Complex Numbers

July 22, 2026

A complex number is a 2D vector with an extra rule: it can multiply another point in the plane. That product turns multiplication into rotation and scaling. With conjugation, it also records the angle and signed area between two vectors.

This sounds like a small addition. It changes the kind of object we have. A 2D real vector space gives us addition and multiplication by real scalars. The complex numbers also let us multiply and divide elements of the plane.

Before adding complex multiplication, a 2D vector space already has a standard way to compare directions. The dot product of u = (a, b) and v = (c, d) is:

u \cdot v = a c + b d = | u | | v | \cos θ

It is positive when the vectors point in similar directions, zero when they are perpendicular, and negative when they point apart. Its value also grows with both vector lengths.

Cosine similarity removes those lengths. It keeps only directional alignment:

\frac{u \cdot v}{| u | | v |} = \cos θ

This gives a value from -1 to 1. Complex multiplication will add the part these measures omit: which side one vector lies on, and how much signed area the pair spans.

2D real vector space	Complex numbers
Elements look like (x, y)	Elements look like a + bi
Adds vectors component by component	Adds numbers component by component
Multiplies a vector by a real scalar	Multiplies two complex numbers
Has no required vector by vector product	Forms a field

The same plane

The match between them is exact if we only look at addition and real scaling:

a + b i ⟷ (a, b)

For example, 3 + 2i is the point (3, 2). Adding 1 + i moves one unit right and one unit up. As real vector spaces, ℂ and ℝ² are isomorphic. Each has two real directions.

One point, two readings. The difference appears when we ask what operations are allowed.

Complex multiplication rotates and scales

A vector space does not tell us what (1, 2)(3, 4) means. We can add these vectors or scale them, but a product needs a new definition.

Complex numbers come with that definition. Since i² = -1:

(a + b i) (c + d i) = (a c - b d) + (a d + b c) i

×	c	di
a	a × cac	a × diadi
bi	bi × cbci	bi × dibd i² = -bd

Blue cells contribute to the real part. Green cells contribute to the imaginary part.

Notice that ac - bd is not the dot product. The minus sign comes from multiplying the two imaginary terms:

(b i) (d i) = b d i^{2} = - b d

The dot product ac + bd appears only when we conjugate one input first. Conjugation changes bi to -bi, so the imaginary terms contribute +bd instead. The later comparison section uses exactly this sign change.

This rule works with addition and gives every nonzero number a reciprocal. It is not part of a plain vector space. We chose extra algebraic structure.

The structure has a clear geometric meaning. Write a nonzero complex number in polar form:

z = r (\cos θ + i \sin θ) = r e^{i θ}

The bridge between the trigonometric and exponential forms is Euler's formula:

e^{i θ} = \cos θ + i \sin θ

When |z| = 1, we have r = 1 and may write z = e^iθ. For a general complex number, the factor r is still needed. Setting θ = π gives Euler's identity:

e^{i π} + 1 = 0

Multiplication by z scales every length by r and rotates every angle by θ. If r = 1, it only rotates. Multiplying by i rotates 90 degrees counterclockwise.

Here w = z₁ and z = z₂. Normal multiplication produces wz = z₁z₂ by adding their angles.

Conjugation turns multiplication into comparison

Normal complex multiplication combines two rotations and scales. If z₁ has angle θ₁ and z₂ has angle θ₂, their product has angle θ₁ + θ₂. This is useful when one complex number acts as a transformation on another:

z_{1} z_{2} = | z_{1} | | z_{2} | e^{i (θ_{1} + θ_{2})}

Sometimes we do not want to combine the vectors. We want to compare them. Conjugating z₁ reverses its angle from θ₁ to -θ₁. Multiplication then subtracts the first angle from the second:

\overline{z_{1}} z_{2} = | z_{1} | | z_{2} | e^{i (θ_{2} - θ_{1})}

The result now describes the direction of z₂ relative to z₁. If both vectors rotate by the same angle, this result does not change. Use normal multiplication to apply or combine rotations. Use the conjugated product to measure relative angle, alignment, orientation, or signed area.

Let z₁ = a + bi and z₂ = c + di. Take the complex conjugate of z₁ first, then multiply it by z₂:

\overline{z_{1}} z_{2} = (a - b i) (c + d i) = (a c + b d) + (a d - b c) i

This is different from the ordinary product z₁z₂, whose real part is ac - bd. A complex conjugate flips the sign of the imaginary part. It is not the multiplicative inverse. The inverse also divides by the squared modulus:

{z_{1}}^{- 1} = \frac{\overline{z_{1}}}{| z_{1} |^{2}}

The real part is the dot product. It measures alignment:

a c + b d = | z_{1} | | z_{2} | \cos Δ θ

The imaginary part is the determinant. It measures signed area and orientation:

a d - b c = | z_{1} | | z_{2} | \sin Δ θ

Here Δθ = θ₂ - θ₁. A positive imaginary part means z₂ lies counterclockwise from z₁, using this order. Swapping the inputs flips the sign.

The dot product becomes the horizontal coordinate of z₃. The signed area becomes its vertical coordinate. Together they form z₃ = z₁z₂.

Encoded, not conserved

It is tempting to say that the conjugate product conserves angle and area. It does not preserve them through a change. It encodes the relative angle and combined scale of the original pair.

If z₁ = r₁e^iθ₁ and z₂ = r₂e^iθ₂, then Euler's formula gives:

\overline{z_{1}} z_{2} = r_{1} r_{2} e^{i (θ_{2} - θ_{1})}

Its magnitude is r₁r₂. Its argument is the signed angle from z₁ to z₂. Its real and imaginary parts split that same information into alignment and area.

The short version: ℝ² and ℂ are the same real vector space, but not the same algebraic object. Complex multiplication turns a point into a rotate-and-scale operation. With conjugation, it also packs the dot product and signed area into one number.

Outlook: why this trick stops at the plane

One complex number has exactly two real coordinates. This is why its geometry fits the 2D plane so well. The rule i² = -1 supplies one fixed quarter-turn, and complex multiplication builds every rotation and scaling from it.

Sadly, there is no matching number system for every ℝ^N that keeps all the useful properties of ℂ. We cannot give an arbitrary real vector space a multiplication that is commutative, associative, distributive, and divisible by every nonzero element.

Even-dimensional spaces still have a partial extension. After choosing a complex structure, ℝ²ⁿ can be viewed as ℂⁿ:

R^{2 n} ≅ C^{n}

This choice pairs real directions and defines an operator J with J² = -I. It lets complex scalars act on the space. It does not turn the whole space into one large complex number field. In particular, odd-dimensional real spaces cannot carry such a complex structure because their directions cannot all be paired.

Dimension four offers another clue. The quaternions give ℝ⁴ a useful multiplication and division, but multiplication is no longer commutative. The Frobenius theorem makes the limit precise: ℝ and ℂ are the only finite-dimensional associative, commutative real division algebras. To move beyond the complex plane, some familiar algebraic rule has to go.

Glossary

Vector space: A set whose elements can be added together and multiplied by scalars while following the vector space rules.
Scalar: A number used to scale a vector. The scalars in a 2D real vector space are real numbers.
Field: A number system with addition, subtraction, multiplication, and division by every nonzero element. Both ℝ and ℂ are fields.
Complex conjugate: The reflection of a complex number across the real axis. The conjugate of z = a + bi is z = a - bi.
Modulus: The length of a complex number viewed as a vector: |z| = √(a² + b²).
Argument: The signed angle of a nonzero complex number from the positive real axis, often written arg(z).
Dot product: A length-weighted measure of alignment. For u = (a, b) and v = (c, d), it equals ac + bd.
Cosine similarity: The dot product divided by both vector lengths. It measures directional alignment from -1 to 1.
Determinant: For two 2D vectors, the signed area of their parallelogram. Its sign records their orientation.
Polar form: A complex number written through its length and angle: z = re^iθ.
Conjugated product: The product z₁z₂. Its real part is the dot product, and its imaginary part is the signed area.

Sean's Blog