal and their imaginary parts are equal. Often, write z = x + i y and split into real/imaginary equations. In general, if f ( x , y ) and g ( x , y ) are real functions of the real unknowns x and y , then

f ( x , y ) + i g ( x , y ) = 0 ⟺ f ( x , y ) = 0    and    g ( x , y ) = 0 , (2.37)

because a complex number is zero only when both its real and imaginary parts are zero: one complex equation is two real equations.

Example

Solve ( x + i y ) 2 = 2 i for real x , y .

Solution.

( x + i y ) 2 = ( x 2 − y 2 ) + i ( 2 x y ) = 0 + 2 i (2.38)

Equate real and imaginary parts:

{ x 2 − y 2 = 0 , 2 x y = 2. (2.39)

From x 2 − y 2 = ( x − y ) ( x + y ) = 0 we have y = x or y = − x .

Case 1: y = x . Then 2 x y = 2 ⇒ 2 x 2 = 2 ⇒ x 2 = 1 . Hence x = ± 1 and y = x . This gives 1 + i and − 1 − i .

Case 2: y = − x . Then 2 x y = 2 ⇒ 2 x ( − x ) = 2 ⇒ − 2 x 2 = 2 ⇒ x 2 = − 1 , which is impossible for real x . So no solutions from this branch.

Therefore the solutions are

( x , y ) = ( 1 , 1 ) and ( x , y ) = ( − 1 , − 1 ) , (2.40)

i.e. z = 1 + i and z = − 1 − i .

Check. ( 1 + i ) 2 = 1 + 2 i − 1 = 2 i and ( − 1 − i ) 2 = 1 + 2 i − 1 = 2 i .

(Alternative via polar form.) 2 i = 2 e i π / 2 . Square roots have modulus 2 and arguments π 4 and π 4 + π , giving 2 e i π / 4 = 1 + i and 2 e i 5 π / 4 = − 1 − i .

(Picture.) In the x y -plane, which is the complex plane, the real equation x 2 = y 2 is the pair of lines y = ± x , and the imaginary equation x y = 1 is a hyperbola in the first and third quadrants. The solutions are where the curves meet: exactly the two points ( 1 , 1 ) and ( − 1 , − 1 ) . The line y = − x never meets the hyperbola, which is why Case 2 gave nothing. The two square roots are opposite points, ± ( 1 + i ) , as square roots always are.

2.2.5Graphs (Geometry in the Complex Plane)

Identify loci via z = x + i y , | z | = x 2 + y 2 , and distances:

| z − a | = R   is a circle of radius  R  centered at  a ∈ ℂ . (2.41)
Re z = c   is the vertical line  x = c ; Im z = c   is the horizontal line  y = c . (2.42)
Example
  1. | z − 1 | = 2 is the set of points at distance 2 from 1 , i.e. the circle ( x − 1 ) 2 + y 2 = 4 .

  2. arg ⁡ z = π / 4 is the ray y = x with x > 0 — a ray, not the whole line, because the argument of a point on the opposite side is π / 4 − π = − 3 π / 4 .

  3. | z − 1 | < | z + 1 | is the set of points closer to 1 than to − 1 , which is the half-plane x > 0 .

2.3Complex Infinite Series

Video Lecture 4, Part 5: Complex Infinite Series (14:43)

DefinitionComplex Series

A complex series is an expression of the form

∑ n = 0 ∞ z n , z n ∈ ℂ . (2.43)

Let

S N = ∑ n = 0 N z n (2.44)

be the sequence of partial sums. If the limit

lim N → ∞ S N = S (2.45)

exists and is a finite complex number S , then the series converges, and we write

∑ n = 0 ∞ z n = S . (2.46)

If the limit does not exist or is not finite, the series diverges.

TheoremConvergence of Complex Series

Let z n = a n + i b n with a n , b n ∈ ℝ . Then

∑ n = 0 ∞ z n   converges ⟺ ∑ n = 0 ∞ a n   and   ∑ n = 0 ∞ b n   both converge . (2.47)

In that case, the sum satisfies

∑ n = 0 ∞ z n = ∑ n = 0 ∞ a n + i ∑ n = 0 ∞ b n . (2.48)
Remark

In practice, check convergence of the real and imaginary parts separately, or use absolute convergence to simplify the test.

TheoremAbsolute Convergence in ℂ

A complex series ∑ z n is said to be absolutely convergent if the series

∑ n = 0 ∞ | z n | (2.49)

converges, where

| z n | = ( Re z n ) 2 + ( Im z n ) 2 . (2.50)

If ∑ | z n | converges, then the complex series ∑ z n also converges. In other words,

absolute convergence   ⟹   convergence . (2.51)

Moreover, any standard convergence test for real, non-negative series may be applied to the series ∑ | z n | .

Worked Examples

ExampleRatio Test

Test

∑ n = 0 ∞ ( 1 + i ) n 2 n . (2.52)

Here

ρ = lim n → ∞ | ( 1 + i ) n + 1 / 2 n + 1 ( 1 + i ) n / 2 n | = | 1 + i 2 | = 2 2 < 1. (2.53)

Thus the series converges absolutely (and hence converges). In fact it is a geometric series with ratio q = 1 + i 2 , so its sum is

∑ n = 0 ∞ ( 1 + i 2 ) n = 1 1 − q = 2 1 − i = 1 + i . (2.54)

In the plane the partial sums spiral into the point 1 + i : each new term is the previous one turned by 45 ∘ and shrunk by the factor 1 / 2 (Figure 2.3, left).

ExampleAlternating parts

Test

∑ n = 1 ∞ i n n . (2.55)

The terms cycle through i , − 1 , − i , 1 , … . Real part: ∑ ± 1 n alternating — converges by the alternating series test. Imaginary part: ∑ ± 1 n alternating — also converges. Hence the series converges (conditionally, not absolutely).

Written out exactly, with i 2 = − 1 , i 3 = − i , i 4 = 1 , the even terms give the real part and the odd terms the imaginary part:

∑ n = 1 ∞ i n n = ∑ n = 1 ∞ ( − 1 ) n 2 n + i ∑ n = 1 ∞ ( − 1 ) n − 1 2 n − 1 . (2.56)

Both are alternating series whose terms decrease to zero, while ∑ | z n | = ∑ 1 / n is a divergent p -series ( p = 1 2 ). The ratio test cannot decide here: ρ = | i | = i i ¯ = 1 . (Note that | i | = 1 , not i : a modulus is never imaginary.) In the plane the partial sums turn by a quarter at every step and close in only slowly, which is what conditional convergence looks like (Figure 2.3, right).

Partial sums of two complex series, drawn as walks in the complex plane (each segment is one term). Left: (1+i2)^n, whose steps turn by 45^ and shrink by 1/2, spirals quickly into its sum 1+i (red). Right: i^n/n turns by a quarter at every step, but its steps shrink only like 1/n: it still closes in on one point (red), slowly. That slowness is the mark of conditional convergence.
Figure 2.3. Partial sums of two complex series, drawn as walks in the complex plane (each segment is one term). Left: ∑ ( 1 + i 2 ) n , whose steps turn by 45 ∘ and shrink by 1 / 2 , spirals quickly into its sum 1 + i (red). Right: ∑ i n / n turns by a quarter at every step, but its steps shrink only like 1 / n : it still closes in on one point (red), slowly. That slowness is the mark of conditional convergence.
ExampleGeometric Series
∑ n = 0 ∞ ( r e i θ ) n (2.57)

converges iff | r | < 1 , in which case

∑ n = 0 ∞ ( r e i θ ) n = 1 1 − r e i θ . (2.58)

Quick Facts

For the ratio test applied to ∑ z n :

2.4Complex Power Series: Disk of Convergence

Video Lecture 4, Part 6: The Disk of Convergence (20:39)

DefinitionComplex Power Series

A complex power series centered at z 0 has the form

∑ n = 0 ∞ c n ( z − z 0 ) n , c n ∈ ℂ . (2.59)

It converges for all z within some disk centered at z 0 , and diverges for all z outside that disk.

2.4.1Radius and Disk of Convergence

TheoremRadius of Convergence (Ratio Test)

If

ρ = lim n → ∞ | c n + 1 c n | exists, then R = 1 ρ , (2.60)

with conventions:

ρ = 0 ⇒ R = ∞ , ρ = ∞ ⇒ R = 0. (2.61)

A complex power series converges absolutely when

| z − z 0 | < R , (2.62)

and diverges when

| z − z 0 | > R . (2.63)

The set

{ z ∈ ℂ : | z − z 0 | < R } (2.64)

is called the disk of convergence. On the boundary | z − z 0 | = R , convergence must be tested separately. With z = x + i y and z 0 = x 0 + i y 0 , the condition | z − z 0 | < R reads

( x − x 0 ) 2 + ( y − y 0 ) 2 < R 2 , (2.65)

the inside of the circle of radius R about z 0 . (The right side is R 2 , not R : the left side is a squared distance.) The interval a − R < x < a + R of a real power series is simply the part of this disk that lies on the real axis.

ExampleMaclaurin Series for ln ⁡ ( 1 + z )
ln ⁡ ( 1 + z ) = ∑ n = 1 ∞ ( − 1 ) n + 1 n z n . (2.66)

Ratio test:

ρ = lim n → ∞ | ( − 1 ) n + 2 z n + 1 / ( n + 1 ) ( − 1 ) n + 1 z n / n | = lim n → ∞ | z | n n + 1 = | z | . (2.67)

Thus R = 1 , so the disk of convergence is | z | < 1 .

The boundary circle. On | z | = 1 write z = e i θ ; the terms ( − 1 ) n + 1 e i n θ / n have modulus 1 / n → 0 . For a complex series that is not enough, and the alternating series test (a test for real series) does not apply to complex terms. Test points instead:

  • z = 1 : 1 − 1 2 + 1 3 − ⋯ is alternating and converges, to ln ⁡ ( 2 ) .

  • z = − 1 : every term is ( − 1 ) n + 1 ( − 1 ) n n = − 1 n , so the series is − ( 1 + 1 2 + 1 3 + ⋯ ) and diverges, as it must, since ln ⁡ ( 1 + z ) is ln ⁡ ( 0 ) there.

  • z = i : the real part is ∑ n = 1 ∞ ( − 1 ) n + 1 2 n = 1 2 ln ⁡ ( 2 ) and the imaginary part is ∑ n = 1 ∞ ( − 1 ) n − 1 2 n − 1 = π 4 , both alternating, so the series converges, to

    ln ⁡ ( 1 + i ) = 1 2 ln ⁡ ( 2 ) + i π 4 = ln ( 2 e i π / 4 ) , (2.68)

    which is exactly the logarithm of 1 + i = 2 e i π / 4 .

In fact the series converges at every point of the circle except z = − 1 (a finer test, Dirichlet's, shows it). One bad point on the circle is what limits the radius to 1 (Figure 2.4).

The disk of convergence |z| &lt; 1 of the series of (1+z). On the boundary circle the series converges at z = 1 and z = i (and at every other point) except z = -1, where it becomes minus the harmonic series and (1+z) itself is undefined.
Figure 2.4. The disk of convergence | z | < 1 of the series of ln ⁡ ( 1 + z ) . On the boundary circle the series converges at z = 1 and z = i (and at every other point) except z = − 1 , where it becomes minus the harmonic series and ln ⁡ ( 1 + z ) itself is undefined.
ExampleShifted Center
∑ n = 0 ∞ ( z + 1 − i ) n 3 n . (2.69)

Ratio test:

ρ = lim n → ∞ | ( z + 1 − i ) n + 1 / 3 n + 1 ( z + 1 − i ) n / 3 n | = | z + 1 − i | 3 . (2.70)

Thus convergence requires | z + 1 − i | < 3 . The disk is centered at z 0 = − 1 + i with radius R = 3 .

RemarkProperties of the Disk of Convergence
  • If R = ∞ , the power series converges for all z ∈ ℂ . In this case, the function defined by the series is called entire.

  • If R = 0 , the series converges only at the center z = z 0 .

  • For all z with | z − z 0 | < R , convergence is uniform on compact subsets of the disk. Term-by-term differentiation and integration are therefore valid within this region (see Theorem in Section 2.4).

2.5Euler's Formula

Video Lecture 4, Part 7: Euler's Formula (25:15)

TheoremEuler's Formula

For any real angle θ ,

e i θ = cos ⁡ ( θ ) + i sin ⁡ ( θ ) . (2.71)

This identity connects the complex exponential with the trigonometric functions and plays a central role in oscillations, waves, quantum mechanics, and electrical engineering.

Proof Using Power Series

RemarkStrategy of the Proof

We will:

  1. Start from the power series definition of e x and substitute x = i θ .

  2. Expand ( i θ ) n and separate the terms with even and odd n .

  3. Simplify powers of i and identify the real and imaginary parts with the known series for cos ⁡ ( θ ) and sin ⁡ ( θ ) .

Step 1. Start from the series for e x .

For any real (or complex) number x ,

e x = ∑ n = 0 ∞ x n n ! . (2.72)

Substitute x = i θ (with θ ∈ ℝ ):

e i θ = ∑ n = 0 ∞ ( i θ ) n n ! = ( i θ ) 0 0 ! + ( i θ ) 1 1 ! + ( i θ ) 2 2 ! + ( i θ ) 3 3 ! + ( i θ ) 4 4 ! + ⋯ . (2.73)

Step 2. Write out the first few terms explicitly.

e i θ = 1 + i θ + ( i θ ) 2 2 ! + ( i θ ) 3 3 ! + ( i θ ) 4 4 ! + ( i θ ) 5 5 ! + ⋯ . (2.74)

Now expand the powers of i :

i 0 = 1 , i 1 = i , i 2 = − 1 , i 3 = − i , i 4 = 1 , i 5 = i ,   … (2.75)

So we obtain:

e i θ = 1 + i θ + i 2 θ 2 2 ! + i 3 θ 3 3 ! + i 4 θ 4 4 ! + i 5 θ 5 5 ! + ⋯ = 1 + i θ + ( − 1 ) θ 2 2 ! + ( − i ) θ 3 3 ! + ( 1 ) θ 4 4 ! + ( i ) θ 5 5 ! + ⋯ . (2.76)

Step 3. Group real and imaginary terms.

Group all the terms without i (real part) and all the terms with a factor i (imaginary part):

e i θ = ( 1 − θ 2 2 ! + θ 4 4 ! − θ 6 6 ! + ⋯ ) ⏟ real part + i ( θ − θ 3 3 ! + θ 5 5 ! − θ 7 7 ! + ⋯ ) ⏟ imaginary part . (2.77)

Thus we can write

e i θ = A ( θ ) + i B ( θ ) , (2.78)

where

A ( θ ) = 1 − θ 2 2 ! + θ 4 4 ! − θ 6 6 ! + ⋯ , (2.79)
B ( θ ) = θ − θ 3 3 ! + θ 5 5 ! − θ 7 7 ! + ⋯ . (2.80)

Step 4. Recall the power series for cos ⁡ ( θ ) and sin ⁡ ( θ ) .

From earlier results on Maclaurin series:

cos ⁡ ( θ ) = ∑ n = 0 ∞ ( − 1 ) n θ 2 n ( 2 n ) ! = 1 − θ 2 2 ! + θ 4 4 ! − θ 6 6 ! + ⋯ , (2.81)
sin ⁡ ( θ ) = ∑ n = 0 ∞ ( − 1 ) n θ 2 n + 1 ( 2 n + 1 ) ! = θ − θ 3 3 ! + θ 5 5 ! − θ 7 7 ! + ⋯ . (2.82)

Comparing term by term, we see

A ( θ ) = cos ⁡ ( θ ) , B ( θ ) = sin ⁡ ( θ ) . (2.83)

Therefore,

e i θ = A ( θ ) + i B ( θ ) = cos ⁡ ( θ ) + i sin ⁡ ( θ ) , (2.84)

which completes the proof.

Geometric Meaning

Euler’s formula shows that e i θ lies on the unit circle in the complex plane:

| e i θ | = 1 , arg ⁡ ( e i θ ) = θ . (2.85)

Thus multiplying by e i θ corresponds to a rotation by angle θ about the origin.

The complex number e^i lies on the unit circle, at angle from the positive real axis. Multiplying any complex number by e^i rotates it by .
Figure 2.5. The complex number e i θ lies on the unit circle, at angle θ from the positive real axis. Multiplying any complex number by e i θ rotates it by θ .

The series itself can be watched as it adds up. Each term of ∑ ( i θ ) n / n ! is the previous one multiplied by i θ / n : turned by a quarter turn (the factor i ) and rescaled. For θ = π the partial sums walk 1 , then up by π , then left by π 2 / 2 , then down, then right, in steps that eventually shrink because n ! wins, and they close in on the point − 1 of the unit circle: e i π = cos ⁡ ( π ) + i sin ⁡ ( π ) = − 1 .

The partial sums of e^i = n=0^ (i)^nn! as a walk in the complex plane, one segment per term. Every step turns a quarter to the left of the previous one, because each term is the previous one times i/n. The early steps grow, the later ones shrink, and the walk lands on -1 (red) on the unit circle (gray).
Figure 2.6. The partial sums of e i π = ∑ n = 0 ∞ ( i π ) n n ! as a walk in the complex plane, one segment per term. Every step turns a quarter to the left of the previous one, because each term is the previous one times i π / n . The early steps grow, the later ones shrink, and the walk lands on − 1 (red) on the unit circle (gray).

Applications of Euler's Formula

RemarkPolar Form

Any nonzero complex number z can be written as

z = r e i θ , r = | z | ,   θ = arg ⁡ ( z ) . (2.86)

This is the polar form of z .

RemarkMultiplication and Division

If z 1 = r 1 e i θ 1 and z 2 = r 2 e i θ 2 , then

z 1 z 2 = r 1 r 2 e i ( θ 1 + θ 2 ) , z 1 z 2 = r 1 r 2 e i ( θ 1 − θ 2 ) . (2.87)

Thus multiplication adds angles, and division subtracts angles.

RemarkRoots of Unity

The n -th roots of 1 are

z k = e i 2 π k n , k = 0 , 1 , … , n − 1 , (2.88)

which are equally spaced points on the unit circle.

Euler's formula also turns every oscillation into a complex number, its phasor. Adding two waves then becomes adding two complex numbers; this is worked out in Section 2.14.2, among the applications at the end of the chapter.

Example

Write 1 + i in polar form.

Solution.

The point 1 + i sits in the first quadrant, so arctan needs no quadrant correction:

r = 1 2 + 1 2 = 2 , θ = arctan ( 1 1 ) = π 4 . (2.89)

Thus

1 + i = 2 e i π / 4 . (2.90)

Polar form is what makes powers and roots easy; we use this result in Section 2.8.

Example

Compute ( 1 + i ) 8 .

Solution.

Using the polar form just found, raising to a power multiplies the angle by the exponent and raises the modulus to it:

( 1 + i ) 8 = ( 2 e i π / 4 ) 8 = ( 2 ) 8 e i 8 π / 4 = 16 e i 2 π = 16. (2.91)

The angle 8 π / 4 = 2 π is a full turn, which returns to the positive real axis — so the answer is real. Doing this by repeated multiplication in rectangular form would take eight expansions.

Example

Find the cube roots of 8 e i π .

Solution.

Here r = 8 and θ = π . Taking the cube root divides the angle by 3 , but the angle is only defined up to 2 π , so all three choices must be included:

z k = 8 1 / 3 e i ( π + 2 π k ) / 3 = 2 e i ( π + 2 π k ) / 3 , k = 0 , 1 , 2. (2.92)

Writing out the three values,

2 e i π / 3 , 2 e i π , 2 e i 5 π / 3 . (2.93)

They all have modulus 2 and are spaced 2 π / 3 apart, so they sit at the vertices of an equilateral triangle on the circle of radius 2 .

2.6Elementary Functions of Complex Numbers

We extend familiar real-valued functions such as the exponential, logarithmic, and trigonometric functions to complex inputs z ∈ ℂ .

These generalizations are most naturally defined using their power series expansions, since power series:

Polynomials and Rational Functions

Polynomials and quotients of polynomials in z are handled by direct substitution and algebra.

Example

Take for example f ( z ) = z 2 + 1 z − 3 . Find f ( i − 2 ) .

f ( i − 2 ) = ( i − 2 ) 2 + 1 ( i − 2 ) − 3 = − 1 − 4 i + 4 + 1 i − 5 = 4 − 4 i i − 5 . (2.94)

Multiply numerator and denominator by the conjugate − i − 5 :

4 − 4 i i − 5 ⋅ − i − 5 − i − 5 = ( 4 − 4 i ) ( − i − 5 ) ( i − 5 ) ( − i − 5 ) . (2.95)

Expand the numerator and simplify:

( 4 − 4 i ) ( − i − 5 ) = 4 ( − i − 5 ) − 4 i ( − i − 5 ) = − 4 i − 20 + 4 i 2 ⏟ − 4 + 20 i = ( − 24 ) + 16 i . (2.96)

Denominator:

( i − 5 ) ( − i − 5 ) = − i 2 − 5 i + 5 i + 25 = 1 + 25 = 26. (2.97)

Therefore

f ( i − 2 ) = − 24 + 16 i 26 = 8 i − 12 13 . (2.98)

2.7The Exponential Function

Video Lecture 5, Part 1: The Complex Exponential (0:00)

DefinitionComplex Exponential

The exponential function for z ∈ ℂ is defined by the power series

e z = ∑ n = 0 ∞ z n n ! = 1 + z + z 2 2 ! + z 3 3 ! + ⋯ . (2.99)

This series converges for all complex numbers z .

TheoremMain Properties of e z

For all complex numbers z , z 1 , z 2 ∈ ℂ ,

  1. e z 1 + z 2 = e z 1 e z 2 ,

  2. d d z ( e z ) = e z ,

  3. If z = x + i y , then e z = e x ( cos ⁡ ( y ) + i sin ⁡ ( y ) ) .

(a) Show that e z 1 + z 2 = e z 1 e z 2 .

Expand each exponential:

e z 1 + z 2 = 1 + ( z 1 + z 2 ) + ( z 1 + z 2 ) 2 2 ! + ( z 1 + z 2 ) 3 3 ! + ⋯ . (2.100)

Similarly,

e z 1 e z 2 = ( 1 + z 1 + z 1 2 2 ! + z 1 3 3 ! + ⋯ ) ( 1 + z 2 + z 2 2 2 ! + z 2 3 3 ! + ⋯ ) . (2.101)

Multiplying out gives:

e z 1 e z 2 = 1 + ( z 1 + z 2 ) + z 1 2 + 2 z 1 z 2 + z 2 2 2 ! + ⋯ (2.102)

Each coefficient matches term-by-term the expansion of e z 1 + z 2 . In general, the terms of total order n in the product are

∑ k = 0 n z 1 k k ! z 2 n − k ( n − k ) ! = 1 n ! ∑ k = 0 n ( n k ) z 1 k z 2 n − k = ( z 1 + z 2 ) n n ! (2.103)

by the binomial theorem, which is the term of order n of e z 1 + z 2 . Thus,

e z 1 + z 2 = e z 1 e z 2 . (2.104)

(b) Show that d d z ( e z ) = e z .

Differentiate term-by-term:

d d z e z = d d z ( 1 + z + z 2 2 ! + z 3 3 ! + ⋯ ) = 0 + 1 + 2 z 2 ! + 3 z 2 3 ! + ⋯ . (2.105)

Simplify:

d d z e z = 1 + z + z 2 2 ! + z 3 3 ! + ⋯ = e z . (2.106)
d d z ( e z ) = e z . (2.107)

(c) Simplify e z into real and imaginary parts.

Let z = x + i y . Then

e z = e x + i y = e x e i y . (2.108)

Using Euler’s formula:

e i y = cos ⁡ ( y ) + i sin ⁡ ( y ) . (2.109)

So,

e x + i y = e x ( cos ⁡ ( y ) + i sin ⁡ ( y ) ) . (2.110)
Example

Compute e 2 + i e − 1 + 3 i without expanding series.

Solution.

Use property (a) to add exponents first, then property (c) to separate real and imaginary parts:

e 2 + i e − 1 + 3 i = e ( 2 + i ) + ( − 1 + 3 i ) = e 1 + 4 i = e 1 ( cos ⁡ ( 4 ) + i sin ⁡ ( 4 ) ) . (2.111)

Note the division of labour in e x + i y = e x ( cos ⁡ ( y ) + i sin ⁡ ( y ) ) : the real part of the exponent sets the size e x , and the imaginary part sets the angle y . This one fact drives the rest of the chapter.

Because d d t e i ω t = i ω e i ω t , differentiating a complex exponential only multiplies it by a constant. Section 2.14.3 uses this to replace the differential equation of an AC circuit by arithmetic with complex numbers.

2.8Powers and Roots of Complex Numbers

Video Lecture 5, Part 2: Powers and Roots (4:10)

Using the rules for multiplication and division of complex numbers in polar form, let

z = r e i θ , r = | z | ≥ 0 ,   θ = arg ⁡ ( z ) . (2.112)
TheoremPowers of a Complex Number

For any integer n ,

z n = ( r e i θ ) n = r n e i n θ . (2.113)

In words: to take the n -th power of a complex number, raise the modulus to the n -th power and multiply the angle by n .

When r = 1 , this becomes De Moivre's formula:

( e i θ ) n = ( cos ⁡ ( θ ) + i sin ⁡ ( θ ) ) n = cos ⁡ ( n θ ) + i sin ⁡ ( n θ ) . (2.114)
Remark

De Moivre's formula is a machine for trigonometric identities. Expanding the left side of (2.114) for n = 2 gives cos 2 ⁡ ( θ ) − sin 2 ⁡ ( θ ) + 2 i sin ⁡ ( θ ) cos ⁡ ( θ ) , and matching real and imaginary parts against cos ⁡ ( 2 θ ) + i sin ⁡ ( 2 θ ) yields both double-angle formulas at once, with no trigonometry at all.

Theorem n -th Roots of a Complex Number

The n -th root of z , written z 1 / n , is a complex number whose n -th power is z . From (2.113), the n -th roots of z = r e i θ are

z 1 / n = ( r e i θ e i ( 2 m π ) ) 1 / n = r 1 / n e i ( θ + 2 m π ) / n = r n ( cos ⁡ ( θ + 2 m π n ) + i sin ⁡ ( θ + 2 m π n ) ) , (2.115)

where m = 0 , 1 , … , n − 1

Here n = 1 , 2 , 3 , … is a positive integer. In a power the extra turns disappear, e i 2 m n π = 1 , but in a root they do not: e i 2 m π / n ≠ 1 unless m is a multiple of n . So m = 0 , 1 , … , n − 1 give n different roots, m = n gives the first one again, and the n roots all have modulus r 1 / n with angles 2 π / n apart: they are equally spaced on a circle.

Example

Compute ( 1 + i ) 8 .

Write

1 + i = 2 e i π / 4 . (2.116)

Then

( 1 + i ) 8 = ( 2 ) 8 e i 8 π / 4 = 16 e i 2 π = 16. (2.117)
Example

Compute ( cos ⁡ ( π 10 ) + i sin ⁡ ( π 10 ) ) 25 .

Solution.

By De Moivre's formula, with e i π / 10 = cos ⁡ ( π 10 ) + i sin ⁡ ( π 10 ) ,

( e i π / 10 ) 25 = e i 25 π / 10 = e i 5 π / 2 = e i 2 π e i π / 2 = e i π / 2 = cos ⁡ ( π 2 ) + i sin ⁡ ( π 2 ) = i . (2.118)

Each factor turns by π / 10 = 18 ∘ ; twenty-five of them make 450 ∘ , one full turn plus a quarter, ending straight up at i (Figure 2.7).

The powers (e^i/10)^k on the unit circle, k = 1, 2, : each step turns by 18^. After k = 25 steps the point has gone once around and a quarter more, and stands at i.
Figure 2.7. The powers ( e i π / 10 ) k on the unit circle, k = 1 , 2 , … : each step turns by 18 ∘ . After k = 25 steps the point has gone once around and a quarter more, and stands at i .
Example

Compute ( − 3 + i ) 3 .

Here

r = ( − 3 ) 2 + 1 2 = 2 , θ = arg ⁡ ( − 3 + i ) = 5 π 6 . (2.119)

Thus

( − 3 + i ) 3 = 2 3 e i 3 ( 5 π / 6 ) = 8 e i 5 π / 2 = 8 e i ( π / 2 ) = 8 i . (2.120)
Example

Fourth roots of − 16 .

We write

− 16 = 16 e i π . (2.121)

With n = 4 , the roots are

w k = 16 1 / 4 e i ( π + 2 π k ) / 4 = 2 e i ( π / 4 + π k / 2 ) , k = 0 , 1 , 2 , 3. (2.122)

Explicitly:

2 e i π / 4 = 2 ( 1 + i ) , 2 e i 3 π / 4 = 2 ( − 1 + i ) , 2 e i 5 π / 4 = 2 ( − 1 − i ) , 2 e i 7 π / 4 = 2 ( 1 − i ) . (2.123)
Example

Cube roots of 8 e i π .

Here r = 8 , θ = π , n = 3 . Thus

w k = 2 e i ( π + 2 π k ) / 3 , k = 0 , 1 , 2 , (2.124)

i.e.

2 e i π / 3 , 2 e i π , 2 e i 5 π / 3 . (2.125)
Example

Find all the cube roots of 8 , that is, solve z 3 = 8 .

Solution.

Write 8 = 8 e i 2 m π . Then

z = ( 8 e i 2 m π ) 1 / 3 = 2 e i 2 m π / 3 = 2 [ cos ⁡ ( 2 m π 3 ) + i sin ⁡ ( 2 m π 3 ) ] , m = 0 , 1 , 2 : (2.126)
m = 0 :   2 , m = 1 :   2 ( − 1 2 + i 3 2 ) = − 1 + i 3 , m = 2 :   2 ( − 1 2 − i 3 2 ) = − 1 − i 3 . (2.127)

m = 3 gives 2 [ cos ⁡ ( 2 π ) + i sin ⁡ ( 2 π ) ] = 2 again. Check: ( − 1 + i 3 ) 3 = ( 2 e i 2 π / 3 ) 3 = 8 e i 2 π = 8 . Only one of the three roots is real; the principal angle of the third is − 2 π / 3 (the same point as 4 π / 3 ). The three roots sit 120 ∘ apart on the circle of radius 2 , at the corners of an equilateral triangle (Figure 2.8).

The three cube roots of 8 on the circle |z| = 2, spaced 2/3 apart: the corners of an equilateral triangle. Only 2 is real.
Figure 2.8. The three cube roots of 8 on the circle | z | = 2 , spaced 2 π / 3 apart: the corners of an equilateral triangle. Only 2 is real.
ExampleFifth roots of unity

Solve w 5 = 1 .

We write

1 = e i ⋅ 0 . (2.128)

Then the solutions are

w k = e i ( 0 + 2 π k ) / 5 = e i 2 π k / 5 , k = 0 , 1 , 2 , 3 , 4. (2.129)

These five points form a regular pentagon on the unit circle.

2.9Trigonometric Functions of a Complex Variable

Video Lecture 5, Part 3: Sine and Cosine of a Complex Number (8:59)

Recall that Euler’s formula gives

e i θ = cos ⁡ ( θ ) + i sin ⁡ ( θ ) , e − i θ = cos ⁡ ( θ ) − i sin ⁡ ( θ ) . (2.130)

Adding and subtracting these expressions,

cos ⁡ ( θ ) = e i θ + e − i θ 2 , sin ⁡ ( θ ) = e i θ − e − i θ 2 i . (2.131)
TheoremComplex Sine and Cosine

For any complex number z = x + i y ∈ ℂ ,

sin ⁡ ( z ) = e i z − e − i z 2 i , cos ⁡ ( z ) = e i z + e − i z 2 . (2.132)
RemarkExtension to Complex Domain

These formulas extend the real trigonometric functions to complex arguments while preserving their power series definitions and their analytic properties.

Example
  1. cos ⁡ ( 2 i ) = e i ( 2 i ) + e − i ( 2 i ) 2 = e − 2 + e 2 2 ≡ cosh ⁡ ( 2 ) = 3.762 …

    In the same way cos ⁡ ( i ) = e i ⋅ i + e − i ⋅ i 2 = e − 1 + e 1 2 = 1 2 ( 1 e + e ) = 1.543 … A cosine larger than 1 is no paradox: on the real axis | cos ⁡ ( x ) | ≤ 1 , but along the imaginary axis cos ⁡ ( i y ) = e − y + e y 2 grows without bound (Figure 2.9).

  2. sin ( π 2 + i ln ⁡ ( 2 ) )

    sin ( π 2 + i ln ⁡ ( 2 ) ) = e i ( π / 2 + i ln ⁡ ( 2 ) ) − e − i ( π / 2 + i ln ⁡ ( 2 ) ) 2 i = e i π / 2 − ln ⁡ ( 2 ) − e − i π / 2 + ln ⁡ ( 2 ) 2 i = i e − ln ⁡ ( 2 ) + i e ln ⁡ ( 2 ) 2 i ( since  e i π / 2 = i ,   e − i π / 2 = − i ) = i ⋅ 1 2 + 2 i 2 i = ( 1 2 + 2 ) 1 2 = 1 4 + 1 = 5 4 . (2.133)
The cosine on the real axis, (x) (blue), stays between -1 and 1. On the imaginary axis it becomes (iy) = (y) (green), which grows without bound; at y = 1 it is already (i) = 1.543
Figure 2.9. The cosine on the real axis, cos ⁡ ( x ) (blue), stays between − 1 and 1 . On the imaginary axis it becomes cos ⁡ ( i y ) = cosh ⁡ ( y ) (green), which grows without bound; at y = 1 it is already cos ⁡ ( i ) = 1.543 …
ExampleThe Pythagorean identity still holds

Prove that sin 2 ⁡ ( z ) + cos 2 ⁡ ( z ) = 1 for every complex z .

Solution.

sin 2 ⁡ ( z ) = ( e i z − e − i z 2 i ) 2 = e 2 i z + e − 2 i z − 2 − 4 , cos 2 ⁡ ( z ) = ( e i z + e − i z 2 ) 2 = e 2 i z + e − 2 i z + 2 4 . (2.134)

Adding the two, the e 2 i z and e − 2 i z terms cancel and only the constants survive:

sin 2 ⁡ ( z ) + cos 2 ⁡ ( z ) = 2 − e 2 i z − e − 2 i z 4 + e 2 i z + e − 2 i z + 2 4 = 4 4 = 1 . (2.135)

Note this is now an algebraic fact about exponentials, with no triangles anywhere. It holds for all complex z , even where | sin ⁡ ( z ) | exceeds 1 .

Example

Show that d d z cos ⁡ ( z ) = − sin ⁡ ( z ) .

Solution.

d d z ( e i z + e − i z 2 ) = i e i z − i e − i z 2 = − e i z − e − i z 2 i = − sin ⁡ ( z ) . (2.136)

2.10Hyperbolic Functions

Video Lecture 5, Part 4: Hyperbolic Functions (13:12)

Assume z = i y is purely imaginary. Recall the complex definitions:

sin ⁡ ( z ) = e i z − e − i z 2 i , cos ⁡ ( z ) = e i z + e − i z 2 . (2.137)

Relation with Imaginary Arguments

Compute:

sin ⁡ ( i y ) = e i ( i y ) − e − i ( i y ) 2 i = e − y − e y 2 i = 1 i ⋅ e − y − e y 2 = i e y − e − y 2 ≡ i sinh ⁡ ( y ) , (2.138)

and

cos ⁡ ( i y ) = e i ( i y ) + e − i ( i y ) 2 = e − y + e y 2 ≡ cosh ⁡ ( y ) . (2.139)
TheoremRelation Between Trigonometric and Hyperbolic Functions

For any real number y ,

sin ⁡ ( i y ) = i sinh ⁡ ( y ) , cos ⁡ ( i y ) = cosh ⁡ ( y ) . (2.140)

Equivalently,

sinh ⁡ ( i y ) = i sin ⁡ ( y ) , cosh ⁡ ( i y ) = cos ⁡ ( y ) . (2.141)
DefinitionHyperbolic Functions

For any complex number z ∈ ℂ , define

sinh ⁡ ( z ) = e z − e − z 2 , cosh ⁡ ( z ) = e z + e − z 2 . (2.142)

From these:

tanh ⁡ ( z ) = sinh ⁡ ( z ) cosh ⁡ ( z ) , coth ( z ) = cosh ⁡ ( z ) sinh ⁡ ( z ) , (2.143)
sech ( z ) = 1 cosh ⁡ ( z ) , csch ( z ) = 1 sinh ⁡ ( z ) . (2.144)
Example

Show that cosh 2 ⁡ ( z ) − sinh 2 ⁡ ( z ) = 1

cosh 2 ⁡ ( z ) = ( e z + e − z 2 ) 2 = e 2 z + e − 2 z + 2 4 , (2.145)
sinh 2 ⁡ ( z ) = ( e z − e − z 2 ) 2 = e 2 z + e − 2 z − 2 4 . (2.146)

Hence

cosh 2 ⁡ ( z ) − sinh 2 ⁡ ( z ) = 1. (2.147)

The order matters: sinh 2 ⁡ ( z ) − cosh 2 ⁡ ( z ) = − 1 . Compared with cos 2 ⁡ ( z ) + sin 2 ⁡ ( z ) = 1 , one sign has changed, because cos ⁡ ( i y ) = cosh ⁡ ( y ) while sin ⁡ ( i y ) = i sinh ⁡ ( y ) carries a factor i , and i 2 = − 1 .

The name comes from geometry. The point ( cos ⁡ ( t ) , sin ⁡ ( t ) ) satisfies x 2 + y 2 = 1 and runs around the unit circle; the point ( cosh ⁡ ( t ) , sinh ⁡ ( t ) ) satisfies x 2 − y 2 = 1 and runs along the unit hyperbola (Figure 2.10).

((t), (t)) lies on the circle x^2 + y^2 = 1; ((t), (t)) lies on the right branch of the hyperbola x^2 - y^2 = 1. Hence the names circular and hyperbolic functions.
Figure 2.10. ( cos ⁡ ( t ) , sin ⁡ ( t ) ) lies on the circle x 2 + y 2 = 1 ; ( cosh ⁡ ( t ) , sinh ⁡ ( t ) ) lies on the right branch of the hyperbola x 2 − y 2 = 1 . Hence the names circular and hyperbolic functions.

2.11Logarithms of Complex Numbers

Video Lecture 5, Part 5: The Complex Logarithm (16:52)

If z = e ω , then we define the complex logarithm by ln ⁡ ( z ) = ω .

Logarithm of a Product

Let z 1 = e ω 1 and z 2 = e ω 2 . Then

z 1 z 2 = e ω 1 e ω 2 = e ω 1 + ω 2 ⇒ ln ⁡ ( z 1 z 2 ) = ω 1 + ω 2 = ln ⁡ ( z 1 ) + ln ⁡ ( z 2 ) . (2.148)

This holds for the multivalued logarithm, as an equality between sets of values. For the principal value it can fail: Ln ( − 1 ) + Ln ( − 1 ) = i π + i π = 2 π i , while Ln ( ( − 1 ) ( − 1 ) ) = Ln ( 1 ) = 0 .

DefinitionComplex Logarithm

For a complex number z = r e i θ with r = | z | > 0 ,

ln ⁡ ( z ) = ln ⁡ ( r e i θ ) = ln ⁡ ( r ) + ln ⁡ ( e i θ ) = ln ⁡ ( r ) + i θ . (2.149)

Because the angle θ is not unique (adding 2 π leaves the same point), the logarithm is a multivalued function:

ln ⁡ ( z ) = ln ⁡ ( r ) + i ( θ + 2 π k ) , k ∈ ℤ . (2.150)
RemarkPrincipal Value

We define the principal branch of the logarithm by restricting the argument:

Ln ( z ) = ln ⁡ ( r ) + i Θ , where  − π < Θ ≤ π . (2.151)

Here Θ = Arg ( z ) is the principal argument, defined at the start of this chapter as the unique value of arg ⁡ ( z ) lying in ( − π , π ] .

Example
  1. Find all values of ln ⁡ ( − 1 ) , and then its principal value.

    Solution.

    These are two different questions, and they have two different answers. Take them in that order.

    All values. Write − 1 in polar form. Its modulus is 1 and its argument is π , but any odd multiple of π names the same point:

    − 1 = e i ( 2 n + 1 ) π , n ∈ ℤ . (2.152)

    Feed this into ln ⁡ ( r e i θ ) = ln ⁡ ( r ) + i θ and use ln ⁡ ( 1 ) = 0 . All logarithms of − 1 are therefore

      ln ⁡ ( − 1 ) = ln ⁡ ( 1 ) + i ( 2 n + 1 ) π = i ( 2 n + 1 ) π   , n ∈ ℤ , (2.153)

    that is

    ln ⁡ ( − 1 ) = … ,   − i 3 π ,   − i π ,   i π ,   i 3 π ,   … (2.154)

    The principal value. If instead we ask specifically for the principal logarithm, we are no longer free to choose n : the principal branch demands that the argument lie in ( − π , π ] , and of all the odd multiples of π exactly one does, namely the case n = 0 ,

    Arg ( − 1 ) = π ∈ ( − π , π ] . (2.155)

    Hence

      Ln ( − 1 ) = ln ⁡ ( 1 ) + i Arg ( − 1 ) = i π   . (2.156)

    The lower-case answer is an infinite list; the upper-case answer is one number picked out of that list by a convention. Note also that ln ⁡ ( − 1 ) has no real values at all: the logarithm of a negative number does not exist in ℝ , but it exists, infinitely many times over, in ℂ . In the plane the values sit on the imaginary axis, 2 π apart, like the rungs of a ladder (Figure 2.11).

    The values i(2n+1) of (-1) on the imaginary axis; the principal value Ln(-1) = i is shown in red.
    Figure 2.11. The values i ( 2 n + 1 ) π of ln ⁡ ( − 1 ) on the imaginary axis; the principal value Ln ( − 1 ) = i π is shown in red.
  2. Find all values of ln ⁡ ( 1 + i ) .

    Solution.

    Here | 1 + i | = 2 and arg ⁡ ( 1 + i ) = π / 4 , so

    ln ⁡ ( 1 + i ) = ln ⁡ 2 + i ( π 4 + 2 n π ) , n ∈ ℤ , (2.157)

    with principal value Ln ( 1 + i ) = 1 2 ln ⁡ ( 2 ) + i π / 4 .

Remark 2.11.2All values, and the one we call the answer

The same two-step pattern has now appeared twice — once for a root in Section 2.8 and once for a logarithm — and it is worth seeing the two side by side, because the rule behind them is identical. In each case the several answers come from the argument being fixed only up to whole turns, and the single answer comes from insisting that the argument be the principal one, Arg ( z ) ∈ ( − π , π ] .

Take 8 1 / 3 , the least alarming example there is. Writing 8 = 8 e i 2 π k and taking the cube root divides the angle by 3 :

8 1 / 3 = 8 1 / 3 e i 2 π k / 3 = 2 e i 2 π k / 3 , k = 0 , 1 , 2 , (2.158)

which written out are

2 , 2 e i 2 π / 3 = − 1 + i 3 , 2 e i 4 π / 3 = − 1 − i 3 . (2.159)

The principal one is k = 0 , whose argument 0 lies in ( − π , π ] — it is the familiar real cube root 2 that a calculator returns without mentioning the other two. So the consistent picture is:

All valuesPrincipal value
8 1 / 3 2 ,   − 1 + i 3 ,   − 1 − i 3 2
ln ⁡ ( − 1 ) i ( 2 n + 1 ) π , n ∈ ℤ i π

The two rows differ in how many values there are. A cube root has exactly three, because dividing the angle by 3 makes k and k + 3 give the same point; a logarithm has infinitely many, because adding 2 π i to the answer never repeats. They do not differ in what selects the principal value. The moral is a habit worth forming now: when a question has several answers in ℂ , decide first which of these two questions you are being asked, and then say so in the notation — lower case for the whole family, upper case for the one representative.

2.12Complex Roots and Powers

Video Lecture 5, Part 6: Complex Powers (20:30)

Consider ln ⁡ ( a b ) = b ln ⁡ ( a ) ⇒ a b = e b ln ⁡ ( a ) .

Example
  1. Find all the values of i − 2 i :

    i − 2 i = e − 2 i ln ⁡ ( i ) = e − 2 i [ ln ⁡ ( 1 ) + ( π 2 + 2 π n ) i ] = e 2 ( π 2 + 2 π n ) = e π + 4 π n , n ∈ ℤ . (2.160)

    Every value is a positive real number; the principal value ( n = 0 ) is e π ≈ 23.14 . The reason is geometric: the values of ln ⁡ ( i ) lie on the imaginary axis, and multiplying by − 2 i turns them by − 90 ∘ and stretches them by 2 , onto the real axis (Figure 2.12).

    Three values of (i) (blue) and their images under multiplication by -2i (red): -3, , 5, all real. So every value of i^-2i = e^-2i(i) is real.
    Figure 2.12. Three values of ln ⁡ ( i ) (blue) and their images under multiplication by − 2 i (red): − 3 π , π , 5 π , all real. So every value of i − 2 i = e − 2 i ln ⁡ ( i ) is real.
  2. i 1 / 2 :

    i 1 / 2 = e ( 1 / 2 ) ln ⁡ ( i ) = e 1 2 [ ln ⁡ ( 1 ) + ( π 2 + 2 π n ) i ] = e i ( π 4 + π n ) = e i π / 4 e i π n = ± 1 + i 2 . (2.161)

2.13Inverse trigonometric functions

Video Lecture 5, Part 7: Inverse Functions (23:08)

Let

w = cos ⁡ ( z ) = e i z + e − i z 2 ⇒ z = cos − 1 ⁡ ( w ) . (2.162)
Example

Solve z = cos − 1 ⁡ ( 2 ) .

Solution.

For a real variable this has no solution, since cos ⁡ ( x ) never exceeds 1 . Over ℂ it does. Start from the definition of cos ⁡ ( z ) and put U = e i z :

e i z + e − i z 2 = 2 ⟹ U + U − 1 = 4 . (2.163)

Multiplying through by U turns this into a quadratic:

U 2 − 4 U + 1 = 0 ⟹ U = 4 ± 16 − 4 2 = 4 ± 2 3 2 = 2 ± 3 . (2.164)

So e i z = 2 ± 3 , a positive real number. Taking the complex logarithm, and remembering it is multivalued,

i z = ln ( 2 ± 3 ) + 2 n π i , n ∈ ℤ . (2.165)

Dividing by i (that is, multiplying by − i ),

z = 2 n π − i ln ( 2 ± 3 ) . (2.166)

Finally, note that ( 2 + 3 ) ( 2 − 3 ) = 4 − 3 = 1 , so the two numbers are reciprocals and ln ⁡ ( 2 − 3 ) = − ln ⁡ ( 2 + 3 ) . The two sign choices therefore collapse into a single ± :

z = 2 n π ± i ln ( 2 + 3 ) , n ∈ ℤ . (2.167)

The answer is genuinely complex, with no real part other than 2 n π : asking for an angle whose cosine is 2 forces us off the real axis.

Check: cos ⁡ ( i ln ⁡ ( 2 + 3 ) ) = cosh ⁡ ( ln ⁡ ( 2 + 3 ) ) = ( 2 + 3 ) + ( 2 − 3 ) 2 = 2 . In pictures: along the imaginary axis cos ⁡ ( i y ) = cosh ⁡ ( y ) , which reaches 2 at y = ± ln ⁡ ( 2 + 3 ) ≈ ± 1.317 , and every shift by 2 π gives another pair of solutions (Figure 2.13).

Left: (y) reaches the level 2 at y = (2+3) 1.317. Right: the solutions z = 2n i(2+3) of (z) = 2 in the complex plane, in pairs above and below the real axis, 2 apart.
Figure 2.13. Left: cosh ⁡ ( y ) reaches the level 2 at y = ± ln ⁡ ( 2 + 3 ) ≈ ± 1.317 . Right: the solutions z = 2 n π ± i ln ⁡ ( 2 + 3 ) of cos ⁡ ( z ) = 2 in the complex plane, in pairs above and below the real axis, 2 π apart.

Inverse hyperbolic functions can be written as logarithms in the same way. This explains a formula you may have met in a table of integrals without justification:

∫ d x x 2 + a 2 = sinh − 1 ( x a ) = ln ( x + x 2 + a 2 ) + C . (2.168)
ExampleWhere the integral-table logarithm comes from

Show that sinh − 1 ⁡ ( x / a ) = ln ⁡ ( x + x 2 + a 2 ) − ln ⁡ ( a ) .

Solution.

Let z = sinh − 1 ( x a ) . Then

x a = sinh ⁡ ( z ) = e z − e − z 2 . (2.169)
2 ( x a ) = U − 1 U ( U = e z ) . (2.170)
U 2 − 2 ( x a ) U − 1 = 0. (2.171)
a U 2 − 2 x U − a = 0. (2.172)
⇒ U = 2 x ± 4 x 2 − 4 ( a ) ( − a ) 2 a = 2 x ± 4 x 2 + 4 a 2 2 a = x ± x 2 + a 2 a . (2.173)
e z = x ± x 2 + a 2 a ⇒ z = ln ( x ± x 2 + a 2 a ) (for real integrals) . (2.174)

Only the + sign is admissible: with the − sign the bracket is negative, and a real logarithm would not exist. Absorbing the constant − ln ⁡ ( a ) into the constant of integration gives the table entry:

z = ln ( x + x 2 + a 2 ) − ln ⁡ ( a ) . (2.175)

2.14Applications of Complex Numbers

Complex numbers were introduced in this chapter to make − 1 meaningful, which sounds like a formal convenience. Their real value in physics is different and larger: oscillation and rotation are the same thing, and e i θ is the object that says so. Every application below is a consequence of that single fact.

2.14.1Oscillation Is Rotation: Phasors

Video Lecture 6, Part 1: Oscillation Is Rotation (0:00)

Euler's formula, e i ω t = cos ⁡ ( ω t ) + i sin ⁡ ( ω t ) , has a simple geometric meaning. As the time t increases, the complex number z ( t ) = e i ω t keeps the length | z | = 1 while its angle ω t grows steadily, so z moves around the unit circle at the constant angular speed ω . Its real part, the shadow of the moving point on the real axis, is cos ⁡ ( ω t ) . In other words, an oscillation is the shadow of a rotation (Figure 2.14).

This observation turns every oscillation into a complex number. An oscillation of amplitude A and phase ϕ can be written as

A cos ⁡ ( ω t + ϕ ) = Re [ A e i ( ω t + ϕ ) ] = Re [ ( A e i ϕ ) e i ω t ] . (2.176)

The factor e i ω t is the same for every oscillation of frequency ω ; everything that distinguishes one oscillation from another is contained in the complex number A e i ϕ , called the phasor of the oscillation. Its modulus is the amplitude and its argument is the phase. For example, the oscillation 2 cos ⁡ ( ω t + π 3 ) has the phasor 2 e i π / 3 = 1 + i 3 .

Phasors also simplify calculus. Differentiating the rotating arrow gives

d d t e i ω t = i ω e i ω t , (2.177)

so the derivative is obtained by multiplying by i ω . Since i = e i π / 2 , multiplying by i turns an arrow by 90 ∘ without changing its length. This is why the derivative of cos ⁡ ( ω t ) , namely − ω sin ⁡ ( ω t ) = ω cos ⁡ ( ω t + π 2 ) , is the same oscillation shifted forward by a quarter period. Differentiation becomes multiplication, and differential equations for oscillations become algebraic equations for phasors. The rest of this section uses this one idea in different settings.

Figure 2.14. The phasor z = e i ω t (blue) turns at the angular frequency ω , and its shadow on the real axis traces the oscillation cos ⁡ ( ω t ) . In print, four positions of the arrow are numbered, and the same four moments are marked on the graph of cos ⁡ ( ω t ) .

2.14.2Adding Waves

Video Lecture 6, Part 2: Adding Waves (3:08)

Two oscillations of the same frequency add in the simplest possible way: their phasors add. The first example below shows this in general, and the second works it out for a concrete pair of waves.

ExamplePhysics: adding two waves of the same frequency

Two sources produce oscillations of the same frequency ω but different amplitudes and phases,

y 1 ( t ) = A 1 cos ⁡ ( ω t + ϕ 1 ) , y 2 ( t ) = A 2 cos ⁡ ( ω t + ϕ 2 ) . (2.178)

Find their sum. This is the central calculation in interference and in AC circuit analysis.

Solution.

Done with trigonometric identities this is tedious. The complex method is to notice that, by Euler's formula, A cos ⁡ ( ω t + ϕ ) is the real part of A e i ( ω t + ϕ ) . So write each wave as the real part of a complex exponential and add:

y 1 + y 2 = Re [ A 1 e i ( ω t + ϕ 1 ) + A 2 e i ( ω t + ϕ 2 ) ] = Re [ ( A 1 e i ϕ 1 + A 2 e i ϕ 2 ) e i ω t ] . (2.179)

The time dependence e i ω t factors out completely, and everything specific to the two waves is collected in the bracket

A 1 e i ϕ 1 + A 2 e i ϕ 2 ⏟ a single complex number = A e i ϕ , (2.180)

called the phasor of the combined wave. Adding two oscillations has been reduced to adding two complex numbers — that is, to placing two arrows tip to tail in the complex plane. Reading off the modulus and argument of the sum,

A = | A 1 e i ϕ 1 + A 2 e i ϕ 2 | , ϕ = arg ( A 1 e i ϕ 1 + A 2 e i ϕ 2 ) , (2.181)

and the answer is y 1 + y 2 = A cos ⁡ ( ω t + ϕ ) : the sum is another oscillation at the same frequency, with a new amplitude and phase.

Two special cases are worth noting. If ϕ 1 = ϕ 2 the phasors are parallel and A = A 1 + A 2 — constructive interference. If ϕ 2 = ϕ 1 + π they are antiparallel and A = | A 1 − A 2 | — destructive interference, which vanishes entirely when the amplitudes are equal.

ExampleAdding 3 cos ⁡ ( ω t ) and 4 sin ⁡ ( ω t )

Write 3 cos ⁡ ( ω t ) + 4 sin ⁡ ( ω t ) as a single wave A cos ⁡ ( ω t + ϕ ) .

Solution.

To use phasors, both terms must first be written as cosines. A sine is a cosine delayed by a quarter of a period, sin ⁡ ( ω t ) = cos ⁡ ( ω t − π 2 ) , so the phasor of the first term is 3 and the phasor of the second term is 4 e − i π / 2 = − 4 i . The phasor of the sum is the sum of the phasors,

3 − 4 i . (2.182)

Its modulus is the amplitude of the combined wave, and its argument is the phase:

A = 3 2 + 4 2 = 5 , ϕ = − tan − 1 ⁡ ( 4 3 ) = − 53.13 ∘ . (2.183)

Therefore

3 cos ⁡ ( ω t ) + 4 sin ⁡ ( ω t ) = 5 cos ⁡ ( ω t − 53.13 ∘ ) . (2.184)

The amplitude of the sum is 5 , not 3 + 4 = 7 , because the two waves are a quarter of a period out of step and never reach their maxima at the same time. In general, the amplitude of the sum of two waves follows from the law of cosines applied to the two phasors,

A 2 = A 1 2 + A 2 2 + 2 A 1 A 2 cos ⁡ ( ϕ 2 − ϕ 1 ) , (2.185)

which gives A = A 1 + A 2 for waves in phase and A = | A 1 − A 2 | for waves half a period apart, the two special cases of the previous example; every other phase difference gives an amplitude between these two extremes. One can check the result by expanding the right-hand side: 5 cos ⁡ ( ω t − 53.13 ∘ ) = 5 cos ⁡ ( 53.13 ∘ ) cos ⁡ ( ω t ) + 5 sin ⁡ ( 53.13 ∘ ) sin ⁡ ( ω t ) , and 5 cos ⁡ ( 53.13 ∘ ) = 3 , 5 sin ⁡ ( 53.13 ∘ ) = 4 . Figure 2.15 shows the calculation.

Figure 2.15. The phasors 3 (blue) and − 4 i (red), placed tip to tail, add to 3 − 4 i (yellow), of length 5 and angle − φ with tan ⁡ φ = 4 3 . The corresponding oscillations 3 cos ⁡ ( ω t ) and 4 sin ⁡ ( ω t ) add to 5 cos ⁡ ( ω t − φ ) , φ = 53.13 ∘ .

2.14.3AC Circuits: Impedance and Resonance

Video Lecture 6, Part 3: AC Circuits (6:21)

In an alternating-current circuit, the source voltage oscillates as V ( t ) = V 0 cos ⁡ ( ω t ) = Re [ V 0 e i ω t ] , and after a short time every current in the circuit oscillates at the same frequency, I ( t ) = Re [ I 0 e i ω t ] , where the complex amplitude I 0 records both the size of the current and its phase. Each circuit element relates its voltage to its current, and with the factor e i ω t each relation becomes a multiplication:

resistor: V = R I , Z R = R , inductor: V = L d I d t = i ω L I , Z L = i ω L , capacitor: I = C d V d t = i ω C V , Z C = 1 i ω C = − i ω C . (2.186)

The complex number Z , the impedance, plays the role of resistance in Ohm's law V = Z I . The factor i has a physical meaning: multiplying by i = e i π / 2 turns a phasor by a quarter turn. So the voltage across a resistor is in phase with the current, the voltage across an inductor leads the current by a quarter period, and the voltage across a capacitor lags behind it by a quarter period.

For a resistor, an inductor and a capacitor in series, the impedances add:

Z = R + i ( ω L − 1 ω C ) , | I 0 | = V 0 | Z | = V 0 R 2 + ( ω L − 1 ω C ) 2 . (2.189)

Since I 0 = V 0 / Z , the current lags behind the source voltage by the phase angle

arg ⁡ Z = arctan ⁡ ( ω L − 1 / ( ω C ) R ) , (2.190)

which is negative (the current leads) when the capacitor dominates. A differential equation for the current has become arithmetic with complex numbers.

The current is largest when the imaginary part of Z vanishes, that is, when ω L = 1 / ( ω C ) . This happens at the resonance frequency

ω 0 = 1 L C , (2.191)

where | Z | = R . For example, with R = 20   Ω , L = 0.1 H, C = 10   μ F and V 0 = 10 V, the resonance frequency is ω 0 = 1000   s − 1 and the current amplitude there is | I 0 | = 10 / 20 = 0.5 A. At half that frequency, ω = 500   s − 1 , we have ω L = 50   Ω and 1 / ( ω C ) = 200   Ω , so | Z | = 20 2 + 150 2 ≈ 151.3   Ω and | I 0 | ≈ 0.066 A, less than a seventh of its value at resonance. The current is large only near ω 0 (Figure 2.16). This is how a radio selects one station: turning the tuning knob changes C , which moves ω 0 onto the frequency of the desired station, and only that station drives an appreciable current.

Figure 2.16. The series circuit of a source V , a resistor R , an inductor L and a capacitor C , and its resonance curve | I 0 | ( ω ) = V 0 / | Z ( ω ) | for R = 20   Ω , L = 0.1 H, C = 10   μ F and V 0 = 10 V. The current peaks sharply at ω 0 = 1 / L C = 1000   s − 1 , where | I 0 | = 0.5 A; at ω = 500   s − 1 it is only 0.066 A.

2.14.4The Damped Oscillator

Video Lecture 6, Part 4: The Damped Oscillator (9:57)

When friction is present, the exponent in e λ t is no longer purely imaginary. A complex λ then describes two things at once, oscillation and decay, as the following example shows.

ExamplePhysics: the damped oscillator

A mass m on a spring of stiffness k , slowed by a friction force proportional to its velocity, obeys Newton's second law m x ¨ = − k x − b x ˙ , where b is the friction coefficient. Dividing by m , we write the equation as

x ¨ + 2 γ x ˙ + ω 0 2 x = 0 , γ = b 2 m , ω 0 2 = k m . (2.192)

Here ω 0 is the angular frequency the oscillator would have without friction, and γ measures the strength of the friction. Find the motion.

Solution.

We look for a solution of the form x = e λ t , because differentiating an exponential only multiplies it by a constant: x ˙ = λ e λ t and x ¨ = λ 2 e λ t . Substituting into the equation gives

( λ 2 + 2 γ λ + ω 0 2 ) e λ t = 0 . (2.193)

Since e λ t is never zero, λ must satisfy the quadratic equation λ 2 + 2 γ λ + ω 0 2 = 0 , whose roots are

λ = − γ ± γ 2 − ω 0 2 . (2.194)

When the friction is weak, γ < ω 0 , the number under the square root is negative and the roots are complex:

λ = − γ ± i ω d , ω d = ω 0 2 − γ 2 . (2.195)

The physical motion is the real part of A e λ t . Since e λ t = e − γ t e i ω d t , Euler's formula gives

x ( t ) = A e − γ t cos ⁡ ( ω d t ) . (2.196)

The two parts of the complex number λ play separate roles: the real part − γ makes the amplitude decay, and the imaginary part ω d sets the frequency of the oscillation.

For a numerical example, take m = 1 kg, k = 16 N/m and b = 0.5 kg/s. Then ω 0 = 4   s − 1 , γ = 0.25   s − 1 and ω d = 3.992   s − 1 , so the friction hardly changes the frequency. It does, however, steadily reduce the amplitude: after 9 s the amplitude is e − 0.25 ⋅ 9 ≈ 0.105 , about a tenth of its initial value (Figure 2.17).

The complex solution has a simple picture. In the complex plane, e λ t is an arrow that rotates at the angular frequency ω d while its length shrinks as e − γ t , so its tip moves along a spiral toward the origin. The motion x ( t ) is the projection of this arrow onto the real axis. If the friction is strong, γ ≥ ω 0 , the roots λ are real and negative, and the mass returns to equilibrium without oscillating; this is called critical damping when γ = ω 0 and overdamping when γ > ω 0 . Figure 2.17 shows the spiral and, beside it, the damped oscillation it projects.

Figure 2.17. The complex number z = e λ t , with λ = − γ + i ω d . In the complex plane the arrow turns at ω d while it shrinks as e − γ t , so its tip spirals in toward the origin. Its real part is the damped oscillation x = e − γ t cos ⁡ ( ω d t ) , which stays inside the envelope ± e − γ t (dashed). Here γ = 0.25   s − 1 and ω d = 3.992   s − 1 .

2.14.5Roots of Unity at Work: Three-Phase Power

Video Lecture 6, Part 5: Roots of Unity at Work (13:32)

The n th roots of unity, the solutions of z n = 1 , are z k = e 2 π i k / n for k = 0 , 1 , … , n − 1 . They lie at the corners of a regular n -sided polygon inscribed in the unit circle, and they have a remarkable property: their sum is zero. To see why, write them as powers of w = e 2 π i / n and sum the geometric series:

1 + w + w 2 + ⋯ + w n − 1 = w n − 1 w − 1 = 0 , (2.197)

because w n = 1 while w ≠ 1 . Geometrically, the n arrows from the origin to the corners of a regular polygon balance one another. For n = 3 , for instance, 1 + ( − 1 2 + 3 2 i ) + ( − 1 2 − 3 2 i ) = 0 .

This fact is used every day in the electric power grid. A three-phase supply provides three voltages of equal amplitude, each delayed by a third of a period with respect to the previous one:

V k ( t ) = V 0 cos ⁡ ( ω t − 2 π k 3 ) = Re [ V 0 w k e i ω t ] , k = 0 , 1 , 2 , w = e − 2 π i / 3 . (2.198)

Their phasors are V 0 , V 0 w and V 0 w 2 : three arrows 120 ∘ apart. Their sum is therefore

V 0 e i ω t ( 1 + w + w 2 ) = 0 (2.199)

at every instant (Figure 2.18). When the three phases drive equal loads, the three currents are also equal phasors 120 ∘ apart, and they too add to zero. The current that would have to return along a shared wire is then zero, which is why a balanced three-phase system can carry power with very little return current.

Figure 2.18. Three-phase power. The voltage phasors V 0 , V 0 w and V 0 w 2 (blue, red, green) are 120 ∘ apart; placed tip to tail they close a triangle, so their sum is zero. The three voltages, plotted against time, therefore add to zero at every moment.

2.14.6The Free Particle in Quantum Mechanics

Video Lecture 6, Part 6: The Quantum Corkscrew (16:02)

In quantum mechanics a particle is described by a wavefunction, and for a free particle moving along the x axis the wavefunction is

ψ ( x , t ) = e i ( k x − ω t ) = cos ⁡ ( k x − ω t ) + i sin ⁡ ( k x − ω t ) . (2.200)

Here complex numbers are not a convenience: the wavefunction itself is complex. At each point x , ψ is a complex number, which we may draw as an arrow on a small dial (Figure 2.19).

The meaning of the two parts of the exponent becomes clear if we split it as ψ = e i k x e − i ω t . At a fixed point x , the factor e − i ω t is a phasor turning at the rate ω , so every arrow simply turns as time runs. At a fixed time, the factor e i k x turns once every wavelength λ = 2 π / k along x , so neighbouring dials point in different directions. Together, the pattern slides along x : the crests of the real and imaginary parts move at the speed v = ω / k . The size of ψ , on the other hand, is the same everywhere, | ψ | 2 = 1 , so this particle is equally likely to be found anywhere.

The relation between ω and k follows from the Schrödinger equation for a free particle,

i ℏ ∂ ψ ∂ t = − ℏ 2 2 m ∂ 2 ψ ∂ x 2 . (2.201)

Each derivative of the exponential only multiplies it by a constant: ∂ ψ / ∂ t = − i ω ψ , so the left-hand side is i ℏ ( − i ω ) ψ = ℏ ω ψ , and ∂ 2 ψ / ∂ x 2 = ( i k ) 2 ψ = − k 2 ψ , so the right-hand side is ℏ 2 k 2 2 m ψ . Both sides are a number times ψ , and ψ is a solution exactly when

ℏ ω = ℏ 2 k 2 2 m . (2.202)

With the energy E = ℏ ω and the momentum p = ℏ k , this is E = p 2 / 2 m , the kinetic energy of a free particle.

Figure 2.19. The free-particle wavefunction ψ = e i ( k x − ω t ) . Each dial is the complex plane at one point x , and its arrow is the value of ψ there. Below, the real part cos ⁡ ( k x − ω t ) (red) and the imaginary part sin ⁡ ( k x − ω t ) (blue). Neighbouring dials differ by one turn per wavelength λ = 2 π / k , and as time runs the waves move to the right at v = ω / k .
Remark

Notice what happened to the exponential and trigonometric functions in this chapter. Over ℝ they look unrelated: one grows, the others oscillate. Over ℂ they are the same function seen along different directions, tied together by e i z , and cosh and cos differ only by which axis you walk along. Extending to complex numbers did not add complication; it removed a distinction that was never really there.

Summary Table

QuantityFormulaNotes
Rectangular form z = x + i y Best for adding and subtracting.
Polar form z = r e i θ = r ( cos ⁡ ( θ ) + i sin ⁡ ( θ ) ) Best for multiplying, dividing, powers and roots.
Modulus | z | = x 2 + y 2 = z z ¯ | z 1 z 2 | = | z 1 | | z 2 | .
Conjugate z ¯ = x − i y = r e − i θ Reflection across the real axis.
Argument arg ⁡ z = Arg z + 2 n π Multivalued; Arg z ∈ ( − π , π ] .
Euler's formula e i θ = cos ⁡ ( θ ) + i sin ⁡ ( θ ) Ties oscillation to rotation.
Powers z n = r n e i n θ De Moivre when r = 1 .
Roots z 1 / n = r 1 / n e i ( θ + 2 π m ) / n n distinct roots, m = 0 , … , n − 1 .
Trigonometric cos ⁡ ( z ) = e i z + e − i z 2 , sin ⁡ ( z ) = e i z − e − i z 2 i Unbounded for complex z .
Hyperbolic cos ⁡ ( i y ) = cosh ⁡ ( y ) , sin ⁡ ( i y ) = i sinh ⁡ ( y ) Same function, rotated axis.
Logarithm ln ⁡ ( z ) = ln ⁡ | z | + i arg ⁡ z Infinitely many values.
Complex power a b = e b ln ⁡ ( a ) Generally multivalued.

For the Interested Reader

Complex numbers are usually met first in an algebra course, where they can seem like a trick for writing down roots that do not exist. If that is how they still feel, the sources below are worth an hour. Everything listed is free.

Algebra and geometry of complex numbers

Sections 2.1 and 2.2.

Euler's formula and where it leads

Sections 2.5 onward.

3Blue1Brown derives e i θ from a differential equation rather than from two power series, which is closer to how it is actually used in physics. Worth watching if Euler's formula still feels like an accident:

Video

e i π in 3.14 minutes, using dynamics — 3Blue1Brown

Video thumbnail for e^(iπ) in 3.14 minutes, using dynamics — 3Blue1BrownWatch on YouTube

Open video on YouTube

Remark

A caution, as in Chapter 1. Most of these are mathematics sources, concerned with what complex numbers are. In this course they are mainly a device: a way to turn differential equations into algebra and trigonometric identities into arithmetic. If a source spends its time on the algebraic completeness of ℂ , that is interesting but it is not what we will use.

Video lectures

Each section of this chapter links to the part of a lecture that teaches it. The full lectures are below, and all of them are in the course playlist.

Video Lecture 4: Complex Numbers, Complex Series and Euler's Formula (full lecture)

Video Lecture 5: Functions of a Complex Variable (full lecture)

Video Lecture 6: Complex Numbers at Work (full lecture)