As a development of [2] and [3], we construct a "VN-bialgebra" in Vect_k for each k-linear split-semigroupal functor from a suitable monoidal category C to Vect_k. The main aim here is to avoid the customary compactness assumptions on generators of the domain category C (cf. [3]). Please note that the VN-bialgebras in Vect_k defined here are not necessarily von Neumann regular as k-algebras in the usual sense, the prefix "VN-" coming from the Set-based case of von Neumann regular semigroups.
We propose the construction of a VN-core associated to each (k-linear) split semigroupal functor U from a suitable monoidal category C to Vect k , where all our categories, functors, and natural transformations are assumed to be klinear, for a fixed field k. Essentially, the category C must be equipped with a small "U -generator" A carrying some extra duality information and with U A still being finite dimensional for all A in A.
We shall use the term “VN-core” (in Vect k ) to mean a (usual) k-semibialgebra E together with a k-linear endomorphism S such that
The VN-core is called “antipodal” if S(xy) = SySx (and S(1) = 1) for all x, y ∈ E. This minimal type of structure is introduced here in order to avoid compactness assumptions on the generator A ⊂ C and, at the same time, retain the “fusion” operator, namely
satisfying the usual fusion equation [7]. Note that here the fusion operator always has a partial inverse (see [1]).
In §2 we establish sufficient conditions on a functor U in order that
be a VN-core in Vect k (following [2]). This core can be completed to a VN-core End ∨ U ⊕ k with a unit element. In §3 we give several examples of suitable functors U for the theory.
2 The construction of End ∨ U Let C = (C, ⊗, I) be a monoidal category and let U : C → Vect be a functor with both a semigroupal structure, denoted
and a cosemigroupal structure, denoted
We shall suppose also that there exists a small full subcategory A of C with the properties:
UeA y y r r r r r r r r r r r r and
UeA y y r r r r r r r r r r r r where e UA = 1 ⊗ ev in Vect, and r 3 i 3 = 1.
We now define the semibialgebra structure (End ∨ U, µ, δ) on
as in [2] §2, with the isomorphism of k-linear spaces
given (as in [2] §3) by the usual components
where d is the canonical map from a vector space to its double dual. Furthermore, each map
commutes, and
Then we obtain:
Proof. The von Neumann axiom
becomes the diagram (in which we have omitted “⊗”): where ( * ) is the exterior of the diagram
which commutes using (E2) and commutativity of
The first type of example is derived from the idea of a (contravariant) involution on a (small) comonoidal category D. This includes the doubles D = B op + B and D = B op ⊗ B with their respective “switch” maps (where B is a given small comonoidal Vect k -category), or any small comonoidal and compact-monoidal Vect k -category D (such as the category Mat k of finite matrices over k) with the tensor duals of objects now providing an antipode on the comonoidal aspect of the structure rather than on the monoidal part, or any * -algebra structure on a given k-bialgebra (e.g., a C * -bialgebra) with the * -operation providing the antipode.
In each case, an even functor from D to Vect is defined to be a (k-linear) functor F equipped with a (chosen) dinatural isomorphism
If we take the morphisms of even functors to be all the natural transformations between them then we obtain a category
Let A = E(D, Vect fd ) fs be the full subcategory of E consisting of the finitely valued functors of finite support. While this category is generally not compact, it has on it a natural antipode derived from those on D and Vect fd , namely
Of course, there are also examples where A is actually compact, such as those where D is a Hopf algebroid, in the sense of [4], with antipode (-) * = S, in which case each A from D to Vect has a symmetry structure on it. Now let C be the full subcategory of E consisting of the small coproducts in E of objects from A. This category C is easily seen to be monoidal under the pointwise convolution structure from D, and the inclusion A ⊂ C is U -dense for the functor
which is split semigroupal with U A finite dimensional for all A ∈ A. Moreover,
for all A ∈ A. The conditions of ( 5) are easily verified if we define maps
by commutativity of the diagrams
where the exterior of
is a genuine map in C when C is given the pointwise monoidal structure from D. This completes the details of the general example.
In the case where k = C and D has just one object D whose endomorphism algebra is a C * -bialgebra, we have a one-object comonoidal category D with a C-conjugate-linear antipode given by the * -operation. Then the convolution [D, Hilb fd ], where [D,
is a monoidal category, with a C-linear antipode given by
where H • denotes the conjugate-transpose of H ∈ Hilb fd . We now interpret an even functor F to be a functor equipped with a dinatural isomorphism
Let V = (V, ⊗, I) be a (small) braided monoidal category and let B be the klinearization of Semicoalg(V) with the monoidal structure induced from that on V. By analogy with [5], let X ⊂ B be a finite full subcategory of B with I ∈ X and X op promonoidal when
for x, y, z ∈ X . For example (cf. [5]), one could take X to be a (finite) set of non-isomorphic “basic” objects in some braided monoidal category V, where each x ∈ X has a coassociative diagonal map δ : x → x ⊗ x. However, we won’t need the category X to be discrete o
This content is AI-processed based on open access ArXiv data.