---
title: Biproduct Quasi-Hopf Algebras
url: https://www.emergentmind.com/topics/biproduct-quasi-hopf-algebras
type: topic
---

# Biproduct Quasi-Hopf Algebras

Biproduct quasi-Hopf algebras are quasi-Hopf analogues of Radford biproducts, bosonizations, and related smash-type constructions. In the standard formulation, one starts with a quasi-Hopf algebra \(H\) and a Hopf algebra object \(B\) in the braided monoidal category of left Yetter–Drinfeld modules \({}_H^H\mathcal{YD}\); the resulting biproduct \(B\# H\) has underlying vector space \(B\otimes H\), a smash-product multiplication, a quasi-coassociative coalgebra structure, and reassociator inherited from \(H\) [1310.4680]. More broadly, recent work places such objects inside a larger family of biproduct-like quasi-Hopf algebras obtained by partial dualization, where the algebra remains an ordinary smash product but the coproduct and associator are twisted by weak splitting data [2309.04886]. In this sense, the subject lies at the intersection of bosonization, cross products, tensor-categorical reconstruction, and the structure theory of quasi-Hopf bimodules [2508.00597].

## 1. Radford-type origin and quasi-Hopf definition

The classical point of departure is the Radford–Majid bosonization paradigm: a braided Hopf algebra \(B\) inside a Yetter–Drinfeld category over a Hopf algebra \(H\) can be “de-braided” to an ordinary Hopf algebra \(B\rtimes H\). In the quasi-Hopf setting, the same pattern persists with the associator \(\Phi\) of \(H\) inserted into the structure maps. For a quasi-Hopf algebra \(H\) and a Hopf algebra \(B\) in \({}_H^H\mathcal{YD}\), the biproduct \(B\# H\) has multiplication
\[
(b\# h)(b'\# h')=(X^1\cdot b)(X^2h_1\cdot b')\# X^3h_2h',
\]
and reassociator
\[
\Phi_{B\# H}=1_B\otimes X^1\otimes 1_B\otimes X^2\otimes 1_B\otimes X^3,
\]
with \(X^1\otimes X^2\otimes X^3=\Phi\) [2508.00597]. This is the basic formal definition of a biproduct quasi-Hopf algebra.

A complementary structural theorem states that if \(A\) is a quasi-Hopf algebra equipped with quasi-Hopf morphisms \(H\xrightarrow{i}A\xrightarrow{\pi}H\) such that \(\pi i=\operatorname{Id}_H\), then \(A\) is isomorphic to a biproduct \(B\# H\) for some Hopf algebra \(B\) in \({}_H^H\mathcal{YD}\) [2508.00597]. Thus biproduct quasi-Hopf algebras are precisely quasi-Hopf algebras with projection, just as Radford biproducts are Hopf algebras with Hopf projection in the strict setting.

The more general cross-product background is formulated in a braided monoidal category \(\mathcal C\). Given algebra and coalgebra objects \(A,B\in\mathcal C\), morphisms
\[
\psi:B\otimes A\to A\otimes B,\qquad \phi:A\otimes B\to B\otimes A
\]
define a cross product algebra and a cross coproduct coalgebra on \(A\otimes B\); necessary and sufficient conditions characterize when these combine into a bialgebra or Hopf algebra [1109.1947]. Biproducts arise as the special case in which the cross algebra is a smash product and the cross coalgebra is a smash coproduct.

## 2. Yetter–Drinfeld reconstruction, coinvariants, and preantipodes

The quasi-Hopf theory of biproducts is anchored in structure theorems for bicomodule algebras and quasi-Hopf bimodules. If \(H\) is a quasi-Hopf algebra, \(B\) an \(H\)-bicomodule algebra, and \(v:H\to B\) a morphism of \(H\)-bicomodule algebras, then the coinvariants \(B^{co(H)}\) acquire the structure of an algebra in the left-left Yetter–Drinfeld category over \(H\), and there is an isomorphism of \(H\)-bicomodule algebras
\[
B^{co(H)}\# H \cong B.
\]
This is the quasi-Hopf analogue of the Radford–Majid decomposition \(B^{coH}\# H\) [1310.4680]. The theorem extends to weak Hopf algebras and braided Hopf algebras, but in the quasi-Hopf case it gives the canonical mechanism by which a braided factor is extracted from a larger bicomodule algebra.

At the level of quasi-Hopf bimodules, the key notion is the preantipode. A quasi-bialgebra admits a preantipode if and only if every quasi-Hopf bimodule decomposes as
\[
M \cong M^{\mathrm{co}A}\otimes A,
\]
where \(M^{\mathrm{co}A}\) denotes the space of coinvariants defined via the corresponding idempotent \(T\) [1501.06061]. The same work proves that the preantipode is unique, stable under gauge transformation, and exists for every quasi-Hopf algebra; if \((s,\alpha,\beta)\) is a quasi-antipode, then
\[
S(a)=\beta\,s(a)\,\alpha
\]
is a preantipode [1501.06061].

These results are structurally decisive for biproduct quasi-Hopf algebras. They explain why bosonization remains available when coassociativity is weakened: the braided factor is recovered as a coinvariant object inside a monoidal, rather than strictly tensorial, framework. They also clarify a common source of confusion. In the quasi-Hopf setting, the analogue of the classical Hopf-module decomposition is not expressed by ordinary coinvariants alone; the projection operator \(T\) and the preantipode are intrinsic parts of the reconstruction data [1501.06061].

## 3. Partial dualization and biproduct-like quasi-Hopf algebras

A major extension of the bosonization picture is provided by partial dualization. Let \(H\) be a finite-dimensional Hopf algebra and \(B\subseteq H\) a left coideal subalgebra. Writing \(B^+=B\cap\ker\varepsilon\), one considers the smash-product algebra
\[
A=(H/B^+H)^*\# B.
\]
Using a partially admissible mapping system \((\pi,\gamma^*)\), one reconstructs on this algebra a quasi-Hopf structure whose representation category is tensor-equivalent to the dual tensor category
\[
\mathsf{Rep}(H)_{\mathsf{Rep}(B)}^*,
\]
hence \(A\) is categorically Morita equivalent to \(H\) [2309.04886].

The striking feature is that the algebra structure is ordinary:
\[
(f\# b)(g\# c)=\sum f(b_{(1)}\rightharpoonup g)\# b_{(2)}c,
\]
while the quasi-Hopf behaviour is concentrated in the coalgebra, associator, and antipode [2309.04886]. The paper emphasizes that this is the main deviation from classical biproducts: multiplication is the usual smash product, but \(\Delta\) is only quasi-coassociative and is rectified by an explicitly constructed associator \(\Phi\).

This construction recovers several strict Hopf-theoretic examples as special cases. For matched pairs of finite groups, the partial dual of \(k(F\bowtie G)\) along \(kF\) is exactly the bismash product Hopf algebra \(kG\# kF\). For bosonizations \(H=B\rtimes A\), the right partial dual becomes a bosonization \(B'\rtimes A\), where \(B'=(B^*)^{\mathrm{op},\mathrm{cop}}\) is dually paired with \(B\) in the Yetter–Drinfeld category [2309.04886]. When the splitting maps are strict—equivalently, when the relevant maps are bialgebra or Hopf morphisms—the associator is trivial and one recovers an ordinary Hopf algebra. This makes precise the statement that Radford biproducts are strict partial duals, whereas general partial duals are weakly split biproduct-like quasi-Hopf algebras [2309.04886].

## 4. Cross products, normality conditions, and quasitriangular factorization

The strict Hopf-theoretic theory of cross products supplies much of the formal template for biproduct quasi-Hopf algebras. In a braided monoidal category, a cross product algebra \(A\#_\psi B\) and a cross product coalgebra \(A\#^\phi B\) are determined by transfer morphisms \(\psi\) and \(\phi\), and there are necessary and sufficient conditions for \(A\otimes B\) to be a bialgebra or Hopf algebra. Normality and conormality of \(\psi\) and \(\phi\) then distinguish smash products, smash coproducts, Radford biproducts, and double cross products [1109.1947]. In particular, a cross product bialgebra is a Radford biproduct precisely when its algebra structure is a smash product and its coalgebra structure is a smash coproduct [1109.1947].

For smash biproduct bialgebras \(A_{\tau}\times_{\sigma}B\), quasitriangular structures admit a refined factorization. Under the hypothesis that \(\sigma\) is right conormal, quasitriangularity is equivalent to the existence of normalized elements
\[
W\in B\otimes B,\qquad X\in A\otimes B,\qquad Y\in B\otimes A,\qquad Z\in A\otimes A
\]
satisfying nineteen explicit identities, and the \(R\)-matrix of the smash biproduct is reconstructed from these four factors [2505.04188]. The same framework specializes to Radford biproducts, bicrossproducts, and duals of double cross products under additional normality or conormality assumptions [2505.04188].

This factorization theory is formulated for bialgebras rather than quasi-bialgebras, but the paper explicitly presents it as a structural blueprint for quasi-Hopf generalization. A plausible implication is that, in quasi-Hopf biproducts, one should again expect the braiding data to split into “pure \(A\)-”, “pure \(B\)-”, and mixed terms, now modified by associator contributions. That perspective is consistent with the way quasi-Hopf bosonizations insert \(\Phi\) into multiplication, comultiplication, and antipode formulas [2505.04188].

## 5. Rank-2 classification and explicit families

The first general classification of genuinely quasi-Hopf biproducts in small rank concerns Hopf algebras of dimension \(2\) inside \({}_H^H\mathcal{YD}\). For a quasi-Hopf algebra \(H\) over a field of characteristic different from \(2\), every \(2\)-dimensional Hopf algebra in \({}_H^H\mathcal{YD}\) is isomorphic to exactly one of two types. The first is the trivial group Hopf algebra \(k[C_2]\) with trivial Yetter–Drinfeld structure. The second is a non-trivial family \(B_{\sigma,v}\), determined by an algebra morphism \(\sigma:H\to k\) and an element \(v\in H\) satisfying \(\varepsilon(v)=1\), \(\sigma(v)=-1\), together with the Yetter–Drinfeld compatibility relations recorded in the classification theorem [2508.00597].

For \(B_{\sigma,v}\), with basis \(\{1,n\}\), the braided Hopf structure is
\[
n^2=0,\qquad \Delta(n)=n\otimes 1+1\otimes n,\qquad S(n)=-n,
\]
while the Yetter–Drinfeld structure is
\[
h\cdot n=\sigma(h)n,\qquad \lambda(n)=v\otimes n.
\]
Thus the only non-trivial \(2\)-dimensional braided factor is nilpotent rather than group-like [2508.00597].

As a consequence, any quasi-Hopf algebra \(A\) equipped with a projection onto \(H\) and free of rank \(2\) as a right \(H\)-module is isomorphic either to \(H_g\cong k[C_2]\otimes H\) or to a non-trivial bosonization \(H^{(0)}_{\sigma,v}\) generated by \(H\) and a nilpotent element \(\theta\) satisfying
\[
\theta^2=0,\qquad h\theta=\sigma(h_1)\theta h_2,
\]
with explicit formulas for \(\Delta(\theta)\) and \(S(\theta)\) in terms of \(\Phi\), \(\sigma\), and \(v\) [2508.00597].

The same paper develops extensive families of examples. For quasi-Hopf algebras \(k^G_\omega\) obtained from function algebras on finite groups with reassociator defined by a \(3\)-cocycle \(\omega\), non-trivial \(2\)-dimensional braided Hopf algebras are classified by central elements \(g\in Z(G)\) and functions \(\varphi:G\to k^\times\) satisfying
\[
\varphi(e)=1,\qquad \varphi(g)=-1,\qquad b_\omega^g=\delta\varphi,
\]
where
\[
b_\omega^g(x,y)=\frac{\omega(g,x,y)\omega(x,y,g)}{\omega(x,g,y)}.
\]
The analysis is carried out for finite abelian groups, the Klein four group, and double dihedral groups, yielding new classes of basic quasi-Hopf algebras of even dimension and new classes of tensor categories [2508.00597]. By contrast, for quasi-Hopf algebras with radical of codimension two, non-trivial rank-\(2\) braided Hopf algebras generally do not exist; in those cases only the trivial \(k[C_2]\)-type biproduct survives, except in Nichols-like twist-equivalent situations [2508.00597].

## 6. Doubles, quantum-group analogues, and related structures

Finite-dimensional quasi-Hopf analogues of Taft algebras provide a second major source of biproduct-like phenomena. The quasi-Hopf algebras \(A(n,s,q)\) have algebra relations
\[
g_2^n=1,\qquad x^n=0,\qquad g_2xg_2^{-1}=qx,
\]
but their comultiplication and reassociator are twisted by a group \(3\)-cocycle. Their Drinfeld doubles satisfy
\[
D(A(n,s,q))\cong Q_sU_q(\mathfrak{sl}_2),
\]
producing quasi-Hopf analogues of small quantum groups [1202.1631]. Structurally, \(A(n,s,q)\) behaves like a cocycle-deformed bosonization of a rank-\(1\) nilpotent part over a cyclic group, and its double behaves like a quasi-Hopf double bosonization [1202.1631].

The even–odd dichotomy is especially significant. For odd \(n\), the double \(D(A(n,q))\) is twist equivalent to Lusztig’s small quantum group \(u_q(\mathfrak{sl}_2)\). For even \(n\), this fails: the quasi-Hopf analogue \(Qu_q(\mathfrak{sl}_2)\) is not twist equivalent to any Hopf algebra [1202.1631]. This resolves a natural misconception. Quasi-Hopf biproduct-type constructions are not merely notational variants of strict Hopf bosonizations; their associators can encode cohomological obstructions that cannot be removed by gauge transformation.

A related strict-Hopf development concerns differential structures. Strongly bicovariant differential graded algebras have been constructed for all four flavours of cross product Hopf algebras, including biproducts \(A{\cdot\kern-.33em\triangleright\!\!<}B\), and the canonical coaction on the braided factor is shown to be differentiable [1904.02662]. Although this is formulated for Hopf algebras rather than quasi-Hopf algebras, it identifies a natural geometric companion to bosonization and indicates how canonical coactions can constrain differential calculi on inhomogeneous quantum groups [1904.02662].

Taken together, these results show that biproduct quasi-Hopf algebras form a coherent but heterogeneous class. In the strict limit they recover Radford biproducts, bosonizations, bicrossproducts, and bismash products. Away from that limit they incorporate weakly split partial duals, projected quasi-Hopf algebras, and cocycle-deformed quantum-group analogues, all governed by the same interaction between braided Yetter–Drinfeld data, smash-type algebra structures, and non-trivial associators [2309.04886].

Source: https://www.emergentmind.com/topics/biproduct-quasi-hopf-algebras