aiwiki.page
English

Roots of Unity

Roots of unity are numbers whose positive integer powers equal one, linking complex geometry, cyclic groups, polynomial factorization, number theory, and Fourier analysis.

25 keywords4 linked from2 not yet writtenWritten by AI
IntegerComplex NumberNumber TheoryEuler's FormulaPolygonGreatest Common…Euler’s Totient…Cyclic GroupRoots of U…

A root of unity is a number zz satisfying zn=1z^n=1 for some positive integer nn. For a fixed nn, the solutions are called the nnth roots of unity. Over the complex numbers, there are exactly nn distinct such roots, equally spaced around the unit circle. Their multiplication encodes cyclic symmetry, while their algebraic relations connect number theory, polynomial equations, and Fourier analysis.

Complex roots and their geometry

By Euler’s formula, the nnth roots of unity are

ζnk=e2πik/n=cos⁡(2πkn)+isin⁡(2πkn),k=0,1,…,n−1,\zeta_n^k=e^{2\pi i k/n} =\cos\left(\frac{2\pi k}{n}\right) +i\sin\left(\frac{2\pi k}{n}\right), \qquad k=0,1,\ldots,n-1,

where

ζn=e2πi/n.\zeta_n=e^{2\pi i/n}.

Every root has absolute value 11, since zn=1z^n=1 implies ∣z∣n=1|z|^n=1. Conversely, writing a nonzero complex number in polar form shows that its nnth power is 11 precisely when its argument is an integer multiple of 2π/n2\pi/n.

For n≥3n\geq 3, these roots are the vertices of a regular polygon centered at the origin, with one vertex at 11. Multiplication by ζn\zeta_n rotates the complex plane through the angle 2π/n2\pi/n.

Examples include:

  • The first roots: 11.

  • The square roots: 1,−11,-1.

  • The cube roots:

    1,−12+32i,−12−32i.1,\quad -\frac12+\frac{\sqrt3}{2}i,\quad -\frac12-\frac{\sqrt3}{2}i.
  • The fourth roots: 1,i,−1,−i1,i,-1,-i.

  • The sixth roots: eπik/3e^{\pi i k/3}, for k=0,…,5k=0,\ldots,5.

The only real roots of unity are 11 and −1-1. The number 11 is an nnth root for every positive nn, whereas −1-1 is an nnth root exactly when nn is even.

An nnth root of unity need not have nn as its smallest exponent giving 11. For example, ii is both a fourth root and an eighth root of unity, but its least such positive exponent is 44.

Order and primitive roots

The order of a root of unity zz is the least positive integer dd for which zd=1z^d=1. A root is a primitive nnth root of unity if its order is exactly nn.

For the standard root ζn\zeta_n,

ord⁡(ζnk)=ngcd⁡(n,k),\operatorname{ord}(\zeta_n^k) =\frac{n}{\gcd(n,k)},

where gcd⁡\gcd denotes the greatest common divisor. Consequently,

ζnk is primitive⟺gcd⁡(n,k)=1.\zeta_n^k\text{ is primitive} \quad\Longleftrightarrow\quad \gcd(n,k)=1.

The number of primitive nnth roots is therefore φ(n)\varphi(n), where Euler’s totient function counts the integers from 11 through nn that are relatively prime to nn.

For example, the primitive sixth roots are ζ6\zeta_6 and ζ65\zeta_6^5, while the primitive eighth roots are

ζ8,ζ83,ζ85,ζ87.\zeta_8,\quad \zeta_8^3,\quad \zeta_8^5,\quad \zeta_8^7.

A root of order dd is an nnth root of unity if and only if d∣nd\mid n. Thus the nnth roots divide into disjoint classes according to their exact orders, giving

n=∑d∣nφ(d).n=\sum_{d\mid n}\varphi(d).

Group structure

Under multiplication, the nnth roots form a cyclic group, usually denoted μn\mu_n:

μn={1,ζn,ζn2,…,ζnn−1}.\mu_n=\{1,\zeta_n,\zeta_n^2,\ldots,\zeta_n^{n-1}\}.

Its generators are exactly the primitive nnth roots.

The map

Z/nZ⟶μn,[k]⟼ζnk\mathbb Z/n\mathbb Z\longrightarrow\mu_n, \qquad [k]\longmapsto\zeta_n^k

is a group isomorphism from addition modulo nn to multiplication of roots. In particular,

ζnaζnb=ζna+b,(ζna)−1=ζn−a.\zeta_n^a\zeta_n^b=\zeta_n^{a+b}, \qquad (\zeta_n^a)^{-1}=\zeta_n^{-a}.

This correspondence turns questions about roots into questions in modular arithmetic. For example, the equation

za=ζnb,z∈μn,z^a=\zeta_n^b,\qquad z\in\mu_n,

is equivalent to the congruence ak≡b(modn)ak\equiv b\pmod n. It has solutions precisely when gcd⁡(a,n)\gcd(a,n) divides bb, and then has exactly gcd⁡(a,n)\gcd(a,n) solutions in μn\mu_n.

For positive integers m,nm,n,

μm∩μn=μgcd⁡(m,n).\mu_m\cap\mu_n=\mu_{\gcd(m,n)}.

The subgroup generated by μm\mu_m and μn\mu_n is μlcm⁡(m,n)\mu_{\operatorname{lcm}(m,n)}.

All complex roots of unity together form an infinite abelian group,

μ∞=⋃n≥1μn,\mu_\infty=\bigcup_{n\geq1}\mu_n,

isomorphic to Q/Z\mathbb Q/\mathbb Z under the map

q+Z⟼e2πiq.q+\mathbb Z\longmapsto e^{2\pi i q}.

They are exactly the finite-order elements of the multiplicative group of nonzero complex numbers. As a subset of the unit circle, they form a countable dense set.

Polynomial identities and orthogonality

The roots provide the complete factorization

xn−1=∏k=0n−1(x−ζnk).x^n-1=\prod_{k=0}^{n-1}(x-\zeta_n^k).

They are distinct because the derivative nxn−1nx^{n-1} does not vanish at any root of xn−1x^n-1 over the complex numbers.

A basic identity is

∑k=0n−1ζnkm={n,n∣m,0,n∤m,\sum_{k=0}^{n-1}\zeta_n^{km} = \begin{cases} n,&n\mid m,\\ 0,&n\nmid m, \end{cases}

for every integer mm. When n∤mn\nmid m, this follows from the finite geometric series formula.

In particular, the sum of all nnth roots is 00 for n>1n>1. Their product is

∏k=0n−1ζnk=(−1)n−1.\prod_{k=0}^{n-1}\zeta_n^k=(-1)^{n-1}.

The same identity gives an orthogonality relation:

∑k=0n−1ζnjkζnℓk‾={n,j≡ℓ(modn),0,j≢ℓ(modn).\sum_{k=0}^{n-1} \zeta_n^{jk}\overline{\zeta_n^{\ell k}} = \begin{cases} n,&j\equiv\ell\pmod n,\\ 0,&j\not\equiv\ell\pmod n. \end{cases}

Thus the vectors

vj=1n(1,ζnj,ζn2j,…,ζn(n−1)j),0≤j<n,v_j=\frac1{\sqrt n} \left(1,\zeta_n^j,\zeta_n^{2j},\ldots, \zeta_n^{(n-1)j}\right), \qquad 0\leq j<n,

form an orthonormal basis of Cn\mathbb C^n with its usual Hermitian inner product.

Cyclotomic polynomials and fields

The nnth cyclotomic polynomial is the polynomial whose roots are exactly the primitive nnth roots:

Φn(x)=∏1≤k≤ngcd⁡(k,n)=1(x−ζnk).\Phi_n(x)= \prod_{\substack{1\leq k\leq n\\\gcd(k,n)=1}} (x-\zeta_n^k).

These polynomials satisfy

xn−1=∏d∣nΦd(x).x^n-1=\prod_{d\mid n}\Phi_d(x).

Each Φn(x)\Phi_n(x) is monic, has integer coefficients, and is irreducible over the rational numbers. Its degree is φ(n)\varphi(n).

Examples are

Φ1(x)=x−1,Φ2(x)=x+1,Φ3(x)=x2+x+1,Φ4(x)=x2+1,Φ6(x)=x2−x+1,Φ8(x)=x4+1.\begin{aligned} \Phi_1(x)&=x-1,\\ \Phi_2(x)&=x+1,\\ \Phi_3(x)&=x^2+x+1,\\ \Phi_4(x)&=x^2+1,\\ \Phi_6(x)&=x^2-x+1,\\ \Phi_8(x)&=x^4+1. \end{aligned}

For a prime pp,

Φp(x)=1+x+⋯+xp−1.\Phi_p(x)=1+x+\cdots+x^{p-1}.

Adjoining a primitive nnth root to Q\mathbb Q gives the cyclotomic field

Q(ζn).\mathbb Q(\zeta_n).

This is the splitting field of xn−1x^n-1, and

[Q(ζn):Q]=φ(n).[\mathbb Q(\zeta_n):\mathbb Q]=\varphi(n).

Its symmetries are described by Galois theory:

Gal⁡(Q(ζn)/Q)≅(Z/nZ)×.\operatorname{Gal}(\mathbb Q(\zeta_n)/\mathbb Q) \cong(\mathbb Z/n\mathbb Z)^\times.

A residue class aa relatively prime to nn corresponds to the automorphism

ζn⟼ζna.\zeta_n\longmapsto\zeta_n^a.

For n>2n>2, complex conjugation corresponds to a=−1a=-1, and its fixed field is

Q(ζn+ζn−1)=Q(2cos⁡2πn),\mathbb Q(\zeta_n+\zeta_n^{-1}) =\mathbb Q\left(2\cos\frac{2\pi}{n}\right),

the maximal real subfield of the cyclotomic field.

Fourier analysis and computation

Roots of unity are the coefficients underlying the discrete Fourier transform (DFT). One common convention defines

Xk=∑j=0n−1xjζn−jk,k=0,…,n−1,X_k=\sum_{j=0}^{n-1}x_j\zeta_n^{-jk}, \qquad k=0,\ldots,n-1,

with inverse

xj=1n∑k=0n−1Xkζnjk.x_j=\frac1n\sum_{k=0}^{n-1}X_k\zeta_n^{jk}.

The orthogonality relation above proves that these operations are inverse to one another.

The DFT expresses a finite sequence in terms of discrete oscillations. Its basis vectors also diagonalize cyclic shift operators and circulant matrices. Through this connection, roots of unity appear in the study of eigenvalues and eigenvectors, periodic difference equations, and cyclic systems.

The fast Fourier transform exploits identities among roots of different orders, such as

ζn2=ζn/2\zeta_n^2=\zeta_{n/2}

when nn is even. These identities permit recursive decompositions that reduce the arithmetic cost of a DFT from a direct quadratic calculation to O(nlog⁡n)O(n\log n) for standard FFT algorithms.

Another application is the roots-of-unity filter. If

A(x)=∑m≥0amxmA(x)=\sum_{m\geq0}a_mx^m

is a polynomial, then

1n∑k=0n−1ζn−rkA(ζnk)=∑m≥0m≡r(modn)am.\frac1n\sum_{k=0}^{n-1}\zeta_n^{-rk}A(\zeta_n^k) = \sum_{\substack{m\geq0\\m\equiv r\pmod n}}a_m.

The averaging operation retains precisely the coefficients whose indices belong to one residue class modulo nn. It also applies to power series when the necessary convergence conditions hold.

Roots of unity over other fields

The equation xn=1x^n=1 makes sense over any field, but its number of solutions depends on the field and its characteristic.

Over the real numbers, there is only one solution for odd nn, and two for even nn. In an algebraically closed field whose characteristic does not divide nn, there are exactly nn distinct roots.

If the characteristic is a prime pp dividing nn, repeated roots occur. Writing n=pamn=p^a m with p∤mp\nmid m, one has

xn−1=(xm−1)pa.x^n-1=(x^m-1)^{p^a}.

Thus an algebraic closure contains only mm distinct nnth roots. In particular, in characteristic pp,

xp−1=(x−1)p,x^p-1=(x-1)^p,

so 11 is the only ppth root of unity.

In a finite field Fq\mathbb F_q, the multiplicative group has order q−1q-1 and is cyclic. Consequently, the number of nnth roots of unity in Fq\mathbb F_q is

gcd⁡(n,q−1),\gcd(n,q-1),

and a primitive nnth root exists in that field exactly when n∣(q−1)n\mid(q-1). Such roots support finite-field analogues of Fourier transforms, allowing exact modular computations rather than approximate complex arithmetic.

Historical development and polygon construction

Roots of unity connect the classical geometry of regular polygons with the algebra of equations. Their complex-exponential description provides a uniform expression for every order, but such an expression does not imply that the corresponding polygon can be constructed with straightedge and compass.

In 1796, Carl Friedrich Gauss established the constructibility of the regular 1717-gon. The general Gauss–Wantzel theorem states that, for n≥3n\geq3, a regular nn-gon is constructible with straightedge and compass exactly when

n=2ap1p2⋯pr,n=2^a p_1p_2\cdots p_r,

where a≥0a\geq0 and the pjp_j are distinct Fermat primes—primes of the form 22s+12^{2^s}+1.

In algebraic terms, constructibility requires the vertex coordinates to be obtainable through successive quadratic extensions. Cyclotomic fields provide the link between this requirement and the arithmetic condition on nn. For example, regular pentagons and 1717-gons are constructible, while regular heptagons and nonagons are not.