📝 Original Info
- Title: On Rational Pairings of Functors
- ArXiv ID: 1003.3221
- Date: 2010-03-17
- Authors: ** Bachuki Mesablishvili, Robert Wisbauer **
📝 Abstract
In the theory of coalgebras $C$ over a ring $R$, the rational functor relates the category of modules over the algebra $C^*$ (with convolution product) with the category of comodules over $C$. It is based on the pairing of the algebra $C^*$ with the coalgebra $C$ provided by the evaluation map $\ev:C^*\ot_R C\to R$. We generalise this situation by defining a {\em pairing} between endofunctors $T$ and $G$ on any category $\A$ as a map, natural in $a,b\in \A$, $$\beta_{a,b}:\A(a, G(b)) \to \A(T(a),b),$$ and we call it {\em rational} if these all are injective. In case $\bT=(T,m_T,e_T)$ is a monad and $\bG=(G,\delta_G,\ve_G)$ is a comonad on $\A$, additional compatibility conditions are imposed on a pairing between $\bT$ and $\bG$. If such a pairing is given and is rational, and $\bT$ has a right adjoint monad $\bT^\di$, we construct a {\em rational functor} as the functor-part of an idempotent comonad on the $\bT$-modules $\A_{\rT}$ which generalises the crucial properties of the rational functor for coalgebras. As a special case we consider pairings on monoidal categories.
💡 Deep Analysis
Deep Dive into On Rational Pairings of Functors.
In the theory of coalgebras $C$ over a ring $R$, the rational functor relates the category of modules over the algebra $C^*$ (with convolution product) with the category of comodules over $C$. It is based on the pairing of the algebra $C^*$ with the coalgebra $C$ provided by the evaluation map $\ev:C^*\ot_R C\to R$. We generalise this situation by defining a {\em pairing} between endofunctors $T$ and $G$ on any category $\A$ as a map, natural in $a,b\in \A$,
$$\beta_{a,b}:\A(a, G(b)) \to \A(T(a),b),$$
and we call it {\em rational} if these all are injective. In case $\bT=(T,m_T,e_T)$ is a monad and $\bG=(G,\delta_G,\ve_G)$ is a comonad on $\A$, additional compatibility conditions are imposed on a pairing between $\bT$ and $\bG$. If such a pairing is given and is rational, and $\bT$ has a right adjoint monad $\bT^\di$, we construct a {\em rational functor} as the functor-part of an idempotent comonad on the $\bT$-modules $\A_{\rT}$ which generalises the crucial properties of the rationa
📄 Full Content
arXiv:1003.3221v1 [math.CT] 16 Mar 2010
ON RATIONAL PAIRINGS OF FUNCTORS
BACHUKI MESABLISHVILI AND ROBERT WISBAUER
Abstract. In the theory of coalgebras C over a ring R, the rational functor relates the
category of modules over the algebra C∗(with convolution product) with the category
of comodules over C. It is based on the pairing of the algebra C∗with the coalgebra C
provided by the evaluation map ev : C∗⊗R C →R.
We generalise this situation by defining a pairing between endofunctors T and G on any
category A as a map, natural in a, b ∈A,
βa,b : A(a, G(b)) →A(T(a), b),
and we call it rational if these all are injective. In case T = (T, mT , eT ) is a monad and
G = (G, δG, εG) is a comonad on A, additional compatibility conditions are imposed on
a pairing between T and G. If such a pairing is given and is rational, and T has a right
adjoint monad T⋄, we construct a rational functor as the functor-part of an idempotent
comonad on the T-modules AT which generalises the crucial properties of the rational
functor for coalgebras. As a special case we consider pairings on monoidal categories.
Contents
1.
Introduction
1
2.
Preliminaries
3
3.
Pairings of functors
5
4.
Rational functors
12
5.
Pairings in monoidal categories
15
6.
Entwinings in monoidal categories
19
References
31
1. Introduction
The pairing of a k-vector space V with its dual space V ∗= Hom(V, k) provided by the
evaluation map V ∗⊗V →k can be extended from base fields k to arbitrary base rings A.
Then it can be applied to the study of A-corings C to obtain a faithful functor from the
category of C-comodules to the category of C∗-modules.
The purpose of this paper it to
extend these results to (endo)functors on arbitrary categories. We begin by recalling some
facts from module theory.
1.1. Pairing of modules. Let C be a bimodule over a ring A and C∗= HomA(C, A) the
right dual. Then C ⊗A −and C∗⊗A −are endofunctors on the category AM of left A-modules
and the evaluation
ev : C∗⊗A C →A,
f ⊗c 7→f(c),
induces a pairing between these functors. For left A-modules X, Y , the map
αY : C ⊗A Y →AHom(C∗, Y ),
c ⊗y 7→[f 7→f(c)y],
induces the map
βX,Y : AHom(X, C ⊗A Y )
−→
AHom(X, AHom(C∗, Y )),
X
f→C ⊗A Y
7−→
X
f→C ⊗A Y
αY
−→AHom(C∗, Y ).
1
2
BACHUKI MESABLISHVILI AND ROBERT WISBAUER
Clearly βX,Y is injective for all left A-modules X, Y if and only if αY is a monomorphism
(injective) for any left A-module Y , that is, CA is locally projective (see [1], [8, 42.10]).
Now consider the situation above with some additional structure.
1.2. Pairings for corings. Let C = (C, ∆, ε) be a coring over the ring A, that is, C is an
A-bimodule with bimodule morphisms coproduct ∆: C →C ⊗A C and counit ε : C →A.
Then the right dual C∗= HomA(C, A) has a ring structure by the convolution product for
f, g ∈C∗, f ∗g = f ◦(g ⊗A IC) ◦∆(convention opposite to [8, 17.8]) with unit ε, and we
have a pairing between the comonad C ⊗A −and the monad C∗⊗A −on AM. In this case,
HomA(C∗, −) is a comonad on AM and αY considered in 1.1 induces a comonad morphism
α : C ⊗−→HomA(C∗, −). We have the commutative diagrams
(1.1)
C∗⊗A C∗⊗A C
I⊗I⊗∆/
∗⊗I
C∗⊗A C∗⊗A C ⊗A C
I⊗ev⊗I/ C∗⊗A C
ev
C
ε⊗I
o
ε
zvvvvvvvvvv
C∗⊗A C
ev
/ A
.
The Eilenberg-Moore category MHom(C∗,−) of Hom(C∗, −)-comodules is equivalent to the
category C∗M of left C∗-modules (e.g. [5, Section 3]) and thus α induces a functor
CM →MHom(C∗,−) ≃C∗M
which is fully faithful if and only if the pairing (C∗, C, ev) is rational, that is αY is monomorph
for all Y ∈AM (see [8, 19.2 and 19.3]). Moreover, α is an isomorphism if and only if the
categories CM and C∗M are equivalent and this is tantamount to CA being finitely generated
and projective.
In Section 2 we recall the notions and some basic facts on natural transformations between
endofuctors needed for our investigations.
Weakening the conditions for an adjoint pair of functors, a pairing of two functors T : A →B
and G : B →A is defined as a map βa,b : A(a, G(b)) →A(T (a), b), natural in a ∈A, b ∈B
of two functors between arbitrary categories is defined in Section 3 (see 3.1) and it is called
rational if all the βa,b are injective maps. For pairing of monads T with comonads G on a
category A, additional conditions are imposed on the defining natural transformations (see
3.2). These imply the existence of a functor ΦP : AG →AT from the G-comodules to the
T-modules (see 3.5), which is full and faithful provided the pairing is rational (see 3.7). Of
special interest is the situation that the monad T has a right adjoint T⋄and the last part of
Section 3 is dealing with this case.
Referring to these results, a rational functor RatP : AT →AT is associated with any
rational pairing in Section 4. This leads to the definition of rational T -modules and under
some additional conditions they form a coreflective subcatgeory of AT (see 4.8).
The application of the general notions of pairings to monoidal categories is outlined in
Section 5. The resulting formalism is very close to the module case considered in
…(Full text truncated)…
Reference
This content is AI-processed based on ArXiv data.