Hopf–Galois Structures

• Hopf–Galois structures are defined by an isomorphism L ⊗ H → Endₖ(L) and generalize classical Galois actions using cocommutative Hopf algebras.
• The Greither–Pareigis theorem establishes a bijection between these structures and regular subgroups normalized by the left regular representation, linking field theory with abstract group types.
• Techniques involving holomorphs, skew braces, and gamma functions offer practical methods for counting and classifying Hopf–Galois structures, with applications spanning extension theory and quantum algebras.

Hopf–Galois structures are Hopf-theoretic analogues of classical Galois actions on finite field extensions. For a finite separable extension $L/K$, a Hopf–Galois structure consists of a finite cocommutative Hopf algebra $H$ over $K$ acting on $L$ so that the canonical map $L\otimes_K H\to \operatorname{End}_K(L)$ is an isomorphism; in the Galois case with group $G=\operatorname{Gal}(L/K)$, the classical example is $H=K[G]$ (Ezome et al., 2020, Koch et al., 2017). The modern classification begins with the Greither–Pareigis correspondence: if $E/K$ is the Galois closure of $L/K$, $G=\operatorname{Gal}(E/K)$, and $H$0, then Hopf–Galois structures on $H$1 are in bijection with regular subgroups $H$2 normalized by $H$3; in the Galois case $H$4, so one works with regular subgroups of $H$5 normalized by the left regular representation $H$6 (Ezome et al., 2020, Koch et al., 2017). The abstract group $H$7 is the associated group, and its isomorphism class is called the type of the Hopf–Galois structure (Tsang, 2018).

1. Greither–Pareigis classification and the notion of type

Let $H$8 be a finite Galois extension with Galois group $H$9. Write $K$0 for the full permutation group on the underlying set of $K$1, and define the left and right regular representations by

$K$2

Both $K$3 and $K$4 are regular subgroups: they act freely and transitively on $K$5 (Tsang, 2018, Truman, 2022).

A subgroup $K$6 is regular if the evaluation map

$K$7

is bijective, equivalently if the action is transitive and free (Tsang, 2018). In the Galois case, the Greither–Pareigis theorem says that Hopf–Galois structures on $K$8 are in bijection with regular subgroups $K$9 normalized by $L$0; the corresponding Hopf algebra is

$L$1

or, more generally for a separable extension with Galois closure $L$2,

$L$3

under the diagonal action of $L$4 on $L$5 (Koch et al., 2017, Ezome et al., 2020).

This formulation separates the field-theoretic extension from the abstract group type of the acting symmetry. A Hopf–Galois structure is of type $L$6 when the corresponding regular subgroup is abstractly isomorphic to $L$7, and it is of isomorphic type, or of type $L$8, when the associated group is isomorphic to the original Galois group (Tsang, 2018). This distinction is fundamental throughout the subject: many extensions admit structures whose acting group is not isomorphic to the classical Galois group, and much of the literature is concerned with determining which types occur, how many occur, and how their internal Hopf-theoretic properties differ.

2. Holomorphs, regular embeddings, and counting formulas

A central simplification is the passage from regular subgroups of a symmetric group to regular subgroups of a holomorph. For a finite group $L$9,

$L\otimes_K H\to \operatorname{End}_K(L)$0

inside $L\otimes_K H\to \operatorname{End}_K(L)$1 (Tsang, 2018). Byott’s reformulation replaces the search for regular subgroups of $L\otimes_K H\to \operatorname{End}_K(L)$2 normalized by $L\otimes_K H\to \operatorname{End}_K(L)$3 with the search for regular subgroups of $L\otimes_K H\to \operatorname{End}_K(L)$4 isomorphic to $L\otimes_K H\to \operatorname{End}_K(L)$5. If

$L\otimes_K H\to \operatorname{End}_K(L)$6

and

$L\otimes_K H\to \operatorname{End}_K(L)$7

then

$L\otimes_K H\to \operatorname{End}_K(L)$8

when $L\otimes_K H\to \operatorname{End}_K(L)$9 (Tsang, 2018). In particular, for type $G=\operatorname{Gal}(L/K)$0,

$G=\operatorname{Gal}(L/K)$1

For separable extensions, Byott’s translation theorem recasts the Greither–Pareigis classification in terms of transitive subgroups of $G=\operatorname{Gal}(L/K)$2. If $G=\operatorname{Gal}(L/K)$3, $G=\operatorname{Gal}(L/K)$4, and $G=\operatorname{Gal}(L/K)$5, then Hopf–Galois structures of type $G=\operatorname{Gal}(L/K)$6 correspond to transitive embeddings $G=\operatorname{Gal}(L/K)$7 with stabilizer matching $G=\operatorname{Gal}(L/K)$8; the counting formula becomes

$G=\operatorname{Gal}(L/K)$9

where $H=K[G]$0 (Darlington, 5 Aug 2025).

This holomorph language is also the natural setting for computational work, for skew-brace interpretations, and for explicit enumeration problems. It makes the automorphism group of $H=K[G]$1 visible in the counting problem and frequently converts field-theoretic questions into problems about transitive or regular subgroup embeddings.

3. Sub-Hopf algebras, intermediate fields, and the Hopf–Galois correspondence

Hopf–Galois theory has an analogue of the classical Galois correspondence, but in general it is weaker. If $H=K[G]$2 is a Hopf–Galois structure on a finite extension $H=K[G]$3, Chase–Sweedler showed that the map

$H=K[G]$4

is injective and inclusion-reversing, but it need not be surjective (Ezome et al., 2020). In the Greither–Pareigis setting, sub-Hopf algebras correspond to $H=K[G]$5-normalized subgroups of the regular group $H=K[G]$6: if $H=K[G]$7, then the sub-Hopf algebras of $H=K[G]$8 are precisely the $H=K[G]$9 with $E/K$0 normalized by $E/K$1 (Ezome et al., 2020).

When $E/K$2 is Galois and $E/K$3 is regular and normalized by $E/K$4, a normal subgroup $E/K$5 that is also normalized by $E/K$6 gives a Hopf subalgebra

$E/K$7

If $E/K$8, then $E/K$9 for a unique subgroup $L/K$0, and the subgroup $L/K$1 is described concretely by the orbit formula

$L/K$2

The map $L/K$3 is injective but not surjective in general (Koch et al., 2017). Moreover, $L/K$4 is Hopf–Galois with respect to

$L/K$5

so the same abstract group $L/K$6 appears as a semi-regular subgroup of $L/K$7 and as a regular subgroup of $L/K$8 (Koch et al., 2017).

This tower structure extends to exact sequences. If $L/K$9 and both $G=\operatorname{Gal}(E/K)$0 and $G=\operatorname{Gal}(E/K)$1 are normalized by $G=\operatorname{Gal}(E/K)$2, then there is a short exact sequence of Hopf algebras

$G=\operatorname{Gal}(E/K)$3

and, when $G=\operatorname{Gal}(E/K)$4,

$G=\operatorname{Gal}(E/K)$5

The exactness comes from faithful flat descent applied to the exact sequence $G=\operatorname{Gal}(E/K)$6 over the Galois extension $G=\operatorname{Gal}(E/K)$7 (Koch et al., 2017).

A particularly sharp manifestation of failure of surjectivity is the notion of a minimal Hopf–Galois structure: the corresponding Hopf algebra has exactly two sub-Hopf algebras, namely $G=\operatorname{Gal}(E/K)$8 and $G=\operatorname{Gal}(E/K)$9 itself. The group-theoretic criterion is that the regular subgroup $H$00 have no proper nontrivial subgroup normalized by $H$01 (Ezome et al., 2020). For non-abelian simple Galois group $H$02, one of the two Hopf–Galois structures is minimal: its sub-Hopf algebras correspond only to the normal subgroups of $H$03, so only $H$04 and $H$05 occur even though the intermediate field lattice can be much larger (Ezome et al., 2020). This makes precise the fact that the Hopf–Galois correspondence depends on the chosen Hopf–Galois structure, not only on the extension.

4. Brace-theoretic, combinatorial, and orbit-theoretic methods

A major development in the subject is the translation between regular subgroups of holomorphs and skew braces. Given a $H$06-stable regular subgroup $H$07, the bijection $H$08, $H$09, transports the group law of $H$10 to a second operation $H$11 on the underlying set of $H$12, and $H$13 becomes a skew brace, where $H$14 is the original group law on $H$15 (Truman, 2022, Childs, 2019). Conversely, skew braces with multiplicative group $H$16 correspond to $H$17-stable regular subgroups of $H$18. This perspective is especially effective for counting stable subgroups and for comparing different Hopf–Galois structures on the same extension.

One natural equivalence relation is $H$19-conjugation. If $H$20 is $H$21-stable and regular, define

$H$22

Then $H$23 is again a $H$24-stable regular subgroup, and two structures are called $H$25-conjugate if one arises from the other in this way (Truman, 2022). The induced Hopf algebras are isomorphic, and the actions are related by

$H$26

Accordingly, an intermediate field $H$27 is realizable with respect to $H$28 if and only if $H$29 is realizable with respect to $H$30 (Truman, 2022). The same paper shows that freeness of an ambiguous ideal over its associated order is invariant across a $H$31-conjugacy class of Hopf–Galois structures.

For non-abelian characteristically simple groups $H$32, the inner part of the theory is described by fixed point free pairs $H$33 of endomorphisms. Byott–Childs’ formula gives

$H$34

When $H$35 is finite non-abelian simple, every endomorphism of $H$36 lies in the coordinate-permuting class $H$37, and Tsang encodes pairs $H$38 by a graph $H$39 on $H$40. In this setting, $H$41 is fixed point free if and only if $H$42 is a tree, so counting Hopf–Galois structures becomes a problem in labeled tree enumeration (Tsang, 2018).

For extensions of degree $H$43 with $H$44, a different parametrization uses gamma functions

$H$45

satisfying

$H$46

These gamma functions are in one-to-one correspondence with regular subgroups of $H$47, and the paper develops lifting, gluing, and duality methods for handling them in the case where the Sylow $H$48-subgroups are elementary abelian (Campedel et al., 2023). This provides a conceptual route from explicit group structure to complete classification of Hopf–Galois structures and skew braces in that family.

5. Major classification results for important families

Several large classes of finite groups admit complete or near-complete classification results. The following statements organize some of the main ones.

Family	Result	Source
Quasisimple $H$49	$H$50 and $H$51 for all $H$52 of order $H$53	(Tsang, 2020)
$H$54, $H$55 finite non-abelian simple	$H$56 and $H$57	(Tsang, 2018)
Squarefree order $H$58	Explicit formula for $H$59 for all groups $H$60 of order $H$61	(Alabdali et al., 2019)
$H$62 Galois groups	Only specified types occur; for $H$63, only $H$64; for $H$65, only $H$66 and $H$67	(Crespo et al., 2017)
Almost simple $H$68 with socle $H$69 of prime index $H$70	$H$71	(Tsang, 2019)
Degree $H$72, elementary abelian Sylow $H$73-subgroup	Complete classification by regular subgroups of holomorphs and gamma functions	(Campedel et al., 2023)

For finite quasisimple $H$74, the rigidity is exact: a finite Galois extension with Galois group $H$75 admits exactly two Hopf–Galois structures, both of type $H$76, corresponding to the left and right regular representations (Tsang, 2020). This extends the earlier simple-group result and shows that quasisimplicity is still strong enough to rule out all non-isomorphic types.

For non-abelian characteristically simple groups $H$77, the behavior is different: the number of structures of type $H$78 grows rapidly with $H$79, but all of them still arise from inner automorphisms. Tsang proves

$H$80

so

$H$81

For $H$82, this recovers the simple-group count $H$83 (Tsang, 2018).

Squarefree degree admits a complete arithmetic classification. Every group of squarefree order $H$84 is of the form

$H$85

with $H$86, $H$87, and $H$88, and for groups $H$89 of order $H$90 the number of Hopf–Galois structures is given by an explicit formula involving $H$91, $H$92, and prime-indexed sets $H$93 (Alabdali et al., 2019). The paper also gives complete tables for $H$94.

For symmetric and alternating groups, the allowed types can be sharply constrained. If $H$95 is Galois with group $H$96, then the only Hopf–Galois types are $H$97 and $H$98. If the Galois group is $H$99, then the only types are

$K$00

If the Galois group is $K$01, only $K$02 occurs, and if it is $K$03, only $K$04 and $K$05 occur. More generally, for $K$06 or $K$07 with $K$08, there are no cyclic Hopf–Galois structures (Crespo et al., 2017).

For almost simple groups $K$09 with socle $K$10 of prime index $K$11, the possible types are strongly restricted, and the type $K$12 can be counted explicitly: $K$13 This shows that the existence and multiplicity of this type are governed by the order-$K$14 elements outside the socle (Tsang, 2019).

6. Quantitative phenomena, parallel extensions, and generalizations

The weak nature of the Hopf–Galois correspondence can be quantified. For a $K$15-Galois extension with Hopf–Galois structure of type $K$16, the Galois correspondence ratio is

$K$17

equivalently the proportion of intermediate fields lying in the image of the Hopf–Galois correspondence (Childs, 2019). In skew-brace language, this becomes a count of $K$18-stable subgroups of the additive group. The paper computes this ratio for several families, including radical algebras and Zappa–Szép products. In particular, for a bi-skew brace of squarefree order $K$19 with $K$20 and $K$21, one gets

$K$22

when $K$23 is odd squarefree, while

$K$24

which tends to $K$25 as $K$26 (Childs, 2019). This exhibits a large asymmetry between the two directions of the same bi-skew brace.

Parallel extensions introduce another qualitative phenomenon. If $K$27 is a separable extension of degree $K$28 with Galois closure $K$29, a subextension $K$30 is parallel to $K$31 when $K$32. Darlington shows that there are transitive subgroups corresponding to an extension admitting a Hopf–Galois structure but having a parallel extension that admits no Hopf–Galois structures at all; once such an example exists, it extends to an infinite family (Darlington, 2024). By contrast, in degree $K$33 with $K$34 distinct odd primes, no such example exists: if one extension of degree $K$35 in a fixed Galois closure admits a Hopf–Galois structure of type $K$36, then every parallel extension of the same degree also admits a Hopf–Galois structure of type $K$37 (Darlington, 2024).

The computational side of the subject has also become systematic. An algorithm based on transitive subgroups of holomorphs classifies and enumerates Hopf–Galois structures, skew bracoids, skew braces, almost classically Galois structures, and structures with bijective Hopf–Galois correspondence in several low-degree cases (Darlington, 5 Aug 2025). Its guiding principle is precisely the Byott translation: rather than searching in the full symmetric group, it works holomorph by holomorph, then identifies equivalent transitive subgroups across different types $K$38.

Finally, Hopf–Galois ideas extend beyond finite field extensions in the strict Greither–Pareigis sense. Hopf–Galois algebras, or quantum torsors, are algebras $K$39 equipped with a map

$K$40

satisfying torsor identities, and they are equivalent to right Hopf–Galois objects over suitable Hopf algebras (Zheng et al., 2020). The theory extends to generalized ambiskew polynomial rings and to Poisson Hopf–Galois algebras, where Ore-type extensions and Poisson polynomial extensions admit necessary and sufficient criteria for the Hopf–Galois structure to extend (Bichon et al., 2019, Zheng et al., 2020). This suggests that the classical field-extension problem sits inside a broader noncommutative and Poisson framework in which holomorphs, skew-primitives, and descent remain central organizing principles.