Introduction
Rota-Baxter operators and Hopf algebras are core structures in algebra with profound theoretical implications and broad applications. The Rota-Baxter operator was first introduced by Baxter in 1960 to characterize the Spitzer identity in stochastic analysis. Later, it was systematically axiomatized by Rota, and Cartier investigated the structure of free Rota-Baxter algebras, clarifying the intrinsic connections between such operators, shuffle products and multiple zeta values. Since then, it has developed into an independent algebraic branch [1, 2, 3]. In mathematical physics, Connes and Kreimer revealed the deep relation between Hopf algebras and the renormalization procedure in quantum field theory, which provided the core motivation for the interdisciplinary study of these two structures [4]. Since the beginning of the 21st century, Guo and his collaborators have systematically combined Rota-Baxter operators with classical Hopf algebras, greatly advancing the development of the field. His monograph gives a comprehensive summary of the fundamental theories in this direction [5, 6].
Hopf algebras originated from Hopf’s topological research on Lie group manifolds. The algebraic foundations were gradually laid by Borel, Cartier, Milnor and Moore, and the modern axiomatic system together with Sweedler’s notation was finally established by Sweedler. He introduced core tools including convolution products and convolution inverses, making Hopf algebras an independent algebraic discipline separated from topology [7, 8, 9]. The theoretical framework of modern Hopf algebras was further refined in Radford’s textbook [10]. In 2000, Turaev introduced Hopf π-algebras and Hopf π-coalgebras (collectively called Hopf group (co)algebras) indexed by groups, extending classical Hopf algebras to the group-graded setting. Hopf π-coalgebras can be used to construct invariants of principal π-bundles and 3-manifolds, and their representation categories naturally form crossed π-categories. When the indexing group is commutative, such structures reduce to π-colored ribbon Hopf algebras in the sense of Ohtsuki [11, 12, 13]. Virelizier further systematically established the basic theories of Hopf π-coalgebras, including integrals, traces and quasitriangular structures, laying a rigorous algebraic foundation for their topological applications [14]. It should be noted that existing foundational research on group-graded Hopf structures mainly focuses on the coalgebra side, and the core properties related to convolution operations and antipode operators in the framework of Hopf π-algebras have not been systematically investigated.
In 2021, Goncharov established a complete theory of Rota-Baxter operators on cocommutative Hopf algebras, constructed the correspondence among group structures, Lie algebras and Rota-Baxter operators, and revealed the intrinsic relation between these operators and solutions of the Yang–Baxter equation [15]. In the group-graded direction, Ma et al. introduced classical weight-λ Rota-Baxter operators into the framework of Hopf group (co)algebras in 2022, and established the theory of Rota-Baxter Hopf T-algebras, covering graded versions of Rota-Baxter family algebras and Rota-Baxter bialgebras [16]. This work follows the classical definitional paradigm of Rota-Baxter operators on associative algebras, and belongs to the graded extension of the classical Rota-Baxter system. It belongs to a different research line from Goncharov’s Hopf-type Rota-Baxter operators, and the two definitional systems are incompatible with each other.
The aim of this paper is to systematically extend the Rota-Baxter operator theory on cocommutative Hopf algebras established by Goncharov to the framework of Hopf π-algebras, and to establish a group-graded version that is fully compatible with the classical Hopf-type results. During the derivation, we find that this extension cannot be directly based on existing literature: due to the lack of systematic conclusions on basic properties such as convolution operations and antipode compatibility on the Hopf π-algebra side, we need to build the required theoretical framework step by step starting from basic tools.
Given that group-graded Hopf structures involve multiple group indices, families of graded multiplications, graded comultiplications, graded antipodes and multi-layer composite maps, the notational system is hierarchically complex and prone to index confusion and mapping ambiguity. Section 2 is devoted to establishing a unified set of notational conventions throughout the paper, covering basic isomorphisms of vector spaces, layered naming rules for π-algebras, coalgebras and Hopf π-algebras, as well as the graded version of the Heyneman–Sweedler notation. This notational system is fully self-consistent within the scope of this paper. Although it contains a relatively large number of entries, it fundamentally standardizes the use of notations in subsequent derivations and ensures the rigor and readability of the proofs. § 3 collects the basic definitions and preliminaries of Hopf π-algebras, including the rigorous axioms of π-algebras, Hopf π-algebras, generalized graded homomorphisms, as well as basic constructions such as opposite π-algebras, serving as the starting point of the whole paper. § 4 constructs convolution π-algebras induced by coalgebras and π-algebras, verifies their graded algebra axioms, and proves that coalgebra homomorphisms and π-algebra homomorphisms can respectively induce homomorphisms of convolution π-algebras, so as to establish a complete graded convolution calculus. Using this as the core tool, we derive convolution inverse formulas for two typical classes of maps, and further obtain the antipode compatibility criterion for generalized homomorphisms of Hopf π-algebras. We also prove that the antipode operator is an anti-homomorphism of graded algebras and an anti-homomorphism of coalgebras, and establish the bijectivity of the antipode and the Hopf π-algebra structure of the coopposite graded structure in the cocommutative case. These properties are necessary prerequisites for constructing graded Rota-Baxter operators, and have not been systematically discussed in existing literature on Hopf π-algebras. On this basis, Section 5 establishes the axiomatic system of Rota-Baxter operators on cocommutative Hopf π-algebras that is fully compatible with Goncharov’s system, gives the definitions of two families of graded operators of weight 1 and weight −1, and proves that they form a bijective correspondence via the antipode. We construct several explicit examples to verify the existence and richness of the structure, and discuss the basic properties of the operators as well as related derived structures.
Notations
In this section, we establish the notational conventions adopted throughout the paper. The primary purpose of these uniform conventions is to simplify the exposition and avoid any notational ambiguity. A precise understanding of the symbols defined below is essential for following the arguments and results presented in subsequent sections.
Throughout this paper, we work over a ground field K of characteristic zero, unless otherwise specified. The symbol π always denotes an arbitrary group serving as the index set, and p,q,r always stand for arbitrary elements of π.
For any vector space V over K, we fix the following notation:
(i) idV denotes the identity map on V;
(ii) V∗ denotes its dual space HomK(V,K);
(iii) ϕ together with its inverse ϕ−1 denote the canonical linear isomorphism between V and K⊗V={k⊗v∣k∈K, v∈V}, given by ϕ(v)=1K⊗v and ϕ−1(k⊗v)=k⋅v for all v∈V and k∈K;
(iv) ψV together with its inverse ψV−1 denote the canonical linear isomorphism between HomK(K,V) and V, given by ψV(f)=f(1K) and (ψV−1(v))(k)=k⋅v for all f∈HomK(K,V), v∈V and k∈K.
The notation conventions for families of graded algebraic structures below are used consistently throughout the paper.
(a) A π-algebra is denoted by
Hπ:=({Hp}p∈π,{mp,q}p,q∈π,μ).
We use superscripts to distinguish distinct π-algebras sharing the same index set. For example, two distinct π-algebras are given by
Hπ1:=({Hp1}p∈π,{mp,q1}p,q∈π,μ1),Hπ2:=({Hp2}p∈π,{mp,q2}p,q∈π,μ2).
When the index set is replaced by π1 or π2, the corresponding algebras are denoted respectively by
Hπ1:=({Hp}p∈π1,{mp,q}p,q∈π1,μ),Hπ2:=({Hp}p∈π2,{mp,q}p,q∈π2,μ).
Superscripts for distinguishing distinct algebras are independent of the choice of index set. For example, two distinct π1-algebras are written as
Hπ11:=({Hp1}p∈π1,{mp,q1}p,q∈π1,μ1),Hπ12:=({Hp2}p∈π1,{mp,q2}p,q∈π1,μ2).
We also use different capital base letters to refer to different families of π-algebras. When multiple base letters appear in the same context, the multiplication morphisms and unit morphism are labelled with the corresponding base letter to avoid ambiguity. For example, a π-algebra with base letter A takes the form
Aπ:=({Ap}p∈π,{mp,qA}p,q∈π,μA).
All preceding notation rules for superscripts and index sets apply identically to Aπ and any other letter-labelled π-algebras.
(b) A coassociative counital coalgebra is denoted by
C:=(C,Δ,ε).
We use different calligraphic capital letters to denote distinct coalgebras; for example, a second coalgebra is written as
D:=(D,ΔD,εD).
We also use subscript indices to refer to indexed families of coalgebras. For instance, a family of coalgebras indexed by π takes the form
{Hp}p∈π={(Hp,Δp,εp)}p∈π.
When there are multiple such families indexed by π, they are distinguished by superscripts. For example, two distinct families of coalgebras indexed by π are given by
{Hp1}p∈π:={(Hp1,Δp1,εp1)}p∈π,{Hp2}p∈π:={(Hp2,Δp2,εp2)}p∈π.
We adopt the Heyneman–Sweedler notation for comultiplications throughout this paper. For a coalgebra (C,Δ,ε) and any element c∈C, we write Δ(c)=c(1)⊗c(2). For the indexed family of coalgebras {(Hp,Δp,εp)}p∈π, given any p∈π and any h∈Hp, we write Δp(h)=h(1,p)⊗h(2,p).
(c) A Hopf π-algebra is denoted by
Hπ:=({Hp,Δp,εp}p∈π,{mp,q}p,q∈π,μ,{Sp}p∈π).
We use superscripts to distinguish distinct Hopf π-algebras sharing the same index set. For example, two distinct Hopf π-algebras are given by
Hπ1:=({Hp1,Δp1,εp1}p∈π,{mp,q1}p,q∈π,μ1,{Sp1}p∈π),
Hπ2:=({Hp2,Δp2,εp2}p∈π,{mp,q2}p,q∈π,μ2,{Sp2}p∈π).
When the index set is replaced by π1 or π2, the corresponding Hopf algebras are denoted respectively by
Hπ1:=({Hp,Δp,εp}p∈π1,{mp,q}p,q∈π1,μ,{Sp}p∈π1),
Hπ2:=({Hp,Δp,εp}p∈π2,{mp,q}p,q∈π2,μ,{Sp}p∈π2).
Superscripts for distinguishing distinct algebras are independent of the choice of index set. For example, two distinct Hopf π1-algebras are written as
Hπ11:=({Hp1,Δp1,εp1}p∈π1,{mp,q1}p,q∈π1,μ1,{Sp1}p∈π1),
Hπ12:=({Hp2,Δp2,εp2}p∈π1,{mp,q2}p,q∈π1,μ2,{Sp2}p∈π1).
(d) A Rota–Baxter Hopf π-algebra is denoted by
RHπ:=({Hp,Δp,εp}p∈π,{mp,q}p,q∈π,μ,{Sp}p∈π,{Bp}p∈π).
Preliminaries
This section collects the preliminary knowledge required throughout the paper, including the axiomatic definitions of graded algebraic structures and several basic lemmas. These results serve as a necessary foundation for the subsequent construction of convolution π-algebras and the study of Rota–Baxter operators.
We first recall the rigorous definitions of π-algebras, Hopf π-algebras and related basic concepts.
Definition 3.1. (a) A graded algebra indexed by π (abbreviated as π-algebra) Hπ consists of a family of vector spaces {Hp}p∈π, a family of bilinear operator {mp,q:Hp⊗Hq→Hpq}p,q∈π and a linear operator μ:K→He such that for all p,q,r∈π,
mpq,r∘(mp,q⊗idHr)=mp,qr∘(idHp⊗mq,r), mp,e∘(idHp⊗μ)∘ϕ=idHp=me,p∘(μ⊗idHp)∘ϕ. (b) A graded Hopf algebra indexed by π (abbreviated as Hopf π-algebra) Hπ consists of a π-algebra Hπ together with a family of coassociative counital coalgebras {Hp}p∈π and a family of linear operators {Sp:Hp→Hp−1}p∈π such that each multiplication map mp,q and the unit μ are coalgebra homomorphisms, i.e.
Δpq∘mp,q=(mp,q⊗mp,q)∘σ23∘(Δp⊗Δq),mK∘(εp⊗εq)=εpq∘mp,q, Δe∘μ=(μ⊗μ)∘ΔK,εe∘μ=idK and for every p∈π,
mp−1,p∘(Sp⊗idHp)∘Δp=μ∘εp=mp,p−1∘(idHp⊗Sp)∘Δp. (c) Let Φ:π1→π2 be a group homomorphism (isomorphism).
(i) A generalized graded algebra homomorphism (isomorphism) from a π1-algebra Hπ11 to a π2-algebra Hπ22 is a family of linear maps {fp:Hp1→HΦ(p)2}p∈π1 such that for all p,q∈π1,
mΦ(p),Φ(q)2∘(fp⊗fq)=fpq∘mp,q1,fe∘μ1=μ2. (ii) A generalized graded Hopf algebra homomorphism (isomorphism) from a Hopf π1-algebra Hπ11 to a Hopf π2-algebra Hπ22 is a generalized graded algebra homomorphism (isomorphism) {fp:Hp1→HΦ(p)2}p∈π1 from Hπ11 to Hπ22, satisfying that for any p∈π1, fp is a coalgebra homomorphism from Hp1 to HΦ(p)2, and for any p∈π1, fp−1∘Sp1=SΦ(p)2∘fp.
(iii) When π1=π2=π, the above cases reduce to the classical π-algebra homomorphism (isomorphism) and Hopf π-algebra homomorphism (isomorphism), respectively.
(d) A Hopf π-algebra Hπ is called commutative if π is an abelian group and for all p,q∈π, mq,p∘σp,q=mp,q; it is called cocommutative if for every p∈π, Δp=Δpcop.
(e) A π-subalgebra {Gp}p∈π,{mp,q}p,q∈π,μ of the π-algebra Hπ satisfies the following conditions:
(i) For all p∈π, Gp is a subspace of Hp;
(ii) For all p,q∈π, mp,q(Gp⊗Gq)⊆Gpq;
(iii) μ(1K)∈Ge.
(f) A Hopf π-subalgebra {Gp,Δp,εp}p∈π,{mp,q}p,q∈π,μ,{Sp}p∈π of the Hopf π-algebra Hπ satisfies the following conditions:
(i) For any p∈π, (Gp,Δp,εp) is a subcoalgebra of Hp;
(ii) ({Gp}p∈π,{mp,q}p,q∈π,μ) is a π-subalgebra of Hπ;
(iii) For any p∈π, Sp(Gp)⊆Gp−1.
Based on the above definitions, we first present the conclusion on the opposite π-algebra, a fundamental construction in the graded setting.
Lemma 3.2. If Hπ is a π-algebra, then
Hπop:=({Hpop=Hp−1}p∈π,{mp,qop=mq−1,p−1∘σp−1,q−1}p,q∈π,μ) is also a π-algebra.
Proof.
mp,qop∘(mq,rop⊗idHr−1)=mr−1,q−1p−1∘σq−1p−1,r−1∘(mq−1,p−1∘σp−1,q−1⊗idHr−1)=mr−1q−1,p−1∘(mr−1,q−1∘σq−1,r−1⊗idHp−1)=mr−1,q−1p−1∘(idHr−1⊗mq−1,p−1)∘σ13=mr−1q−1,p−1∘(mr−1,q−1⊗idHp−1)∘σ13=mr−1q−1,p−1∘σp−1,r−1q−1∘(idHp−1⊗mr−1,q−1∘σq−1,r−1)=mp,qrop(idHp−1⊗mq,rop)
mp,eop∘(idHp−1⊗μ)∘ϕ=me,p−1∘σp−1,e∘(idHp−1⊗μ)∘ϕ=me,p−1∘(μ⊗idHp−1)∘ϕ=idHp−1=mp−1,e∘(idHp−1⊗μ)∘ϕ=mp−1,e∘σe,p−1∘(μ⊗idHp−1)∘ϕ=me,pop∘(μ⊗idHp−1)∘ϕ
Extending the conclusion on opposite structures to the case of Hopf π-algebras, we obtain the following result.
Lemma 3.3. If Hπ is a Hopf π-algebra, then Hπop,cop:=({Hpop,cop=Hp−1,Δpop,cop=Δp−1cop,εpop,cop=εp−1}p∈π,{mp,qop,cop=mq−1,p−1∘σp−1,q−1}p,q∈π,μ,{Spop,cop=Sp−1}p∈π) is also a Hopf π-algebra.
Proof. By Lemma 1, ({Hpop,cop}p∈π,{mp,qop,cop}p,q∈π,μ) is a π-algebra, and {(Hpop,cop,Δpop,cop,εpop,cop)}p∈π is a family of coalgebras. For any p,q∈π, we have:
Δpqop,cop∘mp,qop,cop=Δq−1p−1cop∘mq−1,p−1∘σp−1,q−1=σ∘Δq−1p−1∘mq−1,p−1∘σp−1,q−1=σ∘(mq−1,p−1⊗mq−1,p−1)∘σ23∘(Δq−1⊗Δp−1)∘σp−1,q−1=σ∘(mq−1,p−1⊗mq−1,p−1)∘(σp−1,q−1⊗σp−1,q−1)∘σ23∘(Δp−1⊗Δq−1)=(mq−1,p−1∘σp−1,q−1⊗mq−1,p−1∘σp−1,q−1)∘σ23∘(Δp−1op⊗Δq−1cop)=(mp,qop,cop⊗mp,qop,cop)∘σ23∘(Δpop,cop⊗Δqop,cop),
εpqop,cop∘mp,qop,cop=εq−1p−1∘mq−1,p−1∘σp−1,q−1=mK∘(εq−1⊗εp−1)∘σp−1,q−1=mK∘(εp−1⊗εq−1)=mK∘(εpop,cop⊗εqop,cop),
meanwhile, Δeop,cop∘μ=(μ⊗μ)∘ΔK and εeop,cop∘μ=idK, i.e. both {mp,qop,cop}p,q∈π and μ are coalgebra homomorphisms.
For any p∈π, we have:
mp−1,pop,cop∘(Spop,cop⊗idHp)∘Δpop,cop=mp−1,p∘σp,p−1∘(Sp−1⊗idHp)∘Δp−1=mp−1,p∘σp,p−1∘(Sp−1⊗idHp)∘σ∘Δp−1=mp−1,p∘(idHp⊗Sp−1)∘Δp−1=u∘εp−1=u∘εpop,cop,
similarly, mp,p−1op,cop∘(idp⊗Spop,cop)∘Δpop,cop=u∘εpop,cop. □
The following lemma will be used in the construction of concrete examples in subsequent sections.
Lemma 3.4. [10] Let U, V be two vector spaces. Define a linear operator
π:HomK(U,K)⊗V→HomK(U,V),π(u∗⊗v)(u)=u∗(u)v for any u∗⊗v∈HomK(U,K)⊗V and u∈U.
Then π is a linear isomorphism from HomK(U,K)⊗V onto Homf(U,V), where Homf(U,V) denotes the space of finite-rank linear operators from U to V.
Convolution -Algebras
This section establishes the complete theoretical framework of convolution π-algebras, which serves as the core technical tool for the analysis of Hopf π-algebra structures and Rota–Baxter operators throughout the paper.
We present the definition of convolution π-algebra and take it as the main tool to study Hopf π-algebra.
Theorem 4.1. Let Hπ be a π-algebra and let C be a coassociative counital coalgebra. Define a family of vector spaces {HomK(C,Hp)}p∈π and a family
of bilinear operators
{HπCmp,q#:HomK(C,Hp)⊗HomK(C,Hq)→HomK(C,Hpq),HπCmp,q#(f⊗g)=fHπC#p,qg=mp,q∘(f⊗g)∘Δ}p,q∈π
and a linear operator
HπCμ#:K→HomK(C,He),HπCμ#(1K)=μ∘ε
then ({HomK(C,Hp)}p∈π,{HπCmp,q#}p,q∈π,HπCμ#) forms a π-algebra. It is called the convolution π-algebra induced by the coalgebra C and the π-algebra Hπ, denoted by Conv(C,Hπ). This notation is used throughout the rest of the paper.
Proof. Take any p,q,r∈π, f∈HomK(C,Hp), g∈HomK(C,Hq), j∈HomK(C,Hr). Then
(mpq,r#∘(HπCmp,q#⊗idHr))(f⊗g⊗j)=mpq,r∘(mp,q∘(f⊗g)∘Δ⊗j)∘Δ=mpq,r∘(mp,q⊗idHr)∘(f⊗g⊗j)∘(Δ⊗idC)∘Δ=mp,qr∘(idHp⊗mq,r)∘(f⊗g⊗j)∘(idC⊗Δ)∘Δ=mp,qr∘(f⊗mq,r∘(g⊗j)∘Δ)∘Δ=(HπCmp,qr#∘(idHp⊗HπCmq,r#))(f⊗g⊗j).
(mp,e#∘(idHomK(C,Hp)⊗HπCμ#)∘ϕ)(f)=mp,e#(f⊗μ∘ε)=mp,e∘(f⊗μ∘ε)∘Δ=mp,e∘(μ⊗idHomK(C,Hp))∘ϕ∘f∘ϕ−1∘(ε⊗idC)∘Δ=idHomK(C,Hp)(f)=mp,e∘(idHomK(C,Hp)⊗μ)∘ϕ∘f∘ϕ−1∘(idC⊗ε)∘Δ=mp,e∘(μ∘ε⊗f)∘Δ=mp,e#(μ∘ε⊗f)=(mp,e#∘(HπCμ#⊗idHomK(C,Hp))∘ϕ)(f).
□
Example 4.2. Let Hπ be a π-algebra.
(a) Given any coassociative counital coalgebra C, define a family of linear operators
{βp:HomK(C,K)⊗Hp→HomK(C,Hp),βp(c∗⊗h)(c)=c∗(c)h}p∈π,
then ({Im(βp)}p∈π,HπCmp,q#,HπCμ#) is a π-subalgebra of Conv(C,Hπ).
(b) In particular, take C=K. Then there is an isomorphism of π-algebras between Conv(K,Hπ) and Hπ.
Proof. (a) By Lemma 3, Im(βp) is a linear subspace of HomK(C,Hp) for each p∈π. Take arbitrary c∗,d∗∈C∗, h∈Hp, g∈Hq and n∈C. We have:
Hπ#mp,q#(βp(c∗⊗h)⊗βq(d∗⊗g))(n)=(mp,q∘(βp(c∗⊗h)⊗βq(d∗⊗g))∘Δ)(n)=c∗(n(1))d∗(n(2))hg=(c∗d∗)(n)hg=βpq(c∗d∗⊗hg)(n)∈Im(βpq),
and Hπ#μ#(1K)=μ∘ε=βe(ε⊗μ(1K))∈Im(βe).
(b) Take a family of linear space isomorphisms
{ψp:HomK(K,Hp)→Hp}p∈π,ψp(f)=f(1K).
For any f∈HomK(K,Hp) and g∈HomK(K,Hq), it holds that
(mp,q∘(ψp⊗ψq))(f⊗g)=mp,q(f(1K)⊗g(1K))=ψpq(mp,q∘(f⊗g)∘ΔK)=(ψpq∘HπKmp,q#)(f⊗g)
and ψe∘HπKμ#=(μ∘εK)(1K)=μ.
In particular, the antipode axiom of Hopf π-algebras can be equivalently characterized by means of convolution inverses.
Homomorphisms between convolution π-algebras are discussed next, which can be naturally induced by homomorphisms of the underlying structures.
Theorem 4.4. Let α:C→D be a coalgebra homomorphism, and let {βp:Hp1→HΦ(p)2}p∈π1 be a generalized graded algebra homomorphism from Hπ11 to Hπ22. Then
(a)
{αp#:HomK(D,Hp1)→HomK(C,Hp1), αp#(f)=f∘α}p∈π1 is a π1-algebra homomorphism from Conv(D,Hπ11) to Conv(C,Hπ11).
(b)
{βp#:HomK(C,Hp1)→HomK(C,HΦ(p)2), βp#(f)=βp∘f}p∈π1 is a generalized graded algebra homomorphism from Conv(C,Hπ11) to Conv(C,Hπ22).
Proof. (1) For any p,q∈π1, f∈HomK(D,Hp1), g∈HomK(D,Hq1), we have
(Hπ11cmp,q#∘(αp#⊗αq#))(f⊗g)=Hπ11cmp,q#((f⊗g)∘(α⊗α))=mp,q1∘(f⊗g)∘(α⊗α)∘ΔC=mp,q1∘(f⊗g)∘ΔC∘α=(Hπ11pmp,q#(f⊗g))∘α=αp,q#(Hπ11pmp,q#(f⊗g))=(αp,q#∘Hπ11cmp,q#)(f⊗g).
and αe#(μ1∘εD)=μ1∘εD∘α=μA∘ε.
(2) For any p,q∈π1, f∈HomK(C,Hp1), g∈HomK(C,Hq1), we have
(βp,q#∘Hπ11cmp,q#)(f⊗g)=βp,q#(Hπ11cmp,q#(f⊗g))=βp,q∘mp,q1∘(f⊗g)∘Δ=mΦ(p),Φ(q)2∘(βp⊗βq)∘(f⊗g)∘Δ=Hπ22cmΦ(p),Φ(q)#((βp⊗βq)∘(f⊗g))=(Hπ22cmΦ(p),Φ(q)#∘(βp#⊗βq#))(f⊗g).
and βe#(μA∘ε)=βe∘μA∘ε=μB∘ε.
Based on the above results on homomorphism induction, the explicit expressions of convolution inverses for two typical classes of maps are presented.
Theorem 4.5. Suppose Hπ1 is a Hopf π1-algebra.
(1) If Aπ2 is a π2-algebra, and {fp:Hp→AΦ(p)}p∈π1 is a generalized graded algebra homomorphism from Hπ1 to Aπ2, then for any p∈π1, fp has a convolution inverse fp−1∘Sp in Conv(Hp,Aπ2).
(2) If C is a coalgebra and fp:C→Hp is a coalgebra homomorphism from C to Hp, then fp∈HomK(C,Hp) has a convolution inverse Sp∘fp in Conv(C,Hπ1).
Proof. (1) For any p∈π1, we have the coassociative counital coalgebra Hp. Since {fp:Hp→AΦ(p)}p∈π1 is a generalized algebra homomorphism from Hπ1 to Aπ2, by Theorem 2, there is a generalized graded algebra homomorphism {fp#}p∈π1 from Conv(Hp,Hπ1) to Conv(Hp,Aπ2).
Since Sp and idp are convolution inverses of each other in Conv(Hp,Hπ1), it follows that fp−1#(Sp)=fp−1∘Sp and fp#(idp)=fp are convolution inverses of each other in Conv(Hp,Aπ2).
(2) It is known that fp is a coalgebra homomorphism from C to Hp, and Hπ1 is a π-algebra, by Theorem 2, there is a convolution π1-algebra homomorphism {frp,#}r∈π from Conv(Hp,Hπ1) to Conv(C,Hπ1).
Since Sp and idp are convolution inverses of each other in Conv(Hp,Hπ), it follows that fp−1p,#(Sp)=Sp∘fp and fp−1p,#(idp)=fp are convolution inverses of each other in Conv(C,Hπ). □
As a direct corollary, the criterion for the antipode compatibility of Hopf π-algebra homomorphisms can be derived.
Corollary 4.6. Let Hπ11 be a Hopf π1-algebra and Hπ22 be a Hopf π2-algebra, and let {fp:Hp1→HΦ(p)2}p∈π1 be a generalized graded algebra homomorphism from Hπ11 to Hπ22. If for each p∈π1, fp is a coalgebra homomorphism from Hp1 to HΦ(p)2, then for any p∈π1, fp−1∘Sp1=SΦ(p)2∘fp.
Proof. For any p∈π1, {fp}p∈π1 is a generalized graded homomorphism from Hπ11 to Hπ22. By Theorem 3, fp:Hp1→HΦ(p)2 is convolution inverse to fp−1∘Sp1 in Conv(Hπ11,Hπ22).
For any p∈π, fp is a coalgebra homomorphism from Hp1 to HΦ(p)2. By Theorem 3, fp:Hp1→HΦ(p)2 is convolution inverse to SΦ(p)2∘fp in Conv(Hπ11,Hπ22).
In summary, for any p∈π1, fp−1∘Sp1=SΦ(p)2∘fp.
The anti-algebra homomorphism property of convolution inverses is stated in the following lemma.
Lemma 4.7. Let Hπ1 be a Hopf π1-algebra and Aπ2 be a π2-algebra. If {fp:Hp→AΦ(p)}p∈π1 is a generalized graded algebra homomorphism from Hπ1 to Aπ2 and for each p∈π1, fp:Hp→AΦ(p) has a convolution inverse fp−1:Hp→AΦ(p)op in Conv(Hπ1,Aπ2); then {fp−1}p∈π1 is a generalized graded algebra homomorphism from Hπ1 to Aπ2op.
Proof. For any p,q∈π, the tensor product coalgebra of Hp and Hq is Hp⊗Hq:=(Hp⊗Hq,Δp,q⊗=σ23∘(Δp⊗Δq),εp,q⊗=mK∘(εp⊗εq)).
Consequently, we have Conv(Hp⊗Hq,Aπ2). For any p,q∈π,
mAπ2#Hp⊗HqΦ(pq)−1,Φ(pq)(fpq−1∘mp,q⊗fpq∘mp,q)=mΦ(pq)−1,Φ(pq)A∘(fpq−1∘mp,q⊗fpq∘mp,q)∘Δp,q⊗=mΦ(pq)−1,Φ(pq)A∘(fpq−1⊗fpq)∘(mp,q⊗mp,q)∘σ23∘(Δp⊗Δq)=mΦ(pq)−1,Φ(pq)A∘(fpq−1⊗fpq)∘Δp,q∘mp,q=μA∘εpq=μ∘mK∘(εp⊗εq)=μA∘εp,q⊗,
mAπ2#Hp⊗HqΦ(pq),Φ(pq)−1(fpq∘mp,q⊗mΦ(p),Φ(q)A,op∘(fp−1⊗fq−1))=mΦ(pq),Φ(pq)−1A∘(fpq∘mp,q⊗mΦ(p),Φ(q)A,op∘(fp−1⊗fq−1))∘Δp,q⊗=mΦ(pq),Φ(pq)−1A∘(mΦ(p),Φ(q)A∘(fp⊗fq)⊗mΦ(p),Φ(q)A,op∘(fp−1⊗fq−1))∘Δp,q⊗=mΦ(pq),Φ(pq)−1A∘(mΦ(p),Φ(q)A⊗mΦ(p),Φ(q)A,op)∘(fp⊗fq⊗fp−1⊗fq−1)∘Δp,q⊗=μA∘mK∘(εp⊗εq)=μA∘εp,q⊗,
that is, fpq−1∘mp,q and mΦ(p)−1,Φ(q)−1A,op∘(fp−1⊗fq−1) are respectively a convolution left inverse and a convolution right inverse of fpq∘mp,q in Conv(Hp⊗Hq,Aπ2).
Therefore, for any p,q∈π, we have mΦ(p)−1,Φ(q)−1A,op∘(fp−1⊗fq−1)=fpq−1∘mp,q.
Since fe and fe−1 are convolution inverses in Conv(He,Aπ2), it follows that
μA=μA∘idK=μA∘εe∘μ=me,eHe,Aπ2∘(fe⊗fe−1)∘μ=me,eA∘(fe⊗fe−1)∘Δe∘μ=me,eA∘(fe⊗fe−1)∘(μ⊗μ)∘ΔK=me,eA∘(fe∘μ⊗fe−1∘μ)∘ΔK=me,eA∘(μA⊗fe−1∘μ)∘ΔK=me,eA∘(μA∘idK⊗idHe∘fe−1∘μ)∘ΔK=me,eA∘(μA⊗idAe)∘(idK⊗fe−1∘μ)∘ΔK=idAe∘fe−1∘μ=fe−1∘μ.
□
Correspondingly, the anti-coalgebra homomorphism property of convolution inverses is presented in the following lemma.
Lemma 4.8. Suppose C is a coassociative counital coalgebra, and Hπ is a Hopf π-algebra. If fp∈HomK(C,Hp) is a coalgebra homomorphism from C to Hp and has a convolution inverse fp−1∈HomK(C,Hp−1) in Conv(C,Hπ), then fp−1 is a coalgebra homomorphism from Ccop to Hp−1.
Proof. It is known that Hπ is a π-algebra. one can easily verify that
Hπ⊗Hπ:=({Hp⊗Hp}p∈π,{mp,q⊗=(mp,q⊗mp,q)∘σ23}p,q∈π,μ⊗=(μ⊗μ)∘ΔK)
forms a π-algebra, and that {Δp:Hp→Hp⊗Hp}p∈π is a π-algebra homomorphism from Hπ to Hπ⊗Hπ. Consequently, we obtain Conv(C,Hπ⊗Hπ). For any p∈π, fp∈HomK(C,Hp), we have
mp,p−1⊗∘(Δp∘fp⊗Δp−1∘fp−1)=mp,p−1⊗∘(Δp∘fp⊗Δp−1∘fp−1)∘Δ=(mp,p−1⊗mp,q)∘σ23∘(Δp⊗Δp−1)∘(fp⊗fp−1)∘Δ=Δe∘mp,q∘(fp⊗fp−1)∘Δ=(μ⊗μ)ΔK=μ⊗∘ε
(ϵHπcmp−1,p#∘((fp−1⊗fp−1)∘ΔCcop⊗Δp∘fp))(d)=(mp−1,p⊗∘((fp−1⊗fp−1)∘ΔCcop⊗Δp∘fp)∘Δ)(d)=(mp−1,p⊗∘((fp−1⊗fp−1)∘ΔCcop⊗(fp⊗fp)∘Δ)∘Δ)(d)=((mp,p−1⊗mp,p−1)∘σ23∘(fp−1⊗fp−1⊗fp⊗fp)∘(ΔCcop⊗Δ)∘Δ)(d)=(mp,p−1⊗mp,p−1)(fp−1(d(12))⊗fp(d(21))⊗fp−1(d(11))⊗fp(d(22)))=mp,p−1(fp−1(d(21))⊗fp(d(22)))⊗mp,p−1(fp−1(d(1))⊗fp(d(3)))=Hπcmp,p−1#(fp−1⊗fp)(d(2))⊗mp,p−1(fp−1(d(1))⊗fp(d(3)))=(μ∘ε)(d(2))⊗mp,p−1(fp−1(d(1))⊗fp(d(3)))=1H⊗Hπcmp,p−1#(fp−1⊗fp)(d)=(1H⊗1H)ε(d)=(μ⊗∘ε)(d)
That is, Δp−1∘fp−1 and (fp−1⊗fp−1)∘ΔCcop are respectively a convolution right inverse and a convolution left inverse of Δp∘fp in Conv(C,Hπ⊗Hπ).
Therefore, for any p∈π,d∈C, we have Δp−1∘(fp−1)=(fp−1⊗fp−1)∘ΔCcop and
εp−1fp−1(d)=εpfp(d1)εp−1fp−1(d2)=εe(fp#p,p−1fp−1)(d)=εe(με(d))=ε(d).
Theorem 4.9. If Hπ is a Hopf π-algebra, then for any p,q∈π,
Spq∘mpq=mp,qop∘(Sp⊗Sq),Se∘μ=μ and for any p∈π,
Δp−1∘Sp=(Sp⊗Sp)∘Δcop,εp−1∘Sp=εp.
Proof. It is known that Hπ is a Hopf π-algebra, then for any p∈π, idHp and Sp are convolution inverses of each other in Conv(Hp,Hπ); Since {idHp:Hp→Hp}p∈π is an endomorphism of Hπ onto itself, by Lemma 4, {Sp}p∈π is a π-algebra endomorphism of Hπ to Hπop. Since for any p∈π, idHp is an endomorphism of Hp onto itself, by Lemma 5, Sp is a coalgebra homomorphism from Hpcop to Hp−1.
The following three results constitute the core content of this paper and serve as the cornerstone for constructing examples of Rota-Baxter Hopf π-algebras in the next section.
Theorem 4.10. Suppose Hπ is a Hopf π-algebra.
(1) If for any p∈π, Sp:Hp→Hp−1 is bijective, then
({Hp,Δpcop,εp}p∈π,{mp,q}p,q∈π,μ,{Spcop=Sp−1−1}p∈π) forms a Hopf π-algebra.
(2) If there exists a family of linear operators {Spcop}p∈π such that
({Hp,Δpcop,εp}p∈π,{mp,q}p,q∈π,μ,{Spcop}p∈π)
forms a Hopf π-algebra, then for any p∈π, Sp is bijective and Spcop=Sp−1−1.
Proof. (1) By Theorem 4, {Sp}p∈π is a π-algebra isomorphism from Hπ to Hπop, then {Fp=Sp−1−1}p∈π is a π-algebra isomorphism from Hπop to Hπ. That is, for any p,q∈π,
mp,q∘(Sp−1⊗Sq−1)=mp,q∘(Fp⊗Fq)=Fpq∘mp,qop=S(pq)−1−1∘mp,qop,
thus for any p∈π, we have
mp−1,p∘(Spop⊗idHp)∘Δpcop=mp−1,p∘(Sp−1−1⊗idHp)∘Δpcop=mp−1,p∘(Sp−1−1⊗Sp−1)∘(idHp⊗Sp)∘Δpcop=Se−1∘mp−1,pop∘(idHp⊗Sp)∘Δpcop=Se∘mp−1,pop∘(idHp⊗Sp)∘Δpcop=Se∘mp−1,p∘(Sp⊗idHp−1)∘Δp=Se∘μ∘ε=με,
similarly, we can prove that
mp,p−1∘(idHp⊗Spop)∘Δpcop=μ∘ε.
(2) Suppose ({Hp,Δpcop,εp}p∈π,{mp,q}p,q∈π,μ,{Spcop}p∈π) is a Hopf π-algebra. Then for any p∈π,
mp−1,p∘(Spop⊗idHp)∘Δpcop=μ∘εp=mp,p−1∘(idHp⊗Spop)∘Δpcop,
furthermore, for any p∈π,
mp,p−1∘(Sp−1∘Spcop⊗Sp)∘Δp=Se∘mp,p−1∘(idHp⊗Spcop)∘Δpcop=Se∘μ∘εp=μ∘εp,
Sp−1∘Spcop is a left inverse of Sp in ({HomK(Hp,Hr)}r∈π,{mr,sp,#}r,s∈π,μp,#), therefore Sp−1∘Spcop=idHp.
Similarly, for any p∈π, we have
mp−1,p∘(Spcop∘Sp−1⊗Sp−1)∘Δp−1=μ∘εp∘Sp−1=μ∘εp−1,
Spcop∘Sp−1 is a left inverse of Sp−1 in ({HomK(Hp,Hr)}r∈π,{mr,sp,#}r,s∈π,μp,#), hence Spcop∘Sp−1=idp−1 □
Corollary 4.11. If Hπ is a cocommutative Hopf π-algebra, then for any p∈π, Sp−1=Sp−1.
Rota-Baxter Hopf π-algebras
Based on the convolution π-algebra tools and antipode properties established in the previous section, this section extends Rota–Baxter operators to cocommutative Hopf π-algebras, presents the axiomatic definitions and weight transition relations, constructs explicit examples, and discusses the basic properties of the operators as well as the derived structures.
Definition 5.1. A Rota–Baxter Hopf π-algebra RHπ consists of a cocommutative Hopf π-algebra Hπ endowed with a family of linear maps, falling into one of the following two weight classes:
(1) Weight 1
The family of maps takes the form {Bp:Hp→Hp−1}p∈π, and satisfies:
(i) For all p∈π and h∈Hp,
Bp(h)(1,p−1)⊗Bp(h)(2,p−1)=Bp(h(1,p))⊗Bp(h(2,p)),εp−1(Bp(h))=εp(h). (ii) For any p,q∈π, h∈Hp and g∈Hq,
Bp(h)Bq(g)=Bqp(h(1,p)Bp(h(2,p))gSp−1Bp(h(3,p))). (2) Weight −1
The family of maps takes the form {Bp:Hp→Hp}p∈π, and satisfies:
(i) For all p∈π and h∈Hp,
Bp(h)(1,p)⊗Bp(h)(2,p)=Bp(h(1,p))⊗Bp(h(2,p)),εp(Bp(h))=εp(h); (ii) For any p,q∈π, h∈Hp and g∈Hq,
Bp(h)Bq(g)=Bpq(Bp(h(1,p))gSpBp(h(2,p))h(3,p)).
The two weight classes of Rota–Baxter structures are not independent of each other; a bijective correspondence between them can be established by means of the antipode operators.
Proposition 5.2. Let RHπ be a Rota–Baxter Hopf π-algebra.
(1) If RHπ is of weight 1, then (Hπ,{Rp=Bp−1∘Sp}p∈π) is a Rota–Baxter Hopf π-algebra of weight −1;
(2) If RHπ is of weight −1, then (Hπ,{Bp=Rp−1∘Sp}p∈π) is a Rota–Baxter Hopf π-*algebra of weight 1.
Proof. For any p,q∈π, h∈Hp and g∈Hq,
Rp(h)Rq(g)=Bp−1Sp(h)Bq−1Sq(g)=Bq−1p−1(Sp(h)(1,p−1)Bp−1(Sp(h)(2,p−1))Sq(g)Sp−1Bp−1(Sp(h)(3,p−1)))=Bq−1p−1(Sp(h(1,p))Bp−1Sp(h(2,p))Sq(g)SpBp−1Sp(h(3,p)))=Bq−1p−1(Sp(h(1,p))Sp−1SpBp−1(h(2,p))Sq(g)SpBp−1Sp(h(3,p)))=Bq−1p−1Spq(Bp−1Sp(h(1,p))gSpBp−1Sp(h(2,p))h(3,p))=Rpq(Rp(h(1,p))gSpRp(h(2,p))h(3,p)).
that is, the family {Rp}p∈π on the cocommutative Hopf π-algebra Hπ forms a Rota–Baxter algebra of weight −1.
Part (2) follows by completely symmetric arguments: starting from the weight −1 Rota–Baxter identity and applying the same antipode properties, one verifies that Bp=Rp−1∘Sp satisfies the weight 1 axioms. □
Example 5.4. (1) Let RHπ be a Rota–Baxter Hopf π-algebra of weight 1. Then the map
B:He→He,h↦Se(h(1,e))BeSe(h(2,e)) is a Rota–Baxter operator of weight 1 on the cocommutative Hopf algebra (He,me,e,μ,Δe,ϵe,Se).
(2) Let Hπ be a Rota–Baxter Hopf π-algebras of weight −1, then the map
B:He→He,h↦h(1,e)BeSe(h(2,e)) is a Rota–Baxter operators of weight −1 on the cocommutative Hopf algebra (He,me,e,μ,Δe,ϵe,Se).
Applying Corollary 2, we have:
Proposition 5.5. For any cocommutative Hopf π-algebra Hπ we have for any p,q∈π, h∈Hpq,
Sqp(h(1,p)Sp(h(2,p))gSp−1Sp(h(3,p)))=Sqp(ϵp(h(1,p))gSp−1Sp(h(2,p)))=Sqp(gSp−1Sp(h))=Sqp(gh)=Sp(h)Sq(g), that is, (Hπ,{Sp}p∈π) is a Rota–Baxter Hopf π-algebra of weight 1.
By virtue of the canonical Rota–Baxter property of the antipode, the most basic group-algebra type example can be constructed directly.
Example 5.6. Let π be an arbitrary group. We construct a cocommutative Hopf π-algebra Hπ as follows.
For each p∈π, the graded component is Hp=K⋅p, the one-dimensional vector space spanned by the group element p. The graded multiplication mp,q:Hp⊗Hq→Hpq is the linear extension of the group multiplication, satisfying mp,q(p⊗q)=pq. The unit map μ:K→He is given by μ(1K)=e, where e denotes the identity element of π.
For each p∈π, the graded comultiplication Δp:Hp→Hp⊗Hp and the graded counit εp:Hp→K are defined by Δp(p)=p⊗p,εp(p)=1, respectively. In other words, every group element is group-like. The antipode family {Sp:Hp→Hp−1} is defined by Sp(p)=p−1.
By Proposition 3, taking Bp=Sp, i.e., Bp(p)=p−1 for all p∈π, yields a Rota–Baxter Hopf π-algebra of weight 1.
Second, by Proposition 2, the weight −1 counterpart of the above operator is the identity family Bp=idHp, given by Bp(p)=p for all p∈π.
Example 5.7. Let π=S3, the symmetric group on three elements. Let V4={e,a,b,c} be the Klein four-group, with multiplication given by a2=b2=c2=e, ab=ba=c, bc=cb=a, ca=ac=b.
The group S3 acts on V4 by group automorphisms, permuting the three order-2 elements a,b,c. Identifying a,b,c with labels 1,2,3 respectively, for any p∈S3 the action satisfies
p⇀e=e,p⇀a=ap(1),p⇀b=ap(2),p⇀c=ap(3), where a1=a,a2=b,a3=c.
We construct a crossed product cocommutative Hopf π-algebra Hπ as follows: For each p∈π, the graded component is Hp=KV4.
The graded multiplication mp,q:Hp⊗Hq→Hpq is given by the composition of group algebra multiplication and the group action, satisfying mp,q(h⊗g)=h⋅(p⇀g) for any h,g∈V4, where ⋅ denotes the multiplication in the group algebra KV4.The unit map μ:K→He satisfies μ(1K)=e.
For each p∈π, the graded comultiplication Δp:Hp→Hp⊗Hp and the graded counit εp:Hp→K satisfy Δp(g)=g⊗g,εp(g)=1 for any g∈V4. The antipode family {Sp:Hp→Hp−1} satisfies Sp(h)=p−1⇀h for any h∈V4.
Based on this Hopf π-algebra, taking the operator family Bp=Sp, i.e., Bp(h)=p−1⇀h for all p∈π and h∈V4, yields a Rota–Baxter Hopf π-algebra of weight 1, the corresponding weight −1 operator is the identity family Bp=idHp, i.e., Bp(h)=h for all p∈π and h∈V4.
This example focuses on the non-trivial graded multiplication induced by a non-abelian index group, with the canonical antipode Rota–Baxter operator.In contrast, we take an abelian index group below, where the crossed product Hopf π-algebra admits multiple distinct Rota–Baxter operators beyond the canonical one.
Example 5.8. Let π=Z/2Z={0,1} denote the additive group of integers modulo 2. Let V4={e,a,b,c} be the Klein four-group in previous example. The group π acts on V4 by group automorphisms: the generator 1∈π swaps a and b while fixing e and c, i.e.,
1⇀e=e,1⇀a=b,1⇀b=a,1⇀c=c.
We construct a crossed-product cocommutative Hopf π-algebra Hπ as follows: For each p∈π, the graded component is Hp=KV4. The graded multiplication mp,q:Hp⊗Hq→Hp+q is defined by mp,q(h⊗g)=h⋅(p⇀g) for any h,g∈V4, where ⋅ denotes the multiplication in the group algebra KV4. The unit map μ:K→H0 is given by μ(1K)=e.
For each p∈π, the graded comultiplication Δp:Hp→Hp⊗Hp and the graded counit εp:Hp→K satisfy Δp(h)=h⊗h, εp(h)=1 for any h∈V4. The antipode family {Sp:Hp→H−p=Hp} satisfies Sp(h)=p⇀h for any h∈V4.
On this Hopf π-algebra, we construct three distinct families of Rota–Baxter operators. For each family, both the weight-1 form and the corresponding weight-(−1) form are presented below.
The operator family of weight 1 is given by Bp(1)(h)=p⇀h for any h∈V4; the associated operator family of weight −1 reads Rp(1)(h)=h for any h∈V4.
The operator family of weight 1 satisfies
Bp(2)(e)=e,Bp(2)(a)=p⇀b,Bp(2)(b)=p⇀a,Bp(2)(c)=c.
The corresponding operator family of weight −1 satisfies
Rp(2)(e)=e,Rp(2)(a)=b,Rp(2)(b)=a,Rp(2)(c)=c.
The operator family of weight 1 satisfies
Bp(3)(e)=e,Bp(3)(a)=p⇀c,Bp(3)(b)=p⇀c,Bp(3)(c)=e.
The corresponding operator family of weight −1 satisfies
Rp(3)(e)=e,Rp(3)(a)=c,Rp(3)(b)=c,Rp(3)(c)=e.
The above examples verify the existence and richness of Rota–Baxter Hopf π-algebra structures. In what follows, we discuss the basic properties of Rota–Baxter operators from three aspects: preservation of canonical structural elements, equivalent characterization in the commutative and cocommutative case, and invariance under automorphisms of the underlying Hopf π-algebra. The above examples confirm the existence and diversity of such structures. We next discuss the preservation properties of the operators with respect to typical structural elements in Hopf algebras.
Proposition 5.9. Suppose RHπ is a Rota–Baxter Hopf π-algebra of weight 1. Then:
(1) If g∈Hp is a group-like element in Hp, then Bp(g) is a group-like element in Hp−1;
(2) Be∘μ=μ;
(3) if a is a primitive element (i.e., a∈He and Δe(a)=a⊗1+1⊗a), then Be(a) is also a primitive element.
Proof. (1) For any group-like element g∈Hp, we have
Δp−1Bp(g)=Bp(g)(1,p−1)⊗Bp(g)(2,p−1)=Bp(g(1,p))⊗Bp(g(2,p))=Bp(g)⊗Bp(g),
and εp−1Bp(g)=εp(g).
(2) Since μ(1K)=1∈He is a group-like element in (He,Δe,εe) by the conclusion of (1), Be(1) is also a group-like element in (He,Δe,εe). By the antipode compatibility condition of Hopf π-algebras, we have
Be(1)SeBe(1)=εeBe(1)1=1=εeBe(1)1=SeBe(1)Be(1).
From the defining identity of a Rota-Baxter Hopf π-algebra, we have
Be(1)Be(1)=Be(1Be(1)1SeBe(1))1=Be(1).
Multiplying both sides on the left by SeBe(1), we obtain
Be(1)=SeBe(1)Be(1)Be(1)=SeBe(1)Be(1)=1.
(3) For any primitive element a∈He, we have
ΔeBe(a)=Be(a)(1,e)⊗Be(a)(2,e)=Be(a(1,e))⊗Be(a(2,e))=(Be⊗Be)[(a⊗1)+(1⊗a)]=Be(a)⊗Be(1)+Be(1)⊗Be(a)=Be(a)⊗1+1⊗Be(a).
Proposition 5.10. If Hπ is a cocommutative and commutative Hopf π-algebra, and {Bp:Hp→Hp−1}p∈π is a family of coalgebra homomorphisms. Then RHπ is a Rota-Baxter Hopf π-algebra of weight 1 if and only if Bpq∘mp,q=mp−1,q−1∘(Bp⊗Bq) for any p,q∈π.
Proof. For any p,q∈π, h∈Hp, g∈Hq, we have
Bqph(1,p)Bp(h(2,p))gSp−1Bp(h(3,p))=Bqph(1,p)Bp(h(2,p))Sp−1Bp(h(3,p))g=Bqph(1,p)Bp(h(21,p))Sp−1Bp(h(22,p))g=Bqph(1,p)Bp(h(2,p))(1,p−1)Sp−1Bp(h(2,p))(2,p−1)g=Bqph(1,p)εp−1Bp(h(2,p))g=Bqph(1,p)εp(h(2,p))g=Bqp(hg)=Bpq(hg).
Proposition 5.11. Suppose RHπ is a Rota-Baxter Hopf π-algebra of weight 1 and {φp}p∈π is a Hopf π-algebra automorphism or anti-automorphism of Hπ. Let {Bp′=φp−1∘Bp∘φp}p∈π. Then (Hπ,{Bp′}p∈π) is also a Rota-Baxter Hopf π-algebra of weight 1.
Proof. When {φp}p∈π is an automorphism, we have
Bp′(h)Bq′(g)=φp−1Bpφp−1(h)φq−1Bqφq−1(g)=φp−1q−1(Bpφp−1(h)Bqφq−1(g))=φp−1q−1Bqp(φp−1(h)(1,p)Bp(φp−1(h)(2,p))φq−1(g)Sp−1Bp(φp−1(h)(3,p)))=φp−1q−1Bqp(φp−1(h(1,p))Bp(φp−1(h(2,p)))φq−1(g)Sp−1Bp(φp−1(h(3,p))))=φq−1p−1Bqp(φp−1(h(1,p))Bpφp−1(h(2,p))φq−1(g)Sp−1Bpφp−1(h(3,p)))=φq−1p−1Bqp(φp−1(h(1,p))Bpφp−1(h(2,p))φq−1(g)φp−1φpSp−1Bpφp−1(h(3,p)))=φq−1p−1Bqp(φp−1(h(1,p))Bpφp−1(h(2,p))φq−1(g)φp−1Sp−1φp−1Bpφp−1(h(3,p)))=φp−1q−1Bqpφq−1p−1(h(1,p))φp−1Bpφp−1(h(2,p))gSp−1φp−1Bpφp−1(h(3,p))=Bqp′(h(1,p))Bp′(h(2,p))gSp−1Bp′(h(3,p)).
The case of anti-automorphism can be proved similarly. □
Furthermore, Rota–Baxter operators can also induce new graded multiplications and antipode families, giving rise to a class of derived Hopf π-algebras.
Theorem 5.12. Let RHπ be a Rota–Baxter Hopf π-algebra. Define two families of linear maps
{mp,q∗:Hp⊗Hq→Hqp, mp,q∗(h⊗g)=h∗p,qg=h(1,p)Bp(h(2,p))gSp−1(Bp(h(3,p)))}p,q∈π and
{Tp:Hp→Hp−1, Tp(h)=Sp−1(Bp(h(1,p)))Sp(h(2,p))Bp(h(3,p))}p∈π. Then the following statements hold:
(1) For any p,q∈π, h∈Hp and g∈Hq, the following identities hold:
Δqp(h∗p,qg)=(h(1,p)∗p,qg(1,q))⊗(h(2,p)∗p,qg(2,q)),εqp(h∗p,qg)=εp(h)εq(g). (2) For any p∈π and h∈Hp, the following identity holds:
εp(h)1=Bp(h(1,p))Bp−1(Tp(h(2,p))). (3) {Hp}p∈π,{mp,q∗}p,q∈π,μ is a πop-algebra.
(4) {Hp,Δp,εp}p∈π,{mp,q∗}p,q∈π,μ,{Tp}p∈π is a cocommutative Hopf πop-algebra, called the descendant Hopf π-algebra of Hπ with respect to {Bp}p∈π.
(5) {Bp}p∈π is a generalized graded Hopf algebra homomorphism from {Hp,Δp,εp}p∈π,{mp,q∗}p,q∈π,μ,{Tp}p∈π to Hπ.
(6) {Hp,Δp,εp}p∈π,{mp,q∗}p,q∈π,μ,{Tp}p∈π,{Bp}p∈π is a Rota–Baxter Hopf π-algebra.
Proof. (1)For any p,q∈π, h∈Hp, g∈Hq, we have
Δqp(h∗p,qg)=Δqp(h(1,p)Bp(h(2,p))gSp−1(Bp(h(3,p))))=(h(11,p)Bp(h(21,p))g(1,q)Sp−1(Bp(h(31,p))))⊗(h(12,p)Bp(h(22,p))g(2,q)Sp−1(Bp(h(32,p))))=(h(11,p)Bp(h(12,p))g(1,q)Sp−1(Bp(h(13,p))))⊗(h(21,p)Bp(h(22,p))g(2,q)Sp−1(Bp(h(23,p))))=(h(1,p)∗p,qg(1,q))⊗(h(2,p)∗p,qg(2,q))
and
εqp(h∗p,qg)=εqp(h(1,p)Bp(h(2,p))gSp−1(Bp(h(3,p))))=εp(h(1,p))εp−1(Bp(h(2,p)))εq(g)εpSp−1(Bp(h(3,p)))=εp(h(1,p))εp(h(2,p))εq(g)εp(h(3,p))=εp(h)εq(g).
(2)For any p∈π and h∈Hp, we have
Bp(h(1,p))Bp−1(Tp(h(2,p)))=Be(h(11,p)Bp(h(12,p))Tp(h(2,p))Sp−1(Bp(h(3,p))))=Be(h(11,p)Bp(h(12,p))Sp−1(Bp(h(21,p)))Sp(h(22,p)))Bp(h(23,p))Sp−1(Bp(h(3,p)))=Be(h(1,p)Bp(h(21,p))Sp−1(Bp(h(22,p)))Sp(h(3,p))Bp(h(41,p))Sp−1(Bp(h(42,p))))=Be(h(1,p)Bp(h(2,p))(1p−1)Sp−1(Bp(h(2,p))(2p−1))Sp(h(3,p))Bp(h(4,p))(1p−1)Sp−1(Bp(h(4,p))(2p−1)))=Be(h(1,p)εp−1Bp(h(2,p))Sp(h(3,p))εp−1Bp(h(4,p)))=Be(h(1,p)εp(h(2,p))Sp(h(3,p))εp(h(4,p)))=Be(h(1,p)Sp(h(2,p)))=εp(h)Be(1)=εp(h)1.
(3) For any p,q∈π, h∈Hp, g∈Hq, by the original Rota–Baxter identity,
Bpq(mp,q∗(h⊗g))=Bp(h)Bq(g),
thus {Bp}p∈π preserves multiplications.For any p,q,r∈π, h∈Hp, g∈Hq, j∈Hr, we have
(mqp,r∗∘(mp,q∗⊗idHr))(h⊗g⊗j)=(h∗p,qg)∗qp,rj=(h∗p,qg)(1,qp)Bqp((h∗p,qg)(2,qp))jS(qp)−1Bqp((h∗p,qg)(3,qp))=(h(1,p)∗p,qg(1,q))Bqp(h(2,p)∗p,qg(2,q))jS(qp)−1Bqp(h(3,p)∗p,qg(3,q))=h(11,p)Bp(h(12,p))g(1,q)Sp−1Bp(h(13,p))Bp(h(2,p))Bq(g(2,q))jS(qp)−1(Bp(h(3,p))Bq(g(3,q)))=h(1,p)Bp(h(2,p))g(1,q)Sp−1Bp(h(3,p))Bp(h(4,p))Bq(g(2,q))jSq−1Bq(g(3,q))Sp−1Bp(h(5,p))=h(1,p)Bp(h(2,p))g(1,q)ε(h(3,p))Bq(g(2,q))jSq−1Bq(g(3,q))Sp−1Bp(h(4,p))=h(1,p)Bp(h(2,p))g(1,q)Bq(g(2,q))jSq−1Bq(g(3,q))Sp−1Bp(h(3,p))=h(1,p)Bp(h(2,p))(g∗q,rj)Sp−1Bp(h(3,p))=h∗p,rq(g∗q,rj)=(mp,rq∗∘(idHp⊗mq,r∗))(h⊗g⊗j)
and
me,p∗(h⊗1)=h∗p,e1=h(1,p)Bp(h(2,p))Sp−1Bp(h(3,p))=h(1,p)εp−1Bp(h(2,p))=h(1,p)εp(h(2,p))=h=1Be(1)hSeBe(1)=1∗e,ph=mp,e∗(1⊗h).
(4) For any p∈π, h∈Hp,
(mp,p−1∗∘(idHp⊗Tp)∘Δp)(h)=h(1,p)∗p,p−1Tp(h(2,p))=h(11,p)Bp(h(12,p))Tp(h(2,p))Sp−1Bp(h(13,p))=h(11,p)Bp(h(12,p))Sp−1Bp(h(21,p))Sp(h(22,p))Bp(h(23,p))Sp−1Bp(h(13,p))=h(1,p)Bp(h(2,p))Sp−1Bp(h(3,p))Sp(h(4,p))Bp(h(5,p))Sp−1Bp(h(6,p))=h(1,p)εp−1Bp(h(2,p))Sp(h(3,p))εp−1Bp(h(4,p))=h(1,p)εp(h(2,p))Sp(h(3,p))εp(h(4,p))=h(1,p)Sp(h(2,p))=εp(h)1=(μ∘εp)(h).
By
Bp(h(1,p))Bp−1(Tp(h(2,p)))=Be(h(1,p)∗p,p−1Tp(h(2,p)))=Be(εp(h)1)=εp(h)1,
we have
Bp−1(Tp(h(1,p)))Bp(h(2,p))=Bp−1Tp(h(1,p))εp−1Bp(h(2,p))Bp(h(3,p))=εp−1Bp(h(1,p))Bp−1Tp(h(2,p))Bp(h(3,p))=Sp−1Bp(h(1,p))Bp(h(2,p))Bp−1Tp(h(3,p))Bp(h(4,p))=Sp−1Bp(h(1,p))εp(h(2,p))Bp(h(3,p))=εp(h)1.
(mp−1,p∗∘(Tp⊗idHp)∘Δp)(h)=Tp(h(1,p))∗p−1,ph(2,p)=Tp(h(1,p))Bp−1Tp(h(2,p))h(3,p)SpBp−1Tp(h(4,p))=Sp−1Bp(h(1,p))Sp(h(2,p))Bp(h(3,p))Bp−1Tp(h(4,p))h(5,p)SpBp−1Tp(h(6,p))=Sp−1Bp(h(1,p))Sp(h(2,p))εp(h(3,p))h(4,p)SpBp−1Tp(h(5,p))=Sp−1Bp(h(1,p))Sp(h(2,p))h(3,p)SpBp−1Tp(h(4,p))=Sp−1Bp(h(1,p))SpBp−1Tp(h(2,p))=Se(Bp−1Tp(h(2,p))Bp(h(1,p)))=εp(h)1=(μ∘εp)(h).
Therefore {Hp,Δp,εp}p∈π,{mp,q∗}p,q∈π,μ,{Tp}p∈π is a Hopf π-coalgebra.
(5) Clearly, Φ:πop→π given by p↦p−1 is a group homomorphism. For any p,q∈π, h∈Hp and g∈Hq, we have
(mΦ(p),Φ(q)∘(Bp⊗Bq))(h⊗g)=Bp(h)Bq(g)=Bpq(h∗p,qg)=(Bpq∘mp,q∗)(h⊗g),
which means that {Bp}p∈π is a generalized graded algebra homomorphism from ({Hp}p∈π,{mp,q∗}p,q∈π,μ) to Hπ.
Combining with Corollary 1, the proposition is proved.
For any p,q∈π, h∈Hp and g∈Hq,
Bqp(h(1,p)∗Bp(h(2,p))∗g∗Sp−1Bp(h(3,p)))=Bp(h(1,p))Bp−1(Bp(h(2,p)))Bq(g)BpSp−1Bp(h(3,p))=Bp(h(1,p))Bp−1Bp(h(2,p))Bq(g)SpBp−1Bp(h(3,p))=Bp(h)∗p−1,q−1Bq(g)(1)
As a complementary result, the image set of the Rota–Baxter operators constitutes a Hopf π-subalgebra of the original Hopf π-algebra.
Proposition 5.13. Let RHπ be a Rota–Baxter Hopf π-algebra, and set Gp=ImBp−1⊆Hp. Then ({Gp}p∈π,{Δp,εp}p∈π,{mp,q}p,q∈π,μ,{Sp}p∈π) forms a Hopf π-subalgebra of Hπ.
Proof. For any p,q∈π, Gp⋅Gq=(Bp−1(Hp−1)⋅Bq−1(Hq−1))⊆Bq−1p−1(Hq−1p−1)=B(pq)−1(H(pq)−1)=Gpq. By Proposition 1, Be(1)=1∈Ge. Hence ({Gp}p∈π,{mp,q}p,q∈π,μ) is a π-subalgebra of Hπ.
By the coalgebra homomorphism property of Bp−1, Δp(Gp)=Δp(Bp−1(Hp−1))=(Bp−1⊗Bp−1)Δp−1(Hp−1)⊆Gp⊗Gp, and εp(Gp)=εp−1(Hp−1)⊆K. Thus {Gp} is a π-subcoalgebra.
By Theorem 6 (5), Bp∘Tp−1=Sp∘Bp−1 for any p∈π. Therefore Sp(Gp)=Sp(Bp−1(Hp−1))=Bp(Tp−1(Hp−1))⊆ImBp=Gp−1.
Conclusion
This paper presents a systematic generalization of the theory of Rota–Baxter operators from classical cocommutative Hopf algebras to the group-graded setting of Hopf π-algebras. The development proceeds in three stages: construction of graded convolution algebras, derivation of core structural properties of Hopf π-algebras, and establishment of the graded Rota–Baxter operator framework.
First, as the central technical tool of the paper, we introduce the notion of a convolution π-algebra induced by a coassociative counital coalgebra and a π-algebra, and verify that it satisfies the axioms of a graded algebra. We further show that coalgebra homomorphisms pull back to graded homomorphisms of convolution π-algebras, and that generalized graded homomorphisms of π-algebras push forward to generalized graded homomorphisms of the corresponding convolution π-algebras. These results provide a unified convolution calculus for the study of Hopf π-algebras.
Second, using convolution π-algebras and the notion of convolution inverse, we derive a series of fundamental properties of Hopf π-algebras. We give explicit formulas for the convolution inverses of two canonical classes of maps, and deduce the antipode compatibility criterion for generalized homomorphisms between Hopf π-algebras. We prove that the antipode of a Hopf π-algebra is both an anti-homomorphism of graded algebras and an anti-homomorphism of coalgebras. For cocommutative Hopf π-algebras, we establish the bijectivity of the antipode and show that the co-opposite graded structure again admits a Hopf π-algebra structure. These results complement and refine the foundational theory of Hopf π-algebras.
Third, based on the antipode and homomorphism properties established above, we develop the axiomatic theory of Rota–Baxter operators on cocommutative Hopf π-algebras, and introduce two families of graded Rota–Baxter operators, of weight 1 and weight −1 respectively. We prove that the two families are in bijective correspondence via the antipode, so that all structural properties transfer between them. We then construct several explicit examples, including the trivial group-graded Hopf π-algebra, crossed product Hopf π-algebras with non-abelian index group S3, and crossed product Hopf π-algebras with abelian index group Z/2Z; the last example yields three distinct families of Rota–Baxter operators, demonstrating the richness of the theory. We also investigate basic properties of the operators: preservation of canonical structural elements such as group-like elements and primitive elements, an equivalent characterization in the commutative and cocommutative case, and invariance under automorphisms and anti-automorphisms of the underlying Hopf π-algebra. Finally, we describe two derived structures induced by a Rota–Baxter operator: the descendant Hopf πop-algebra and the image Hopf π-subalgebra.
This work extends the classical correspondence between Rota–Baxter operators and Hopf algebras to the group-graded setting, and enriches the repertoire of operator structures on Hopf π-algebras. Building on the results of this paper, we are currently generalizing the framework to multiplier Hopf π-algebras. We expect that future research will elucidate the intrinsic relation between graded Rota–Baxter structures and graded Yang–Baxter equations, extending the scope of the classical results.
Funding
This work was supported by the National Natural Science Foundation of China (Grant No. 12505111).
Acknowledgments
The authors appreciate the forthcoming valuable comments and suggestions from the journal editor and anonymous referee, which will further refine the theoretical content and presentation of this article.
Author Contributions
All authors jointly completed the theoretical derivation and example construction of Rota–Baxter Hopf π-algebras. G.S. proposed the research idea and guided the whole research work. X.T. and X.W. finished the main theorem proofs and drafted the original manuscript. Y.Z. sorted out the references and participated in the revision of the full text. All authors have read and agreed to the final published version of the manuscript.
Data Availability Statement
No new data were created.
Conflict of Interest
The authors declare no conflicts of interest.
References
Faculty of Mathematics and Physics, Huai’an University, Huai’an 223003, Jiangsu, China
Faculty of Mathematics and Physics, Huai’an University, Huai’an 223003, Jiangsu, China
Faculty of Mathematics and Physics, Huai’an University, Huai’an 223003, Jiangsu, China
Faculty of Mathematics and Physics, Huai’an University, Huai’an 223003, Jiangsu, China
Email address: guodongs@hau.edu.cn