---
title: Asymptotically Adjoint Good Endofunctors
url: https://www.emergentmind.com/topics/asymptotically-adjoint-good-endofunctors
type: topic
---

# Asymptotically Adjoint Good Endofunctors

Searching arXiv for recent and foundational papers on asymptotically adjoint good endofunctors and related adjunction frameworks.
{"query":"all:(\"asymptotically adjoint good endofunctors\" OR \"Scalability and asymptotic adjunction\" OR \"Representations are adjoint to endomorphisms\" OR \"adjoint functor theorems for infinity-categories\" OR \"Monadic Tower for infinity-Categories\")","max_results":10,"sort_by":"relevance"}
{"query":"\"On generalized morphisms associated to endofunctors of C*-algebras\"","max_results":5,"sort_by":"submittedDate"}
Asymptotically adjoint good endofunctors are a weakened form of adjoint pair developed for endofunctors of \(C^*\)-algebras when strict adjunction is unavailable but a stabilized and asymptotic substitute remains. In this framework, one works not with ordinary hom-sets but with commutative monoids of generalized morphisms built from a class of “good endofunctors,” together with stabilization by compact operators and iterated application of the asymptotic algebra functor. The central point is that a pair \(S,N\) can fail to be adjoint in the classical sense and nevertheless induce natural isomorphisms on asymptotic generalized morphism sets, which is sufficient for applications to Connes–Higson \(E\)-theory, extension theory, and \(K\)-homology [2509.02001; 2510.07883].

## 1. Origin of the notion

The immediate motivation comes from operator \(C^*\)-algebra theory, where adjunctions organize duality and universal properties, but natural functors such as the suspension functor \(S\) do not admit genuine adjoints in the homotopy category of asymptotic homomorphisms \(hAsy\) [2510.07883]. This obstruction leads to the introduction of asymptotic adjunction, a weaker notion that preserves an adjunction-like isomorphism only after stabilization and passage to asymptotic colimits [2509.02001].

The notion was introduced in a broader program that studies endofunctors of \(C^*\)-algebras through generalized morphisms, homotopies of natural transformations, and bimonoidal structure on the class of good endofunctors [2509.02001]. A subsequent geometric realization shows that for every pair of scalable proper metric spaces, the functor of continuous functions and the relative Roe functor associated with the pair are asymptotically adjoint, and that this weaker adjunction is “intuitive and useful in applications” despite being weaker than a genuine adjunction [2510.07883].

A common misconception is to treat asymptotic adjunction as merely a notational reformulation of an ordinary adjunction. The defining data and the resulting isomorphisms show otherwise: the counit lands in the stabilized asymptotic algebra functor rather than the identity, and the triangle identities are replaced by commutativity up to homotopy [2509.02001; 2510.07883].

## 2. Good endofunctors and generalized morphisms

The ambient setting is the category of \(C^*\)-algebras equipped with a special class of endofunctors called good endofunctors. These are described as a class of functors endowed with labelings and compatibilities sufficient to support homotopy-theoretic manipulations, generalized morphism sets, and a bimonoidal structure [2509.02001]. Major examples listed in the literature include the asymptotic algebra functor \(\mathfrak{A}\), the stabilization functor \(K\), functors of continuous functions \(C_{\mathsf{X}}\), tensor-type functors, and relative Roe functors [2509.02001; 2510.07883].

For a pair of \(C^*\)-algebras \(A,B\) and a good endofunctor \(F\), the basic generalized morphism set is
\[
[A,F,B] := \operatorname{Hom}(A,F(B))/\simeq_F,
\]
where \(\simeq_F\) denotes \(F\)-homotopy [2509.02001]. The asymptotic refinement is
\[
[[A,F,B]] := \varinjlim_n [A,F\mathfrak{A}^n \mathbb{K},B],
\]
with transition maps induced by the natural transformation \(\alpha\) into \(\mathfrak{A}\) and stabilization [2510.07883]. The paper on generalized morphisms states that these constructions produce commutative monoids and that the collection of good endofunctors carries a tight bimonoidal structure; it also defines a bilinear associative product
\[
\bullet: [B,GK,C] \times [A,FK,B] \to [A,FGK,C],
\]
generalizing \(E\)-theoretic composition [2509.02001].

This formalism is not auxiliary. It is the domain in which asymptotic adjunction is defined and in which its main consequence—an isomorphism of commutative monoids—holds naturally in both variables [2509.02001].

## 3. Definition of asymptotic adjunction

In the homotopy category of good endofunctors, denoted \(\mathrm{hGEFC}\), an asymptotic adjunction between good endofunctors \(S\) and \(N\) consists of a unit and a counit of the form
\[
\eta: \operatorname{Id} \Rightarrow NS,
\qquad
\varepsilon: SN \Rightarrow \mathfrak{A}K,
\]
subject to homotopy-commutative triangle diagrams [2509.02001]. Equivalently, in the notation used for good endofunctors in the geometric paper, good endofunctors \(S,N\) are asymptotically adjoint if there exist labeled natural transformations
\[
\eta: \operatorname{Id} \Rightarrow NS,
\qquad
\varepsilon: SN \Rightarrow \mathfrak{A}\mathbb{K},
\]
such that
\[
S \xrightarrow{S\eta} SNS \xrightarrow{\varepsilon S} \mathfrak{A}KS
\]
is homotopic to \(\alpha\iota_{00}S\), and
\[
N \xrightarrow{\eta N} NSN \xrightarrow{N\varepsilon} N\mathfrak{A}K
\]
is homotopic to \(N\alpha\iota_{00}\) [2510.07883].

Two features distinguish this from ordinary adjunction. First, the target of the counit is \(\mathfrak{A}K\), not \(\operatorname{Id}\). Second, the triangle identities hold only in the homotopy category of natural transformations [2510.07883]. The literature is explicit that exact inverses in a strict categorical sense do not exist here; the useful inverse statements arise only after passing to stabilized and asymptotic colimits [2510.07883].

This suggests that asymptotic adjunction is best understood as an adjunction internal to a stabilized asymptotic calculus rather than as a defectively implemented classical adjunction.

## 4. Fundamental categorical consequence

The main theorem states that an asymptotic adjunction \(S \mathop{as} N\) yields mutually inverse natural isomorphisms of commutative monoids [2509.02001]:
\[
\Phi: [[SA,\operatorname{Id},B]] \to [[A,N,B]],
\qquad
\Psi: [[A,N,B]] \to [[SA,\operatorname{Id},B]].
\]
In the geometric formulation, this is written as
\[
[[LA,\operatorname{Id},B]] \cong [[A,R,B]]
\]
for asymptotically adjoint good endofunctors \(L\) and \(R\), naturally in \(A\) and \(B\) [2510.07883]. The theorem is the asymptotic analogue of the classical hom-set isomorphism characterizing an adjunction.

The formulas for the inverse maps are explicit. If \(\eta\) and \(\varepsilon\) are the unit and counit, then \(\Phi\) is defined by composing a representative
\[
SA \xrightarrow{\varphi} \mathfrak{A}^n K B
\]
with \(\eta A: A \to NSA\) and then applying \(N\varphi\). The inverse \(\Psi\) starts with
\[
A \xrightarrow{\psi} N\mathfrak{A}^n K B,
\]
applies \(S\psi\), then the counit, and finally passes to one further asymptotic stage \(\mathfrak{A}^{n+1}K B\) [2509.02001].

This is the exact sense in which asymptotic adjunction “retains sufficient categorical properties.” It does not provide a strict hom-set bijection in the original category of \(C^*\)-algebras, but it does provide a natural isomorphism after stabilization and asymptotization [2510.07883]. A plausible implication is that the notion is tailored to bivariant theories whose composition laws already live in such colimit-stabilized environments.

## 5. Geometric realization: continuous functions and relative Roe functors

A principal example is attached to a pair \((\mathsf{X},\mathsf{X}_0)\) of proper metric spaces with \(\mathsf{X}_0\) closed. The continuous functions functor is
\[
C_{\mathsf{X},\mathsf{X}_0}(B):=C_0(\mathsf{X},\mathsf{X}_0,B)
=\{f\in C_0(\mathsf{X},B): f|_{\mathsf{X}_0}=0\},
\]
and it is described as a tensor-type good endofunctor because
\[
C_{\mathsf{X},\mathsf{X}_0}(A)\cong C_0(\mathsf{X},\mathsf{X}_0)\otimes A
\]
for any \(C^*\)-algebra \(A\) [2510.07883].

For a \(\Delta\)-discretization \(X\subset \mathsf{X}\) with
\[
X_0:=\{x\in X:\mathrm{dist}(x,\mathsf{X}_0)<\Delta\},
\]
one forms the uniform Roe algebra \(\mathfrak{M}^u_X(B)\), the ideal \(\mathfrak{M}^u_{X\supset X_0}(B)\) of matrices supported near \(X_0\), and the relative uniform Roe functor
\[
\mathfrak{N}^u_{X,X_0}(B):=
\mathfrak{M}^u_X(B)/\mathfrak{M}^u_{X\supset X_0}(B),
\]
which is proved to be a good endofunctor [2510.07883].

If the pair \((\mathsf{X},\mathsf{X}_0)\) is scalable, then the main theorem of the geometric paper states
\[
C_{\mathsf{X},\mathsf{X}_0} \asymp \mathfrak{N}^u_{X,X_0},
\]
for any coarse discretization \((X,X_0)\) [2510.07883]. The unit and counit are constructed explicitly using the partition of unity coming from the discretization and the scaling map, and the resulting asymptotic adjunction yields a natural isomorphism of monoids
\[
[[C_{\mathsf{X},\mathsf{X}_0}A,\operatorname{Id},B]]
\cong
[[A,\mathfrak{N}^u_{X,X_0},B]]
\]
for all \(C^*\)-algebras \(A,B\) [2510.07883].

This example is the clearest realization of the abstract notion. It translates between a topological functor built from vanishing-at-infinity continuous functions and a coarse-geometric functor built from relative Roe algebras, with the bridge provided not by a strict adjoint pair but by asymptotic adjunction.

## 6. Applications and relation to broader adjointness theories

The principal applications are to Connes–Higson \(E\)-theory, extension theory, and \(K\)-homology. The generalized morphism framework was designed to recover and extend “Kasparov-type” bivariant \(K\)-theories, and asymptotic adjunction provides a new route to identifying their morphism monoids [2509.02001]. In particular, the literature states that for suspension-type situations one obtains an alternative, “\(KK\)-like,” model of \(E\)-theory, and more specifically that
\[
E_0(A,B) = [[SA,\mathfrak{A}K,B]] \cong [[A,N,B]]
\]
when \(N\) is asymptotically right adjoint to suspension in the required sense [2509.02001].

The geometric paper makes these applications concrete. For suspension pairs \((\mathbb{R}^n,\varnothing)\) and their lattice discretizations, the asymptotic adjunction yields unsuspended descriptions
\[
E_0(A,B) \cong [[A,\mathfrak{M}^u_{\mathbb{Z}^2},B]],
\qquad
E_1(A,B) \cong [[A,\mathfrak{M}^u_{\mathbb{Z}},B]],
\]
so that the right-hand sides do not involve suspensions of \(A\) [2510.07883]. It also gives
\[
E_1(A,B) \cong [[SA,\mathbb{K}B]]
\cong [[A,\mathfrak{N}^u_{\mathbb{Z}_+,\{0\}},B]],
\]
linking \(E\)-theory to “extensions with asymptotic coefficients,” and
\[
K_1(\mathsf{X}) \cong [[\mathbb{C},\mathfrak{N}^u_{(\mathcal{O}\mathsf{X})^{discr},\{0\},\mathbb{K}}]]
\]
for cones over compact subsets of a Hilbert sphere, with a generalization to any compact metric space by embedding into a Hilbert sphere [2510.07883].

Within the broader literature on adjointness, asymptotic adjunction occupies a specific position among several distinct responses to the failure or decomposition of strict adjoints. In enriched category theory, the representation functor \(Rep\) from monoids in a monoidal category \(V\) to categories over an enriched category \(C\) is genuinely adjoint to the endomorphism monoid functor \(E\), with
\[
\hom_{Mon(V)}(M,EF)\cong \hom_{Cat/C}(Rep(M),F),
\]
and the same pattern extends to operads and props via enrichments in symmetric and bisymmetric sequences [1904.06987]. In \(\infty\)-category theory, general adjoint functor theorems characterize when a functor admits an adjoint in terms of continuity and solution-set or \(h\)-initial conditions, and for presentable \(\infty\)-categories right adjoints admit canonical decompositions as a coreflection followed by possibly transfinitely many monadic functors [1803.01664; 2104.01816]. In the 2-categorical setting of lax-idempotent pseudomonads, adjointness can be decomposed into \(R\)-cocontinuity and relative \(IR\)-admissibility [2306.10389]. By contrast, the neat reduct functor for representable cylindric algebras provides a sharp non-example: it has no right adjoint, whereas the analogous functor for polyadic algebras is an equivalence [1304.0714].

Taken together, these results show that asymptotically adjoint good endofunctors belong to a larger landscape in which adjointness may be strict, relative, decomposed, or obstructed. Their distinctive feature is that the replacement for strict adjunction is formulated directly in the stabilized asymptotic language appropriate to \(C^*\)-algebraic bivariant theories.

Source: https://www.emergentmind.com/topics/asymptotically-adjoint-good-endofunctors