Representation theory of Hopf algebras

In abstract algebra, a representation of a Hopf algebra is a representation of its underlying associative algebra. That is, a representation of a Hopf algebra H over a field K is a K-vector space V with an action H × V → V usually denoted by juxtaposition ( that is, the image of (h,v) is written hv ). The vector space V is called an H-module.

Properties

The module structure of a representation of a Hopf algebra H is simply its structure as a module for the underlying associative algebra. The main use of considering the additional structure of a Hopf algebra is when considering all H-modules as a category. The additional structure is also used to define invariant elements of an H-module V. An element v in V is invariant under H if for all h in H, hv = ε(h)v, where ε is the counit of H. The subset of all invariant elements of V forms a submodule of V.

Categories of representations as a motivation for Hopf algebras

For an associative algebra H, the tensor product V₁ ⊗ V₂ of two H-modules V₁ and V₂ is a vector space, but not necessarily an H-module. For the tensor product to be a functorial product operation on H-modules, there must be a linear binary operation Δ : H → H ⊗ H such that for any v in V₁ ⊗ V₂ and any h in H,

$hv=\Delta h(v_{(1)}\otimes v_{(2)})=h_{(1)}v_{(1)}\otimes h_{(2)}v_{(2)},$

and for any v in V₁ ⊗ V₂ and a and b in H,

$\Delta (ab)(v_{(1)}\otimes v_{(2)})=(ab)v=a[b[v]]=\Delta a[\Delta b(v_{(1)}\otimes v_{(2)})]=(\Delta a)(\Delta b)(v_{(1)}\otimes v_{(2)}).$

using sumless Sweedler's notation, which is somewhat like an index free form of Einstein's summation convention. This is satisfied if there is a Δ such that Δ(ab) = Δ(a)Δ(b) for all a, b in H.

For the category of H-modules to be a strict monoidal category with respect to ⊗, $V_{1}\otimes (V_{2}\otimes V_{3})$ and $(V_{1}\otimes V_{2})\otimes V_{3}$ must be equivalent and there must be unit object ε_H, called the trivial module, such that ε_H ⊗ V, V and V ⊗ ε_H are equivalent.

This means that for any v in

$V_{1}\otimes (V_{2}\otimes V_{3})=(V_{1}\otimes V_{2})\otimes V_{3}$

and for h in H,

$((\operatorname {id} \otimes \Delta )\Delta h)(v_{(1)}\otimes v_{(2)}\otimes v_{(3)})=h_{(1)}v_{(1)}\otimes h_{(2)(1)}v_{(2)}\otimes h_{(2)(2)}v_{(3)}=hv=((\Delta \otimes \operatorname {id} )\Delta h)(v_{(1)}\otimes v_{(2)}\otimes v_{(3)}).$

This will hold for any three H-modules if Δ satisfies

$(\operatorname {id} \otimes \Delta )\Delta A=(\Delta \otimes \operatorname {id} )\Delta A.$

The trivial module must be one-dimensional, and so an algebra homomorphism ε : H → F may be defined such that hv = ε(h)v for all v in ε_H. The trivial module may be identified with F, with 1 being the element such that 1 ⊗ v = v = v ⊗ 1 for all v. It follows that for any v in any H-module V, any c in ε_H and any h in H,

$(\varepsilon (h_{(1)})h_{(2)})cv=h_{(1)}c\otimes h_{(2)}v=h(c\otimes v)=h(cv)=(h_{(1)}\varepsilon (h_{(2)}))cv.$

The existence of an algebra homomorphism ε satisfying

$\varepsilon (h_{(1)})h_{(2)}=h=h_{(1)}\varepsilon (h_{(2)})$

is a sufficient condition for the existence of the trivial module.

It follows that in order for the category of H-modules to be a monoidal category with respect to the tensor product, it is sufficient for H to have maps Δ and ε satisfying these conditions. This is the motivation for the definition of a bialgebra, where Δ is called the comultiplication and ε is called the counit.

In order for each H-module V to have a dual representation V such that the underlying vector spaces are dual and the operation * is functorial over the monoidal category of H-modules, there must be a linear map S : H → H such that for any h in H, x in V and y in V*,

$\langle y,S(h)x\rangle =\langle hy,x\rangle .$

where $\langle \cdot ,\cdot \rangle$ is the usual pairing of dual vector spaces. If the map $\varphi :V\otimes V^{*}\rightarrow \varepsilon _{H}$ induced by the pairing is to be an H-homomorphism, then for any h in H, x in V and y in V*,

$\varphi \left(h(x\otimes y)\right)=\varphi \left(x\otimes S(h_{(1)})h_{(2)}y\right)=\varphi \left(S(h_{(2)})h_{(1)}x\otimes y\right)=h\varphi (x\otimes y)=\varepsilon (h)\varphi (x\otimes y),$

which is satisfied if

$S(h_{(1)})h_{(2)}=\varepsilon (h)=h_{(1)}S(h_{(2)})$

for all h in H.

If there is such a map S, then it is called an antipode, and H is a Hopf algebra. The desire for a monoidal category of modules with functorial tensor products and dual representations is therefore one motivation for the concept of a Hopf algebra.

Representations on an algebra

A Hopf algebra also has representations which carry additional structure, namely they are algebras.

Let H be a Hopf algebra. If A is an algebra with the product operation μ : A ⊗ A → A, and ρ : H ⊗ A → A is a representation of H on A, then ρ is said to be a representation of H on an algebra if μ is H-equivariant. As special cases, Lie algebras, Lie superalgebras and groups can also have representations on an algebra.

See also

• Tannaka–Krein reconstruction theorem