In mathematics, especially in the area of algebra known as group theory, the holomorph of a group ${\displaystyle G}$, denoted ${\displaystyle \operatorname {Hol} (G)}$, is a group that simultaneously contains (copies of) ${\displaystyle G}$ and its automorphism group ${\displaystyle \operatorname {Aut} (G)}$. It provides interesting examples of groups, and allows one to treat group elements and group automorphisms in a uniform context. The holomorph can be described as a semidirect product or as a permutation group.

If ${\displaystyle \operatorname {Aut} (G)}$ is the automorphism group of ${\displaystyle G}$ then

${\displaystyle \operatorname {Hol} (G)=G\rtimes \operatorname {Aut} (G)}$

where the multiplication is given by

${\displaystyle (g,\alpha )(h,\beta )=(g\alpha (h),\alpha \beta ).}$

(1)

Typically, a semidirect product is given in the form ${\displaystyle G\rtimes _{\phi }A}$ where ${\displaystyle G}$ and ${\displaystyle A}$ are groups and ${\displaystyle \phi :A\rightarrow \operatorname {Aut} (G)}$ is a homomorphism and where the multiplication of elements in the semidirect product is given as

${\displaystyle (g,a)(h,b)=(g\phi (a)(h),ab)}$

which is well defined, since ${\displaystyle \phi (a)\in \operatorname {Aut} (G)}$ and therefore ${\displaystyle \phi (a)(h)\in G}$.

For the holomorph, ${\displaystyle A=\operatorname {Aut} (G)}$ and ${\displaystyle \phi }$ is the identity map, as such we suppress writing ${\displaystyle \phi }$ explicitly in the multiplication given in equation ( 1) above.

For example,

• ${\displaystyle G=C_{3}=\langle x\rangle =\{1,x,x^{2}\}}$ the cyclic group of order 3
• ${\displaystyle \operatorname {Aut} (G)=\langle \sigma \rangle =\{1,\sigma \}}$ where ${\displaystyle \sigma (x)=x^{2}}$
• ${\displaystyle \operatorname {Hol} (G)=\{(x^{i},\sigma ^{j})\}}$ with the multiplication given by:

${\displaystyle (x^{i_{1}},\sigma ^{j_{1}})(x^{i_{2}},\sigma ^{j_{2}})=(x^{i_{1}+i_{2}2^{^{j_{1}}}},\sigma ^{j_{1}+j_{2}})}$ where the exponents of ${\displaystyle x}$ are taken mod 3 and those of ${\displaystyle \sigma }$ mod 2.

Observe, for example

${\displaystyle (x,\sigma )(x^{2},\sigma )=(x^{1+2\cdot 2},\sigma ^{2})=(x^{2},1)}$

and this group is not abelian, as ${\displaystyle (x^{2},\sigma )(x,\sigma )=(x,1)}$, so that ${\displaystyle \operatorname {Hol} (C_{3})}$ is a non-abelian group of order 6, which, by basic group theory, must be isomorphic to the symmetric group ${\displaystyle S_{3}}$.

A group G acts naturally on itself by left and right multiplication, each giving rise to a homomorphism from G into the symmetric group on the underlying set of G. One homomorphism is defined as λ: G → Sym(G), ${\displaystyle \lambda _{g}}$(h) = g·h. That is, g is mapped to the permutation obtained by left-multiplying each element of G by g. Similarly, a second homomorphism ρ: G → Sym(G) is defined by ${\displaystyle \rho _{g}}$(h) = h·g⁻¹, where the inverse ensures that ${\displaystyle \rho _{gh}}$(k) = ${\displaystyle \rho _{g}}$(${\displaystyle \rho _{h}}$(k)). These homomorphisms are called the left and right regular representations of G. Each homomorphism is injective, a fact referred to as Cayley's theorem.

For example, if G = C₃ = {1, x, x² } is a cyclic group of order three, then

• ${\displaystyle \lambda _{x}}$(1) = x·1 = x,
• ${\displaystyle \lambda _{x}}$(x) = x·x = x², and
• ${\displaystyle \lambda _{x}}$(x²) = x·x² = 1,

so λ(x) takes (1, x, x²) to (x, x², 1).

The image of λ is a subgroup of Sym(G) isomorphic to G, and its normalizer in Sym(G) is defined to be the holomorph N of G. For each n in N and g in G, there is an h in G such that n·${\displaystyle \lambda _{g}}$ = ${\displaystyle \lambda _{h}}$·n. If an element n of the holomorph fixes the identity of G, then for 1 in G, (n·${\displaystyle \lambda _{g}}$)(1) = (${\displaystyle \lambda _{h}}$·n)(1), but the left hand side is n(g), and the right side is h. In other words, if n in N fixes the identity of G, then for every g in G, n·${\displaystyle \lambda _{g}}$ = ${\displaystyle \lambda _{n(g)}}$·n. If g, h are elements of G, and n is an element of N fixing the identity of G, then applying this equality twice to n·${\displaystyle \lambda _{g}}$·${\displaystyle \lambda _{h}}$ and once to the (equivalent) expression n·${\displaystyle \lambda _{gh}}$ gives that n(g)·n(h) = n(g·h). That is, every element of N that fixes the identity of G is in fact an automorphism of G. Such an n normalizes ${\displaystyle \lambda _{G}}$, and the only ${\displaystyle \lambda _{g}}$ that fixes the identity is λ(1). Setting A to be the stabilizer of the identity, the subgroup generated by A and ${\displaystyle \lambda _{G}}$ is semidirect product with normal subgroup ${\displaystyle \lambda _{G}}$ and complement A. Since ${\displaystyle \lambda _{G}}$ is transitive, the subgroup generated by ${\displaystyle \lambda _{G}}$ and the point stabilizer A is all of N, which shows the holomorph as a permutation group is isomorphic to the holomorph as semidirect product.

It is useful, but not directly relevant, that the centralizer of ${\displaystyle \lambda _{G}}$ in Sym(G) is ${\displaystyle \rho _{G}}$, their intersection is ${\displaystyle \rho _{Z(G)}=\lambda _{Z(G)}}$, where Z(G) is the center of G, and that A is a common complement to both of these normal subgroups of N.

• ρ(G) ∩ Aut(G) = 1
• Aut(G) normalizes ρ(G) so that canonically ρ(G)Aut(G) ≅ G ⋊ Aut(G)
• ${\displaystyle \operatorname {Inn} (G)\cong \operatorname {Im} (g\mapsto \lambda (g)\rho (g))}$ since λ(g)ρ(g)(h) = ghg⁻¹ (${\displaystyle \operatorname {Inn} (G)}$ is the group of inner automorphisms of G.)
• K ≤ G is a characteristic subgroup if and only if λ(K) ⊴ Hol(G)

• Hall, Marshall Jr. (1959), The theory of groups, Macmillan, MR  0103215
• Burnside, William (2004), Theory of Groups of Finite Order, 2nd ed., Dover, p. 87