Skip to content

Chapter 3 — Polynomials and Rational Functions

3.3 — Complex Numbers

A number system is not discovered complete; it is built, axiom by axiom, until it answers the question the previous one could not.


§3.2 closed with an unresolved question: what are the roots that x2+1x^2+1 lacks in R\mathbb{R}? §1.3 had already pointed out the same phenomenon from another angle, by declaring a2+b2a^2+b^2 irreducible over R\mathbb{R}, and §1.2 had explained it from its deepest cause: no real number squared produces a negative result. Three different sections, three appearances of the same obstacle. Resolving it does not consist in forcing an exception inside R\mathbb{R}, but in building, with the same axiomatic rigor that erected R\mathbb{R} in §1.1, a strictly larger number system in which the obstacle disappears.

§1. Construction: ordered pairs over ℝ × ℝ

Section titled “§1. Construction: ordered pairs over ℝ × ℝ”

The construction starts from a set we already know, R×R\mathbb{R}\times\mathbb{R}, the ordered pairs of real numbers, and equips it with two new operations, chosen not arbitrarily but to achieve a precise goal, which will be revealed in §3.

Definition. The set C\mathbb{C} of complex numbers is R×R\mathbb{R}\times\mathbb{R} equipped with the operations

(a,b)+(c,d):=(a+c, b+d)(a,b) + (c,d) := (a+c,\ b+d) (a,b)⋅(c,d):=(ac−bd, ad+bc)(a,b)\cdot(c,d) := (ac-bd,\ ad+bc)

for all (a,b),(c,d)∈R×R(a,b),(c,d)\in\mathbb{R}\times\mathbb{R}.

The sum is coordinatewise addition, so natural that it requires no comment. The product, on the other hand, is not the coordinatewise product that a first instinct would suggest, (ac,bd)(ac,bd), but this more elaborate combination. The reason for that choice is deliberately postponed until §3: for now, the definition is taken as given, and the rest of this section verifies that, with it, C\mathbb{C} behaves with all the algebraic regularity that R\mathbb{R} already exhibited.


§1.1 stated, for R\mathbb{R}, four laws (commutative, associative, distributive, and the existence of neutral elements and inverses) without giving them, at the time, a collective name. That name is due now: a set equipped with two operations satisfying those four laws is called a field. Under this definition R\mathbb{R} is a field; the purpose of this section is to prove that C\mathbb{C} is one too.

Definition. A field is a set FF with two operations +,⋅+,\cdot such that, for all a,b,c∈Fa,b,c\in F:

  1. Closure: a+b∈Fa+b\in F and a⋅b∈Fa\cdot b\in F.
  2. Commutativity: a+b=b+aa+b=b+a and a⋅b=b⋅aa\cdot b=b\cdot a.
  3. Associativity: (a+b)+c=a+(b+c)(a+b)+c=a+(b+c) and (a⋅b)⋅c=a⋅(b⋅c)(a\cdot b)\cdot c=a\cdot(b\cdot c).
  4. Distributivity: a⋅(b+c)=a⋅b+a⋅ca\cdot(b+c)=a\cdot b+a\cdot c.
  5. Neutral elements: there exist 0,1∈F0,1\in F, with 0≠10\neq1, such that a+0=aa+0=a and a⋅1=aa\cdot1=a.
  6. Inverses: for every a∈Fa\in F there exists −a∈F-a\in F with a+(−a)=0a+(-a)=0; for every a∈Fa\in F with a≠0a\neq0 there exists a−1∈Fa^{-1}\in F with a⋅a−1=1a\cdot a^{-1}=1.

Theorem. C\mathbb{C}, with the operations of §1, is a field.

Proof. Closure is immediate: both operations, applied to pairs of reals, return pairs of reals, because addition, subtraction and multiplication are closed operations in R\mathbb{R}, an axiom already established in §1.1.

The commutativity of ++ and the associativity of ++ are inherited directly, coordinate by coordinate, from the same properties in R\mathbb{R}. For the product, commutativity is verified by direct computation:

(a,b)(c,d)=(ac−bd, ad+bc)=(ca−db, da+cb)=(c,d)(a,b)(a,b)(c,d) = (ac-bd,\ ad+bc) = (ca-db,\ da+cb) = (c,d)(a,b)

where the middle step uses only the commutativity of the product and the sum in R\mathbb{R}.

The neutral elements are 0C=(0,0)0_{\mathbb{C}}=(0,0) and 1C=(1,0)1_{\mathbb{C}}=(1,0): indeed, (a,b)+(0,0)=(a,b)(a,b)+(0,0)=(a,b), and (a,b)(1,0)=(a⋅1−b⋅0, a⋅0+b⋅1)=(a,b)(a,b)(1,0)=(a\cdot1-b\cdot0,\ a\cdot0+b\cdot1)=(a,b). The additive inverse of (a,b)(a,b) is (−a,−b)(-a,-b), by immediate coordinatewise verification.

What remains is the multiplicative inverse, the only nontrivial case. Let (a,b)≠(0,0)(a,b)\neq(0,0), so that a2+b2>0a^2+b^2>0, the sum of a nonzero square and a nonnegative one. Let us propose

(a,b)−1:=(aa2+b2, −ba2+b2)(a,b)^{-1} := \left(\frac{a}{a^2+b^2},\ \frac{-b}{a^2+b^2}\right)

which is well defined precisely because a2+b2≠0a^2+b^2\neq0. Verifying:

(a,b)⋅(aa2+b2,−ba2+b2)=(a2a2+b2−−b2a2+b2, …)=(a2+b2a2+b2, 0)=(1,0)(a,b)\cdot\left(\frac{a}{a^2+b^2},\frac{-b}{a^2+b^2}\right) = \left(\frac{a^2}{a^2+b^2}-\frac{-b^2}{a^2+b^2},\ \ldots\right) = \left(\frac{a^2+b^2}{a^2+b^2},\ 0\right) = (1,0)

where the second coordinate vanishes by a symmetric computation (−aba2+b2+aba2+b2=0\frac{-ab}{a^2+b^2}+\frac{ab}{a^2+b^2}=0) that is omitted because it is analogous.

The associativity of the product and the distributivity are verified by the same kind of direct computation, expanding both sides of the equality in terms of the four coordinates involved and checking that they coincide term by term, via the properties already established for R\mathbb{R}; they are omitted here because they are mechanical and contribute no new idea beyond those already used in verifying commutativity. ■\blacksquare

It is this theorem, and not a notational convention, that legitimizes treating C\mathbb{C} with the same algebraic rules (solving for an unknown, distributing, canceling) that already govern R\mathbb{R}: both are fields, and every consequence of the field axioms proved abstractly, such as the properties of signs in §1.1, holds, without need of a new proof, in any field, C\mathbb{C} included.


§3. The imaginary unit and the notation a+bi

Section titled “§3. The imaginary unit and the notation a+bi”

The theorem of §2 legitimizes the arithmetic of C\mathbb{C}; it still remains to show that this arithmetic achieves what §3.2 left pending, and to explain the notation a+bia+bi that will finally replace ordered pairs.

Proposition. The map φ:R→C\varphi:\mathbb{R}\to\mathbb{C} given by φ(a)=(a,0)\varphi(a)=(a,0) is injective and satisfies φ(a+b)=φ(a)+φ(b)\varphi(a+b)=\varphi(a)+\varphi(b) and φ(ab)=φ(a)φ(b)\varphi(ab)=\varphi(a)\varphi(b) for all a,b∈Ra,b\in\mathbb{R}.

Proof. Injectivity is immediate: (a,0)=(a′,0)(a,0)=(a',0) requires a=a′a=a', by equality of ordered pairs. For the sum, φ(a)+φ(b)=(a,0)+(b,0)=(a+b,0)=φ(a+b)\varphi(a)+\varphi(b)=(a,0)+(b,0)=(a+b,0)=\varphi(a+b). For the product, φ(a)φ(b)=(a,0)(b,0)=(ab−0, 0+0)=(ab,0)=φ(ab)\varphi(a)\varphi(b)=(a,0)(b,0)=(ab-0,\ 0+0)=(ab,0)=\varphi(ab). ■\blacksquare

This proposition allows us to identify each real aa with the complex number (a,0)(a,0) without ambiguity or loss of structure: the sum and product of reals, seen through φ\varphi, are exactly the sum and product in C\mathbb{C} restricted to the first coordinate. In that precise sense, R\mathbb{R} is a subset of C\mathbb{C}, not merely a similar copy.

Definition. The imaginary unit is i:=(0,1)∈Ci:=(0,1)\in\mathbb{C}.

Proposition. i2=(−1,0)i^2=(-1,0).

Proof. By the definition of the product, i2=(0,1)(0,1)=(0⋅0−1⋅1, 0⋅1+1⋅0)=(−1,0)i^2=(0,1)(0,1)=(0\cdot0-1\cdot1,\ 0\cdot1+1\cdot0)=(-1,0). ■\blacksquare

Identifying (−1,0)(-1,0) with the real number −1-1 via φ\varphi, this proposition says exactly i2=−1i^2=-1: the root that x2+1=0x^2+1=0 lacked in R\mathbb{R} exists in C\mathbb{C}, and it is ii, together with −i-i, since (−i)2=i2=−1(-i)^2=i^2=-1 as well. The obstacle pointed out in §1.2, §1.3 and §3.2 is thereby resolved, with this single construction, once and for all.

It remains to justify the standard notation. For an arbitrary (a,b)∈C(a,b)\in\mathbb{C},

(a,b)=(a,0)+(0,b)=(a,0)+(b,0)(0,1)=φ(a)+φ(b)⋅i(a,b) = (a,0)+(0,b) = (a,0)+(b,0)(0,1) = \varphi(a)+\varphi(b)\cdot i

which, identifying φ(a)\varphi(a) with aa and φ(b)\varphi(b) with bb, is precisely a+bia+bi. From here on, this section, and the rest of the book, writes a+bia+bi in place of (a,b)(a,b), with the guarantee that both notations designate exactly the same object and obey exactly the same arithmetic of §2.


The formula for the multiplicative inverse, found in §2 by direct computation on pairs, has a more transparent reading once the notation a+bia+bi is available, and that reading introduces two tools that the rest of the chapter will use frequently.

Definition. The conjugate of z=a+biz=a+bi is zˉ:=a−bi\bar{z}:=a-bi. The modulus of z=a+biz=a+bi is ∣z∣:=a2+b2|z|:=\sqrt{a^2+b^2}.

Proposition. For every z=a+bi∈Cz=a+bi\in\mathbb{C}, zzˉ=∣z∣2z\bar{z}=|z|^2.

Proof. zzˉ=(a+bi)(a−bi)=a2−abi+abi−b2i2=a2−b2(−1)=a2+b2=∣z∣2z\bar{z}=(a+bi)(a-bi)=a^2-abi+abi-b^2i^2=a^2-b^2(-1)=a^2+b^2=|z|^2, using i2=−1i^2=-1 from §3 and that −abi+abi=0-abi+abi=0 by direct cancellation. ■\blacksquare

This identity explains, in retrospect, the inverse formula found in §2: if z≠0z\neq0, then zzˉ=∣z∣2≠0z\bar{z}=|z|^2\neq0 is a positive real number, and therefore

z−1=zˉ∣z∣2=aa2+b2−ba2+b2iz^{-1} = \frac{\bar{z}}{|z|^2} = \frac{a}{a^2+b^2} - \frac{b}{a^2+b^2}i

exactly the expression already proved, now obtained without manipulating ordered pairs directly, by multiplying and dividing by the conjugate. This is the same procedure that, applied systematically, allows us to divide any two complex numbers: wz=wzˉzzˉ=wzˉ∣z∣2\dfrac{w}{z}=\dfrac{w\bar{z}}{z\bar{z}}=\dfrac{w\bar{z}}{|z|^2}, reducing every division to a multiplication followed by a division by a real number.


C\mathbb{C} has been constructed, proved to be a field, and equipped with the arithmetic (conjugate, modulus, division) that will make the rest of the chapter workable. But the question that motivated this section has been resolved only for a particular case: x2+1x^2+1 has roots in C\mathbb{C}. Does the same hold for every polynomial, of any degree? That is the much more general question that opens the next section.