In abstract algebra, the dihedral group $ D_n $ (for $ n \geq 3 $) is the finite group of symmetries of a regular $ n $-gon, encompassing all rotations and reflections that map the polygon onto itself, with a total order of $ 2n $.[1][2] These symmetries arise as rigid motions in the plane, forming a non-abelian group under composition for $ n > 2 $, which serves as a fundamental example of how geometric structures give rise to algebraic objects.[3][1] The elements of $ D_n $ consist of $ n $ rotations generated by a single rotation $ r $ of order $ n $ (by angles $ k \cdot \frac{2\pi}{n} $ for $ k = 0, 1, \dots, n-1 $) and $ n $ reflections, each of order 2, often denoted by $ s_k = r^k s $ where $ s $ is a fixed reflection.[1][3] Algebraically, $ D_n $ admits the presentation $ \langle r, s \mid r^n = s^2 = 1, , s r s^{-1} = r^{-1} \rangle $, capturing the key relation that conjugation by a reflection inverts a rotation.[3][4] This structure highlights $ D_n $'s semidirect product form $ \mathbb{Z}_n \rtimes \mathbb{Z}_2 $, where the cyclic subgroup of rotations is normal and index 2.[5] Dihedral groups exhibit notable properties, including n subgroups of order 2 generated by the reflections (and one additional if n is even, generated by the 180° rotation) and a cyclic subgroup of order $ n $ that is characteristic.[6][7] For small values, $ D_3 $ is isomorphic to the symmetric group $ S_3 $ of order 6, modeling symmetries of an equilateral triangle, while $ D_4 $ of order 8 describes the symmetries of a square and contains the Klein four-group as a normal subgroup.[4][1] Beyond geometry, dihedral groups appear in crystallography for classifying molecular symmetries, in combinatorics via counting distinct objects under group actions, and in representation theory as examples of groups with faithful 2-dimensional representations over the reals.[5][2]

Definition and Elements

Generators and Relations

The dihedral group DnD_n (for n≥3n \geq 3) is defined abstractly as the group generated by two elements: a rotation rr of order nn and a reflection ss of order 2, satisfying the key relation srs−1=r−1s r s^{-1} = r^{-1}. This relation encodes the interaction between rotations and reflections, where conjugation by ss inverts the rotation rr, reflecting the geometric action of a reflection on a rotational symmetry. Since s2=1s^2 = 1, the relation simplifies to srs=r−1s r s = r^{-1}. The standard presentation of DnD_n is ⟨r,s∣rn=s2=1,srs=r−1⟩\langle r, s \mid r^n = s^2 = 1, s r s = r^{-1} \rangle, which specifies the generators and the minimal set of relations that define the group up to isomorphism. This presentation captures all relations in DnD_n because any word in rr and ss can be reduced using these rules to one of the distinct forms rkr^k (for k=0,1,…,n−1k = 0, 1, \dots, n-1) or srks r^k (for k=0,1,…,n−1k = 0, 1, \dots, n-1). To see that there are exactly 2n2n elements, note that the rotations {1,r,r2,…,rn−1}\{1, r, r^2, \dots, r^{n-1}\} form a cyclic subgroup of order nn, and the coset s⟨r⟩={s,sr,sr2,…,srn−1}s \langle r \rangle = \{s, s r, s r^2, \dots, s r^{n-1}\} consists of nn additional distinct elements, as the relation srk=r−kss r^k = r^{-k} s (derived iteratively from srs=r−1s r s = r^{-1}) prevents overlap between the cosets. Thus, the order of DnD_n is 2n2n. This algebraic formulation originated from the study of symmetries of regular polygons in the 19th century and was formalized within the emerging theory of finite groups by Camille Jordan.

Elements and Operations

The finite dihedral group DnD_n (for integer n≥3n \geq 3) consists of 2n2n elements, partitioned into nn rotations and nn reflections. The rotations form a cyclic subgroup generated by rr, the rotation by an angle of 2π/n2\pi/n; explicitly, these are the elements rkr^k for k=0,1,…,n−1k = 0, 1, \dots, n-1, where r0r^0 is the identity element. The reflections are the elements srks r^k for k=0,1,…,n−1k = 0, 1, \dots, n-1, where ss denotes a fixed reflection; each such element corresponds to reflection across an axis followed (or preceded) by a rotation.[3] The group operation is determined by the relations rn=er^n = e (identity), s2=es^2 = e, and sr=r−1ss r = r^{-1} s, which allow all products to be reduced to one of the 2n2n standard forms. Products involving rotations follow the cyclic group law: ra⋅rb=r(a+b)mod  nr^a \cdot r^b = r^{(a + b) \mod n}. A rotation times a reflection yields another reflection: ra⋅(srb)=sr(b−a)mod  nr^a \cdot (s r^b) = s r^{(b - a) \mod n}. Similarly, a reflection times a rotation is $ (s r^a) \cdot r^b = s r^{(a + b) \mod n} $. The product of two reflections is a rotation: $ (s r^a) \cdot (s r^b) = r^{(b - a) \mod n} $; this formula holds uniformly for both even and odd nn.[3] The order of each rotation rkr^k is n/gcd⁡(k,n)n / \gcd(k, n), so all rotation orders divide nn. Every reflection has order 2, as $ (s r^k)^2 = s r^k \cdot s r^k = s (r^k s) r^k = s (s r^{-k}) r^k = r^{-k} r^k = e $. These orders highlight the structure: the rotation subgroup is cyclic of order nn, while the reflections are involutions outside it.[3][8] To illustrate the operations and non-commutativity of DnD_n, consider the smallest non-abelian case D3D_3 (order 6, isomorphic to the symmetric group S3S_3), with elements {e,r,r2,s,sr,sr2}\{e, r, r^2, s, s r, s r^2\}, where rr has order 3 and each reflection has order 2. The multiplication table below enumerates all products (rows left-multiplied onto columns); for example, r⋅s=sr2r \cdot s = s r^2 and s⋅r=r2ss \cdot r = r^2 s, showing rs≠srr s \neq s r. Rotations multiply cyclically along the top-left 3×33 \times 3 block, while reflection products yield rotations or the identity.[9][10]
⋅\cdoteerrr2r^2sssrs rsr2s r^2
eeeerrr2r^2sssrs rsr2s r^2
rrrrr2r^2eesr2s r^2sssrs r
r2r^2r2r^2eerrsrs rsr2s r^2ss
sssssrs rsr2s r^2eerrr2r^2
srs rsrs rsr2s r^2ssr2r^2eerr
sr2s r^2sr2s r^2sssrs rrrr2r^2ee

Matrix Representation

The dihedral group DnD_n of order 2n2n admits a faithful two-dimensional representation over the real numbers, where the generators rr (rotation by 2π/n2\pi/n) and ss (a reflection) are mapped to specific orthogonal matrices. The rotation rr is represented by the matrix
R=(cos⁡(2π/n)−sin⁡(2π/n)sin⁡(2π/n)cos⁡(2π/n)), R = \begin{pmatrix} \cos(2\pi/n) & -\sin(2\pi/n) \\ \sin(2\pi/n) & \cos(2\pi/n) \end{pmatrix},
which corresponds to the linear transformation implementing a counterclockwise rotation by angle 2π/n2\pi/n in the plane.[11][12] For the reflection ss, assuming it is over the x-axis, the matrix is
S=(100−1). S = \begin{pmatrix} 1 & 0 \\ 0 & -1 \end{pmatrix}.
The general reflection srks r^k is obtained by conjugating SS with RkR^k, yielding
Sk=RkS(Rk)−1=(cos⁡(2πk/n)sin⁡(2πk/n)sin⁡(2πk/n)−cos⁡(2πk/n)), S_k = R^k S (R^k)^{-1} = \begin{pmatrix} \cos(2\pi k/n) & \sin(2\pi k/n) \\ \sin(2\pi k/n) & -\cos(2\pi k/n) \end{pmatrix},
which represents reflection over the line at angle πk/n\pi k/n.[11][12] These matrices satisfy the dihedral presentation relations, including rn=Ir^n = I (the identity matrix, as powers of RR cycle through rotations summing to full circles) and s2=Is^2 = I (since S2=IS^2 = I). The key relation srs−1=r−1s r s^{-1} = r^{-1} (or srs=r−1s r s = r^{-1} since s=s−1s = s^{-1}) holds under matrix multiplication: compute SRS=(100−1)(cos⁡θ−sin⁡θsin⁡θcos⁡θ)(100−1)=(cos⁡θsin⁡θ−sin⁡θcos⁡θ)S R S = \begin{pmatrix} 1 & 0 \\ 0 & -1 \end{pmatrix} \begin{pmatrix} \cos\theta & -\sin\theta \\ \sin\theta & \cos\theta \end{pmatrix} \begin{pmatrix} 1 & 0 \\ 0 & -1 \end{pmatrix} = \begin{pmatrix} \cos\theta & \sin\theta \\ -\sin\theta & \cos\theta \end{pmatrix}, where θ=2π/n\theta = 2\pi/n, which equals R−1R^{-1} (rotation by −θ-\theta).[11] This representation embeds DnD_n as a finite subgroup of the orthogonal group O(2)O(2), consisting of isometries of the plane that preserve a regular nn-gon centered at the origin, with determinant ±1\pm 1 distinguishing rotations (determinant 1) from reflections (determinant -1).[11][13] Over the complex numbers, the representation extends naturally by identifying the plane with C\mathbb{C}, where rotations act by multiplication by roots of unity: rr maps to multiplication by ω=e2πi/n\omega = e^{2\pi i / n}, and reflections incorporate complex conjugation, yielding a unitary representation in U(2)U(2). In diagonal form, the two-dimensional irreducible representation over C\mathbb{C} is given by r↦diag⁡(ωk,ω−k)r \mapsto \operatorname{diag}(\omega^k, \omega^{-k}) for suitable kk.[11]

Geometric Interpretations

Symmetries of Regular Polygons

The dihedral group DnD_n of order 2n2n is the group of all isometries of the plane that map a regular nn-gon to itself, preserving its position and orientation up to rotation and reflection. These symmetries consist of nn rotational symmetries and nn reflection symmetries, forming a non-abelian group for n≥3n \geq 3.[14][7] The rotational symmetries form a cyclic subgroup of order nn, generated by a rotation rr through an angle of 2π/n2\pi/n radians (or 360∘/n360^\circ/n) around the center of the polygon. Powers of rr, namely rkr^k for k=0,1,…,n−1k = 0, 1, \dots, n-1, correspond to rotations by multiples of 2π/n2\pi/n, with r0r^0 being the identity. This subgroup is isomorphic to the cyclic group Z/nZ\mathbb{Z}/n\mathbb{Z}.[14][7] The reflection symmetries are achieved by flips across nn axes of symmetry passing through the center of the polygon. For odd nn, each axis passes through a vertex and the midpoint of the opposite side. For even nn, half the axes pass through opposite vertices, and the other half pass through the midpoints of opposite sides. Each reflection ss satisfies s2=es^2 = e (the identity), and composing a reflection with a rotation yields another reflection.[7] In terms of group action, the elements of DnD_n permute the nn labeled vertices of the regular nn-gon. The rotations act as cyclic permutations: the generator rr corresponds to an nn-cycle (1 2 … n)(1\ 2\ \dots\ n), and its powers are products of disjoint cycles depending on nn. Reflections act as permutations that fix vertices on the axis (if any) and swap pairs of vertices symmetric across it. Thus, DnD_n is isomorphic to a subgroup of the symmetric group SnS_n generated by this nn-cycle and a suitable reflection permutation, with the rotational symmetries comprising the even permutations within this embedding when nn is odd.[3][7] For a concrete example, consider the square (n=4n=4), whose symmetry group is D4D_4 of order 8. The rotational symmetries are the identity, a 90∘90^\circ clockwise rotation, a 180∘180^\circ rotation, and a 270∘270^\circ rotation. The reflection symmetries include flips across the two diagonals and across the lines joining midpoints of opposite sides (horizontal and vertical, assuming the square is aligned with the axes). These operations permute the four vertices, such as the 90∘90^\circ rotation cycling them as (1 2 3 4)(1\ 2\ 3\ 4), and a diagonal reflection swapping two pairs of vertices.[14]

Rotations in 3D Space

The dihedral group DnD_n of order 2n2n can be realized as a finite subgroup of the special orthogonal group SO(3)\mathrm{SO}(3), consisting entirely of proper rotations in three-dimensional Euclidean space.[15][16] In this embedding, the elements include nn rotations by angles 2πk/n2\pi k / n (for k=0,1,…,n−1k = 0, 1, \dots, n-1) around a principal axis, forming a cyclic subgroup ⟨r⟩≅Zn\langle r \rangle \cong \mathbb{Z}_n isomorphic to the rotational symmetries of a regular nn-gon in the plane perpendicular to that axis.[17] The remaining nn elements are 180-degree rotations (order-2 elements) around axes lying in the equatorial plane, each perpendicular to the principal axis and spaced evenly at angles π/n\pi / n apart.[16] This structure distinguishes the 3D dihedral rotations from their 2D counterparts, where half the elements are reflections, by interpreting all symmetries as orientation-preserving transformations.[15] These dihedral rotation groups in SO(3)\mathrm{SO}(3) arise as the pure rotational symmetries of certain polyhedra, particularly regular nn-gonal prisms and bipyramids. For n=2n=2, D2D_2 (the Klein four-group) describes 180-degree rotations around three mutually perpendicular axes, as in a rhombohedron with unequal edge lengths.[17] For n=3n=3, D3D_3 is the rotation group of the triangular bipyramid, featuring 120- and 240-degree turns around the axis joining the apexes and three 180-degree flips around axes through the midpoints of opposite equatorial edges.[18] Similar patterns hold for n=4n=4 and n=5n=5, corresponding to the rotational symmetries of the square prism and pentagonal bipyramid, respectively, though these are distinct from the larger polyhedral groups of the Platonic solids like the cube (with octahedral rotations S4S_4) or icosahedron (with A5A_5).[16] In contrast to the full symmetry groups of these polyhedra in O(3)\mathrm{O}(3), which incorporate improper rotations equivalent to reflections combined with inversion, the dihedral subgroups in SO(3)\mathrm{SO}(3) focus exclusively on proper rotations.[15] Rotations in these dihedral subgroups can be represented using axis-angle formalism, where each element is specified by a unit vector u\mathbf{u} along the rotation axis and an angle θ\theta. For the principal cyclic subgroup, θ=2πk/n\theta = 2\pi k / n with u\mathbf{u} along the fixed principal axis (e.g., the zz-axis).[18] The 180-degree rotations use θ=π\theta = \pi and axes u\mathbf{u} in the xyxy-plane, such as u=(cos⁡(πj/n),sin⁡(πj/n),0)\mathbf{u} = (\cos(\pi j / n), \sin(\pi j / n), 0) for j=0,1,…,n−1j = 0, 1, \dots, n-1.[19] Alternatively, unit quaternions provide a double-cover representation in SU(2)\mathrm{SU}(2), mapping to SO(3)\mathrm{SO}(3) via the binary dihedral group, though the focus here remains on the axis-angle description for direct geometric intuition.[17] A concrete example is D4D_4 in SO(3)\mathrm{SO}(3), which includes 90-degree increments around the zz-axis—rotations by 0∘0^\circ, 90∘90^\circ, 180∘180^\circ, and 270∘270^\circ—along with four 180-degree rotations around axes at 0∘0^\circ, 45∘45^\circ, 90∘90^\circ, and 135∘135^\circ in the xyxy-plane; this realizes the rotational symmetries of a square prism.[16] For the cube or octahedron, these D4D_4 rotations form a subgroup of the full octahedral group, highlighting how dihedral symmetries embed within broader polyhedral rotation groups.[15]

Examples of Dihedral Symmetries

The dihedral group $ D_2 $, isomorphic to the Klein four-group, represents the symmetries of a non-square rectangle. These symmetries consist of four elements: the identity, a 180° rotation about the center, reflection across the horizontal midline, and reflection across the vertical midline. This group is abelian and arises in contexts where an object has bilateral symmetries but no rotational symmetry beyond 180°.https://gunn-gatm.github.io/textbook/key_chapters/04%20Rotation%20and%20Reflection%20Groups.pdf[3] The dihedral group $ D_4 $ describes the full symmetry group of a square, with eight elements including rotations by 0°, 90°, 180°, and 270° around the center, as well as four reflections: two across axes through midpoints of opposite sides (horizontal and vertical) and two across the diagonals connecting opposite vertices. These reflections preserve the square's edges and vertices under the group action.https://larryriddle.agnesscott.org/ifs/symmetric/D4example.htm[20] For a regular pentagon, the dihedral group $ D_5 $ of order 10 encompasses five rotations by multiples of 72° and five reflections, each passing through a vertex and the midpoint of the opposite side. This configuration ensures that every reflection axis bisects the pentagon symmetrically.https://math.libretexts.org/Courses/Mount_Royal_University/Abstract_Algebra_I/Chapter_3%3A_Permutation_Groups/3.3%3A_Dihedral_Groups_(Group_of_Symmetries)[3] Beyond convex regular polygons, the regular star polygon {5/2}, or pentagram, exhibits the same symmetry group $ D_5 $, as its rotational and reflectional symmetries align identically with those of the enclosing pentagon, despite the intersecting edges.https://math.stackexchange.com/questions/2393640/the-group-of-symmetries-of-the-regular-pentagram-is-isomorphic-to-what-math-gr In real-world applications, dihedral groups appear in molecular structures; for instance, the ammonia molecule (NH₃) has $ D_3 $ symmetry (isomorphic to the point group $ C_{3v} $), featuring three-fold rotational symmetry among the hydrogen atoms and three reflection planes each containing the nitrogen atom and one hydrogen.http://nsmn1.uh.edu/hunger/class/spring_2012/lectures/lecture_2.pdf[21]

Algebraic Properties

Subgroups and Conjugacy Classes

The conjugacy classes of the dihedral group DnD_n, which has order 2n2n, are determined by the action of conjugation on its rotations and reflections. The identity element forms its own conjugacy class of size 1. For non-identity rotations rkr^k where 1≤k≤n−11 \leq k \leq n-1, the conjugacy class consists of {rk,r−k}\{r^k, r^{-k}\}, which has size 2 unless 2k≡0(modn)2k \equiv 0 \pmod{n}; this occurs only when nn is even and k=n/2k = n/2, in which case {rn/2}\{r^{n/2}\} is a singleton class since rn/2r^{n/2} commutes with all reflections.[3] If gcd⁡(k,n)=1\gcd(k, n) = 1, the rotations rkr^k and r−kr^{-k} are distinct and form a paired class, but for general kk, the pairing still holds within the rotation subgroup, extended by the full group action.[3] Reflections in DnD_n form conjugacy classes that depend on the parity of nn. When nn is odd, all nn reflections {srj∣0≤j<n}\{s r^j \mid 0 \leq j < n\} are conjugate to one another, forming a single class of size nn.[3] When nn is even, the reflections split into two distinct conjugacy classes, each of size n/2n/2: one class consists of reflections across axes through opposite vertices (e.g., {sr2j}\{s r^{2j}\}), and the other across axes through midpoints of opposite edges (e.g., {sr2j+1}\{s r^{2j+1}\}).[3] This splitting arises because conjugation by rotations preserves the parity of the axis type, while reflections swap the classes only within their type.[3] By Lagrange's theorem, every proper subgroup of DnD_n has order dividing 2n2n. The rotation subgroup ⟨r⟩≅Cn\langle r \rangle \cong C_n is cyclic of index 2 and normal in DnD_n, as it is the kernel of the sign homomorphism to {±1}\{ \pm 1 \}.[6] Its subgroups are the cyclic groups ⟨rn/d⟩\langle r^{n/d} \rangle of order dd for each divisor d∣nd \mid n. Additionally, for each d∣nd \mid n, DnD_n contains dihedral subgroups isomorphic to DdD_d, generated by rotations of order dd and suitable reflections. When nn is even, DnD_n also contains Klein four-subgroups, such as ⟨rn/2,s⟩\langle r^{n/2}, s \rangle and ⟨rn/2,sr⟩\langle r^{n/2}, s r \rangle, each isomorphic to V4V_4 and consisting of the identity, the 180-degree rotation, and two reflections from the same conjugacy class if n≡0(mod4)n \equiv 0 \pmod{4} or from different conjugacy classes if n≡2(mod4)n \equiv 2 \pmod{4}.[7]

Center and Normal Subgroups

The center of the dihedral group DnD_n, denoted Z(Dn)Z(D_n), consists of the elements that commute with every element in the group. For odd n≥3n \geq 3, Z(Dn)Z(D_n) is trivial, containing only the identity element. For even n>2n > 2, Z(Dn)={e,rn/2}Z(D_n) = \{e, r^{n/2}\}, where rr is a generator of the rotation subgroup and rn/2r^{n/2} is the central 180-degree rotation, so the center is cyclic of order 2. For the special cases n=1n=1 (cyclic of order 2) and n=2n=2 (Klein four-group of order 4), DnD_n is abelian, hence Z(Dn)=DnZ(D_n) = D_n.[22] The normal subgroups of DnD_n play a key role in its structure, with the cyclic rotation subgroup ⟨r⟩\langle r \rangle always normal as it has index 2. Conjugation by any reflection inverts elements of ⟨r⟩\langle r \rangle, so a subgroup H≤⟨r⟩H \leq \langle r \rangle is normal in DnD_n if and only if it is invariant under inversion. Thus, the normal subgroups contained in ⟨r⟩\langle r \rangle are precisely the cyclic subgroups ⟨rd⟩\langle r^d \rangle where dd divides nn and the subgroup is characteristic under the action of inversion. For even nn, additional normal subgroups include ⟨r2⟩\langle r^2 \rangle (cyclic of index 4 in DnD_n) and two more index-2 subgroups: the dihedral subgroup ⟨r2,s⟩\langle r^2, s \rangle and ⟨r2,rs⟩\langle r^2, rs \rangle, where ss is a reflection. The complete lattice of normal subgroups can be constructed via coset decompositions, revealing that DnD_n has exactly three normal subgroups of index 2 when nn is even and only one (⟨r⟩\langle r \rangle) when nn is odd.[23][24] The derived subgroup Dn′D_n' (commutator subgroup) is generated by all commutators [g,h]=g−1h−1gh[g,h] = g^{-1}h^{-1}gh. For even nn, Dn′=⟨r2⟩D_n' = \langle r^2 \rangle, the cyclic subgroup of even-powered rotations. For odd n≥3n \geq 3, since 2 is invertible modulo nn, Dn′=⟨r⟩D_n' = \langle r \rangle, the full rotation subgroup. This reflects the non-abelian nature of DnD_n for n>2n > 2, with commutators arising primarily from relations between rotations and reflections.[25] Dihedral groups DnD_n are solvable for all n≥1n \geq 1, as they admit a subnormal series with cyclic factors: for example, {e}⊴Dn′⊴⟨r⟩⊴Dn\{e\} \trianglelefteq D_n' \trianglelefteq \langle r \rangle \trianglelefteq D_n when nn is odd, terminating in abelian quotients. However, DnD_n is nilpotent if and only if nn is a power of 2 (i.e., DnD_n is a 2-group), in which case it has a central series reaching the trivial group; for other nn, the presence of odd-order elements prevents nilpotency despite solvability. For instance, D4D_4 (order 8) is nilpotent of class 2, while D3D_3 (order 6) is solvable but not nilpotent.[5]

Representations and Characters

The irreducible representations of the dihedral group DnD_n (of order 2n2n) over the complex numbers C\mathbb{C} consist of one-dimensional and two-dimensional components, whose structure depends on the parity of nn. These representations are completely reducible, and their characters provide a powerful tool for analyzing the group's structure.[26] When nn is odd, DnD_n has exactly two one-dimensional irreducible representations and (n−1)/2(n-1)/2 two-dimensional irreducible representations, for a total of (n+3)/2(n+3)/2 irreducible representations. When nn is even, there are four one-dimensional irreducible representations and n/2−1n/2 - 1 two-dimensional irreducible representations, yielding n/2+3n/2 + 3 irreducible representations in total. In both cases, the number of irreducible representations equals the number of conjugacy classes, as required by the theory of character orthogonality.[27][26] The one-dimensional irreducible representations are as follows. For all nn, there is the trivial representation, where every group element acts as multiplication by 1, and the alternating (or sign) representation, where rotations act as 1 and reflections act as -1. When nn is even, two additional one-dimensional representations arise: one sends reflections through vertices to 1 and through edges to -1, while the other interchanges these signs on the reflection classes. These representations act via the nontrivial homomorphism on the rotation Z2\mathbb{Z}_2 factor, sending rn/2r^{n/2} to (−1)n/2(-1)^{n/2}. These four representations correspond to the homomorphisms from DnD_n to {±1}\{\pm 1\}, reflecting the Klein four-group abelianization for even nn.[27][11] The two-dimensional irreducible representations, labeled by k=1,2,…,⌊(n−1)/2⌋k = 1, 2, \dots, \lfloor (n-1)/2 \rfloor, can be defined on C2\mathbb{C}^2 by
ρk(rm)=(ωkm00ω−km),ρk(s)=(0110), \rho_k(r^m) = \begin{pmatrix} \omega^{k m} & 0 \\ 0 & \omega^{-k m} \end{pmatrix}, \quad \rho_k(s) = \begin{pmatrix} 0 & 1 \\ 1 & 0 \end{pmatrix},
where ω=e2πi/n\omega = e^{2\pi i / n} is a primitive nnth root of unity and ss is any reflection (with srms=r−ms r^m s = r^{-m}). The representation ρ1\rho_1 is faithful, embedding DnD_n into GL(2,C)\mathrm{GL}(2, \mathbb{C}) as the symmetries of the regular nn-gon. The characters of these representations are χk(rm)=ωkm+ω−km=2cos⁡(2πkm/n)\chi_k(r^m) = \omega^{k m} + \omega^{-k m} = 2 \cos(2\pi k m / n) on rotations and χk(s)=0\chi_k(s) = 0 on reflections, since the trace of the reflection matrix vanishes. For the one-dimensional representations, the characters take values ±1\pm 1 on the respective conjugacy classes of rotations and reflections.[26][11] The character table of DnD_n is formed by tabulating these values across the conjugacy classes: the identity class, the remaining rotation classes (pairs {rm,r−m}\{r^m, r^{-m}\} for m≠0m \neq 0), the 180-degree rotation class (when nn even), and the one or two reflection classes. This table encodes the traces of the irreducible representations on each class.[27][28] The irreducible characters {χi}\{\chi_i\} satisfy the orthogonality relations: for distinct irreducibles χi,χj\chi_i, \chi_j,
⟨χi,χj⟩=1∣Dn∣∑g∈Dnχi(g)χj(g)‾=0, \langle \chi_i, \chi_j \rangle = \frac{1}{|D_n|} \sum_{g \in D_n} \chi_i(g) \overline{\chi_j(g)} = 0,
and ⟨χi,χi⟩=1\langle \chi_i, \chi_i \rangle = 1. These relations imply that the characters are orthonormal with respect to the inner product on class functions and form a basis for the space of class functions, whose dimension equals the number of conjugacy classes—thus confirming that the listed representations exhaust the irreducibles.[26][29] The regular representation of DnD_n, which acts on C[Dn]\mathbb{C}[D_n] by left translation, has character χreg(g)=∣Dn∣\chi_{\mathrm{reg}}(g) = |D_n| if g=eg = e and 0 otherwise. By the character decomposition formula, it breaks down as Reg≅⨁i(dim⁡ρi)ρi\mathrm{Reg} \cong \bigoplus_i (\dim \rho_i) \rho_i, where the sum is over all irreducible representations ρi\rho_i; the multiplicity dim⁡ρi\dim \rho_i follows from ⟨χreg,χi⟩=dim⁡ρi\langle \chi_{\mathrm{reg}}, \chi_i \rangle = \dim \rho_i. This decomposition underscores the completeness of the representation list, as the dimensions sum to 2n=∣Dn∣2n = |D_n|.[27][26] As an application, the equality between the number of conjugacy classes and irreducible representations is directly verified using characters: the irreducibles, being linearly independent class functions via orthogonality, span a space of dimension equal to the number of classes, ensuring no additional irreducibles exist. This principle, central to Burnside's theorem on character degrees, applies straightforwardly to DnD_n.[29][26]

Automorphisms and Isomorphisms

Automorphism Group Structure

The automorphism group of the dihedral group DnD_n of order 2n2n, for n>2n > 2, is isomorphic to the holomorph Hol(Zn)\mathrm{Hol}(\mathbb{Z}_n) of the cyclic group Zn\mathbb{Z}_n, which is the semidirect product Zn⋊Aut(Zn)\mathbb{Z}_n \rtimes \mathrm{Aut}(\mathbb{Z}_n) where Aut(Zn)≅(Z/nZ)×\mathrm{Aut}(\mathbb{Z}_n) \cong (\mathbb{Z}/n\mathbb{Z})^\times acts by multiplication on Zn\mathbb{Z}_n.[6] This isomorphism arises because any automorphism of Dn=⟨r,s∣rn=s2=1,srs−1=r−1⟩D_n = \langle r, s \mid r^n = s^2 = 1, srs^{-1} = r^{-1} \rangle is uniquely determined by its action on the rotation subgroup ⟨r⟩≅Zn\langle r \rangle \cong \mathbb{Z}_n, extended compatibly to the reflections, preserving the relation srs−1=r−1srs^{-1} = r^{-1}.[30] The order of Aut(Dn)\mathrm{Aut}(D_n) is thus nϕ(n)n \phi(n), where ϕ\phi denotes Euler's totient function, reflecting the nn choices for the image of ss (one for each reflection) combined with the ϕ(n)\phi(n) automorphisms of ⟨r⟩\langle r \rangle.[6] Generators of Aut(Dn)\mathrm{Aut}(D_n) include the automorphisms from Aut(Zn)\mathrm{Aut}(\mathbb{Z}_n), which map r↦rkr \mapsto r^k for gcd⁡(k,n)=1\gcd(k, n) = 1 while fixing ss. The action includes inversion r↦r−1r \mapsto r^{-1} via the unit -1 in Aut(Zn)\mathrm{Aut}(\mathbb{Z}_n), corresponding to conjugation by ss (an inner automorphism). These capture the full semidirect product structure.[30] The outer automorphism group Out(Dn)=Aut(Dn)/[Inn](/page/Inn)(Dn)\mathrm{Out}(D_n) = \mathrm{Aut}(D_n)/\mathrm{[Inn](/page/Inn)}(D_n) has order ϕ(n)\phi(n) for even n>2n > 2 and ϕ(n)/2\phi(n)/2 for odd n>2n > 2, with structure depending on nn; for example, it is trivial for n=3n=3, Z2\mathbb{Z}_2 for n=4,5,6,7,10n=4,5,6,7,10, and Z2×Z2\mathbb{Z}_2 \times \mathbb{Z}_2 for n=8n=8. As an example, for n=3n=3, D3≅S3D_3 \cong S_3 and Aut(S3)≅S3\mathrm{Aut}(S_3) \cong S_3, which has order 6 matching 3ϕ(3)=3⋅2=63 \phi(3) = 3 \cdot 2 = 6. For n=4n=4, Aut(D4)≅D4\mathrm{Aut}(D_4) \cong D_4, also of order 8, illustrating the self-automorphism property in this even case.[5]

Inner Automorphisms

The inner automorphism group of the dihedral group DnD_n, denoted Inn⁡(Dn)\operatorname{Inn}(D_n), comprises all automorphisms induced by conjugation with elements of DnD_n. For any group GG, Inn⁡(G)\operatorname{Inn}(G) is isomorphic to the quotient G/Z(G)G / Z(G), where Z(G)Z(G) denotes the center of GG.[31] In the case of Dn=⟨r,s∣rn=s2=e,srs−1=r−1⟩D_n = \langle r, s \mid r^n = s^2 = e, s r s^{-1} = r^{-1} \rangle, the center Z(Dn)Z(D_n) is trivial when nn is odd, yielding Inn⁡(Dn)≅Dn\operatorname{Inn}(D_n) \cong D_n of order 2n2n. When nn is even and greater than 2, Z(Dn)={e,rn/2}Z(D_n) = \{e, r^{n/2}\} has order 2, so Inn⁡(Dn)≅Dn/Z(Dn)\operatorname{Inn}(D_n) \cong D_n / Z(D_n) has order nn, forming an index-2 subgroup of the full automorphism group Aut⁡(Dn)\operatorname{Aut}(D_n).[32][3] The conjugation map Dn→Aut⁡(Dn)D_n \to \operatorname{Aut}(D_n) has kernel exactly Z(Dn)Z(D_n). Conjugations by powers of the rotation rr act trivially on the cyclic rotation subgroup ⟨r⟩\langle r \rangle, as it is abelian, but cycle the nn reflections: specifically, conjugation by rr maps the reflection srks r^k to srk+1mod  ns r^{k+1 \mod n}, generating a transitive cyclic action of order nn on the set of reflections when nn is odd. Conjugations by reflections invert elements of ⟨r⟩\langle r \rangle, sending rkr^k to r−kr^{-k}. When nn is even, the reflections partition into two conjugacy classes of size n/2n/2 each—one containing axis-through-vertex reflections like ss, and the other axis-through-edge reflections like srs r—and conjugation by a reflection from one class swaps these classes while preserving the inversion on rotations.[3][33] As an illustrative example, consider D4D_4, the dihedral group of order 8 symmetries of the square. Here, Z(D4)={e,r2}Z(D_4) = \{e, r^2\}, so Inn⁡(D4)≅D4/Z(D4)≅Z2×Z2\operatorname{Inn}(D_4) \cong D_4 / Z(D_4) \cong \mathbb{Z}_2 \times \mathbb{Z}_2, the Klein four-group; the inner automorphisms arise from conjugations that quotient out the central 180-degree rotation, effectively identifying opposite symmetries in the quotient.[34]

Isomorphisms to Other Groups

The dihedral group $ D_1 $, which consists of the identity and a single reflection, is isomorphic to the cyclic group $ \mathbb{Z}_2 $. Similarly, $ D_2 $, the symmetry group of a digon or a line segment with two reflections, is isomorphic to the Klein four-group $ \mathbb{Z}_2 \times \mathbb{Z}_2 $, making it the only non-cyclic abelian dihedral group of even order greater than 2.[35] For $ n = 3 $, the dihedral group $ D_3 $ of order 6 is isomorphic to the symmetric group $ S_3 $, the group of all permutations of three elements, as both are the unique non-abelian groups of order 6.[9] In contrast, $ D_4 $ of order 8 is non-abelian but admits no simple abelian isomorphism; it is presented by generators $ r $ and $ s $ satisfying $ r^4 = s^2 = 1 $ and $ srs^{-1} = r^{-1} $, and it is distinct from the quaternion group $ Q_8 $, the other non-abelian group of order 8, since $ D_4 $ has elements of order 4 outside its center while all non-central elements of $ Q_8 $ have order 4 but the groups differ in their subgroup structures (e.g., $ D_4 $ has three subgroups of order 4 isomorphic to $ \mathbb{Z}_2 \times \mathbb{Z}_2 $, whereas $ Q_8 $ has a unique such subgroup).[36] When $ n = p $ is an odd prime, $ D_p $ of order $ 2p $ is the unique non-abelian group up to isomorphism, as classification of groups of order $ 2p $ yields only the cyclic $ \mathbb{Z}_{2p} $ and this dihedral realization, confirmed by Sylow theorems showing a normal Sylow $ p $-subgroup with a complement of order 2 acting non-trivially.[37] For larger $ n $, such as $ n = 6 $, $ D_6 $ of order 12 is isomorphic to the direct product $ S_3 \times \mathbb{Z}_2 $, reflecting its structure as symmetries of a regular hexagon with an additional central involution.[38] However, for $ n > 3 $, dihedral groups $ D_n $ are generally not isomorphic to symmetric groups $ S_n $ or alternating groups $ A_n $, as the latter have orders $ n! $ and $ n!/2 $ respectively, far exceeding $ 2n $, and their element orders and subgroup lattices differ fundamentally (e.g., $ S_n $ for $ n \geq 4 $ contains transpositions absent in $ D_n $).

Generalizations and Extensions

Infinite Dihedral Group

The infinite dihedral group, denoted $ D_\infty $, is the symmetry group of the real line R\mathbb{R}, consisting of all isometries generated by translations along the line and reflections across points on the line. It serves as the infinite analogue of the finite dihedral groups $ D_n $, where the cyclic rotational subgroup of order $ n $ is replaced by an infinite cyclic group of translations, reflecting the unbounded nature of the line. This group captures the full isometry group of R\mathbb{R} under the Euclidean metric, excluding orientation-reversing transformations beyond reflections.[39] Formally, $ D_\infty $ admits the presentation $ \langle r, s \mid s^2 = 1, , s r s = r^{-1} \rangle $, where $ r $ denotes a generator of infinite order corresponding to a primitive translation (e.g., by distance 1), and $ s $ generates a reflection (e.g., across the origin). The elements of $ D_\infty $ can be partitioned into two cosets: the even elements, which are the translations $ r^k $ for $ k \in \mathbb{Z} $, forming the orientation-preserving symmetries; and the odd elements, which are the reflections $ s r^k $ for $ k \in \mathbb{Z} $, combining a reflection with a translation. Every element is uniquely expressible in this form, and the group operation follows from the relation $ s r = r^{-1} s $.[39][40] Algebraically, $ D_\infty $ is isomorphic to the semidirect product $ \mathbb{Z} \rtimes \mathbb{Z}_2 $, where $ \mathbb{Z}_2 = {1, \sigma} $ acts on $ \mathbb{Z} $ via the inversion homomorphism $ \sigma \cdot k = -k $ for $ k \in \mathbb{Z} $; here, $ \mathbb{Z} $ corresponds to the translation subgroup $ \langle r \rangle $, and $ \mathbb{Z}2 $ to the reflection. The translation subgroup $ \langle r \rangle \cong \mathbb{Z} $ is normal in $ D\infty $ with index 2, as conjugation by $ s $ inverts elements of $ \langle r \rangle $, and it is the unique maximal abelian normal subgroup. All proper nontrivial subgroups are either cyclic (isomorphic to $ \mathbb{Z} $ or $ \mathbb{Z}2 $) or isomorphic to $ D\infty $ itself.[40][41] Unlike the finite dihedral groups, which are compact and arise from symmetries of regular polygons, $ D_\infty $ is non-compact, reflecting the infinite extent of the line. It plays a key role in the classification of frieze groups, the infinite discrete symmetry groups of one-dimensional patterns on a strip in the plane; specifically, three of the seven frieze groups are isomorphic to $ D_\infty $, corresponding to patterns with both translations and reflections but no rotations or glide reflections. These applications extend to crystallographic symmetries in low dimensions, where $ D_\infty $ models glide and reflection symmetries in frieze patterns.[39][42]

Dihedral Groups in Higher Dimensions

In three dimensions, dihedral groups extend to full point groups that incorporate both rotational symmetries of the dihedral group DnD_n and reflections, forming the prismatic or polyhedral symmetry groups used in crystallography and molecular modeling. The group DnhD_{nh} consists of the DnD_n rotations combined with a horizontal mirror plane perpendicular to the principal nn-fold axis, plus nn vertical mirror planes containing the axis, resulting in 4n4n elements; this symmetry describes objects like regular prisms with horizontal caps.[43] Similarly, DndD_{nd} includes the DnD_n rotations along with nn dihedral mirror planes bisecting the angles between adjacent 2-fold axes and an inversion center, yielding 4n4n elements without a horizontal mirror; for odd nn, such as n=3n=3, this is denoted D3dD_{3d}.[43] These groups, along with the pure rotational DnD_n, belong to the 32 crystallographic point groups and serve as the rotational-reflection components of space groups describing crystal lattices.[44] In Coxeter notation, the finite dihedral group DnD_n corresponds to the irreducible Coxeter group of type I2(n)I_2(n) with diagram ∘−n−∘\circ -^n - \circ, or symbolically [n,2][n,2], generated by two reflections with angle π/n\pi/n between their planes. Extensions to higher dimensions involve Coxeter groups with diagrams incorporating additional nodes, such as the prismatic groups in three dimensions denoted [n,2,2][n,2,2], which contain DnD_n as a subgroup corresponding to the base polygon's symmetries.[45] In nn-dimensional Euclidean space, the full symmetry group of the hypercube (or cross-polytope) is the hyperoctahedral group of order 2nn!2^n n!, a signed permutation group that includes dihedral subgroups DkD_k for k≤nk \leq n acting on 2-dimensional subspaces or faces.[46] These dihedral point groups appear as subgroups in the 230 space groups of three-dimensional crystallography, governing the symmetries of molecular crystals and arrangements where rotational and reflectional operations preserve the lattice.[44] For instance, in molecular symmetry, the staggered conformation of ethane ($ \ce{C2H6} $) exhibits D3dD_{3d} symmetry, featuring a 3-fold rotation axis along the C-C bond, three 2-fold axes perpendicular to it, three dihedral mirror planes, and an inversion center, which stabilizes the lowest-energy arrangement.[47] This example illustrates how DndD_{nd}-type groups model torsional barriers and vibrational modes in organic molecules within crystallographic contexts.[43]

Coxeter Group Connections

The dihedral group DnD_n of order 2n2n, which consists of the symmetries of a regular nn-gon, is isomorphic to the Coxeter group of type I2(n)I_2(n), an irreducible finite reflection group of rank 2.[48] This identification arises from the fact that DnD_n can be generated by two reflections whose composition yields a rotation by 2π/n2\pi/n. The corresponding Coxeter diagram for I2(n)I_2(n) consists of two vertices connected by an edge labeled nn, denoted as o−n−oo - n - o, where the label specifies the order of the product of the two generating reflections.[48] In this geometric realization, the generators of I2(n)I_2(n) are two reflections across lines (mirrors) that intersect at an angle of π/n\pi/n, producing a kaleidoscopic action that tiles the plane with 2n2n congruent sectors. This rank-2 structure positions the dihedral groups as the fundamental building blocks among finite Coxeter groups, where higher-rank examples are constructed by extending these diagrams.[48] Dihedral groups appear as subgroups within larger Weyl groups associated to root systems. For instance, in the Weyl group of type An−1A_{n-1}, which is the symmetric group SnS_n, dihedral subgroups are generated by transpositions of adjacent elements, corresponding to reflections in the root system.[48] Similarly, the Weyl group of type BnB_n, the hyperoctahedral group, contains dihedral subgroups generated by pairs of orthogonal reflections, reflecting the signed permutation structure.[48] The Artin group associated to the Coxeter system of type I2(n)I_2(n) provides a braid-like generalization, where the generators satisfy a relation alternating between them nn times, such as στσ⋯=τστ⋯\sigma \tau \sigma \cdots = \tau \sigma \tau \cdots with nn factors on each side.[49] For n=3n=3, this Artin group is isomorphic to the 3-strand braid group B3B_3.[49] Finite Coxeter groups, including the dihedral types I2(m)I_2(m) for m≥3m \geq 3, were fully classified by their irreducible diagrams into infinite families AnA_n, BnB_n, DnD_n and exceptional types E6,E7,E8,F4,G2,H3,H4,I2(m)E_6, E_7, E_8, F_4, G_2, H_3, H_4, I_2(m), with dihedrals serving as the rank-2 cases. This classification, established through the geometric and algebraic properties of reflection representations, underscores the dihedral groups' role in unifying symmetry groups across dimensions.[48]