Gluing of cotorsion pairs via recollements of abelian categories

This paper constructs new cotorsion pairs in an abelian category $\mathcal{A}$ by gluing cotorsion pairs from a recollement's outer components, establishes conditions for their coincidence and completeness, and applies these results to Morita rings while introducing a specific class of recollements with enhanced homological properties.

The Gist

Imagine you are an architect trying to build a massive, complex skyscraper (let's call it Category A). But instead of building it from scratch, you have two smaller, well-understood blueprints: one for the basement (Category A') and one for the penthouse (Category A'').

Your goal is to glue these two blueprints together to create a perfect, stable structure for the whole building. In the world of mathematics, specifically in a field called Representation Theory, these "blueprints" are called Cotorsion Pairs. They are like sets of rules that tell you how to fix broken parts of the building (resolutions) and how different parts relate to each other.

The tool you use to glue these blueprints together is called a Recollement. Think of a recollement as a sophisticated set of elevators and connecting bridges that link the basement, the main building, and the penthouse. It allows you to move things up and down and see how a problem in the basement affects the penthouse.

The Problem: The Old Rules Were Too Strict

Previously, mathematicians had a rule for gluing these blueprints: "To glue them successfully, the elevator going from the penthouse down to the main building must work perfectly (be 'exact')."

However, in many real-world mathematical scenarios (like specific types of rings called Morita rings or Triangular Matrix rings), this elevator is broken or only works partially. The old rules said, "If the elevator is broken, you can't glue the blueprints." This meant many interesting buildings were left unfinished.

The Solution: A New Gluing Technique

The authors of this paper, Jinrui Yang and Yongyun Qin, invented a new, more flexible way to glue these blueprints.

1. The "Good Enough" Elevator: Instead of demanding the elevator be perfect, they found a way to glue the blueprints even if the elevator is slightly wobbly, as long as it doesn't cause a specific type of "leak" (mathematically, as long as a certain derived functor vanishes).
2. The "Monomorphism" Safety Net: They introduced a specific safety condition (called Condition P). Imagine this as a rule that says: "As long as the connection between the penthouse and the main building is a one-way street that doesn't lose any information (a monomorphism), we are safe to glue the blueprints."
   • This condition turns out to be true in many natural situations, like when you are dealing with triangular matrix rings (which look like a triangle of numbers) or Morita rings (which are like a 2x2 grid of numbers with special rules).

What They Achieved

By using this new, flexible gluing method, they were able to:

• Construct New Blueprints: They successfully created new, stable "Cotorsion Pairs" for complex rings that were previously impossible to handle.
• Prove Stability: They showed that if the basement and penthouse blueprints were "perfect" (complete and hereditary), the new whole-building blueprint is also perfect.
• Solve Specific Cases: They applied this to Morita Rings (a specific type of algebraic structure). They found that if the "glue" (a map called $\phi$ or $\psi$) is a one-way street (a monomorphism), you can build these new structures.

The "Aha!" Moment

The most exciting part is that their method works even when the old, strict rules failed.

• Old Way: "The elevator is broken? Stop. No building."
• New Way: "The elevator is broken, but the safety net (Condition P) is holding. Let's build!"

Why It Matters

This is like finding a new way to assemble furniture when the instructions say a specific screw is missing. Instead of giving up, the authors figured out how to use a different tool to make the furniture just as sturdy.

This allows mathematicians to:

• Understand the "shape" of complex algebraic systems better.
• Create new models for solving equations in these systems.
• Connect different areas of math (like module theory and ring theory) that were previously too far apart to link.

In short, they took a rigid, "all-or-nothing" rule for building mathematical structures and turned it into a flexible, "smart-gluing" technique that works in much more difficult and interesting scenarios.

Technical Summary

Here is a detailed technical summary of the paper "Gluing of cotorsion pairs via recollements of abelian categories" by Jinrui Yang and Yongyun Qin.

1. Problem Statement

The paper addresses the problem of constructing new cotorsion pairs in an abelian category $\mathcal{A}$ by "gluing" together existing cotorsion pairs from two related abelian categories, $\mathcal{A}'$ and $\mathcal{A}''$, via a recollement.

• Context: A recollement of abelian categories is a diagram involving six functors $(i_*, i^*, i_!, j_!, j^*, j_*)$ that relates $\mathcal{A}'$, $\mathcal{A}$, and $\mathcal{A}''$. It generalizes the relationship between module categories of a ring and its quotient/subring structures (e.g., triangular matrix rings, Morita rings).
• Limitation of Existing Methods: Previous literature (e.g., Hu-Zhu-Zhu, Chen) established gluing techniques for cotorsion pairs under the strict assumption that the functors $i_!$ or $i_*$ are exact. This condition is highly restrictive; for instance, $i_!$ is exact only in specific cases like comma categories or triangular matrix rings.
• Goal: The authors aim to generalize these gluing techniques by relaxing the exactness requirements on $i_!$ and $i_*$, thereby allowing the construction of cotorsion pairs in a broader class of abelian categories, including general Morita rings.

2. Methodology

The authors employ homological algebra techniques within the framework of recollements.

• Definitions of Glued Classes:
  Given cotorsion pairs $(\mathcal{U}', \mathcal{V}')$ in $\mathcal{A}'$ and $(\mathcal{U}'', \mathcal{V}'')$ in $\mathcal{A}''$, the authors define two specific subcategories in $\mathcal{A}$:

  • $\mathcal{N}_{\mathcal{V}'}^{\mathcal{V}''} = \{ X \in \mathcal{A} \mid i_!X \in \mathcal{V}', j_*X \in \mathcal{V}'', (R^1i_!)(X) = 0 \}$
  • $\mathcal{M}_{\mathcal{U}'}^{\mathcal{U}''} = \{ X \in \mathcal{A} \mid i_*X \in \mathcal{U}', j_*X \in \mathcal{U}'', (L^1i_*)(X) = 0 \}$
  • Note: The conditions $(R^1i_!)(X)=0$ and $(L^1i_*)(X)=0$ replace the strict exactness of the functors.

• Introduction of Condition (P):
  To ensure the glued classes form a valid cotorsion pair and to characterize the orthogonal complements, the authors introduce a specific constraint on the recollement called Condition (P):

  • The counit $\epsilon: j_!j^* \to \text{id}_{\mathcal{A}}$ must be a monomorphism for any projective object $P \in \mathcal{A}$.
  • This condition is shown to be satisfied if $i_*$ or $i_!$ is exact, but also holds in broader contexts (e.g., Morita rings with monomorphism maps).

• Homological Tools:
  The proofs rely heavily on:

  • Adjunction isomorphisms and derived functors ($L_1, R_1$).
  • Snake Lemma applications to analyze exact sequences induced by units and counits.
  • Properties of resolving and coresolving subcategories.
  • Special precovers and preenvelopes to establish completeness.

3. Key Contributions and Results

A. Generalized Gluing Theorems

The paper establishes sufficient conditions for the existence of cotorsion pairs in $\mathcal{A}$ without requiring $i_!$ or $i_*$ to be exact.

• Theorem 3.3:
  • If $(L_1j_!)(\mathcal{U}'') = 0$, then $(^\perp\mathcal{N}_{\mathcal{V}'}^{\mathcal{V}''}, \mathcal{N}_{\mathcal{V}'}^{\mathcal{V}''})$ is a cotorsion pair in $\mathcal{A}$.
  • If $(R_1j_*)(\mathcal{V}'') = 0$, then $(\mathcal{M}_{\mathcal{U}'}^{\mathcal{U}''}, (\mathcal{M}_{\mathcal{U}'}^{\mathcal{U}''})^\perp)$ is a cotorsion pair in $\mathcal{A}$.

B. Characterization and Coincidence

The authors determine when the two constructed pairs coincide to form a single pair $(\mathcal{M}_{\mathcal{U}'}^{\mathcal{U}''}, \mathcal{N}_{\mathcal{V}'}^{\mathcal{V}''})$.

• Proposition 3.7: Under Condition (P), if $(L_1j_!)(\mathcal{U}'') = 0$ and $(R_1j_*)(\mathcal{V}'') = 0$, then:
  $$(^\perp\mathcal{N}_{\mathcal{V}'}^{\mathcal{V}''}, \mathcal{N}_{\mathcal{V}'}^{\mathcal{V}''}) = (\mathcal{M}_{\mathcal{U}'}^{\mathcal{U}''}, (\mathcal{M}_{\mathcal{U}'}^{\mathcal{U}''})^\perp)$$
  This implies that $(\mathcal{M}_{\mathcal{U}'}^{\mathcal{U}''}, \mathcal{N}_{\mathcal{V}'}^{\mathcal{V}''})$ is a cotorsion pair.

C. Heredity and Completeness

• Theorem 3.12: If the original pairs $(\mathcal{U}', \mathcal{V}')$ and $(\mathcal{U}'', \mathcal{V}'')$ are complete and hereditary, and the vanishing conditions on derived functors hold, then the glued pair $(\mathcal{M}_{\mathcal{U}'}^{\mathcal{U}''}, \mathcal{N}_{\mathcal{V}'}^{\mathcal{V}''})$ is also complete and hereditary.
• Theorem 3.9: The glued pair is hereditary if and only if the original pairs are hereditary.

D. Applications to Morita and Triangular Matrix Rings

The theory is applied to Morita rings $\Lambda(\phi, \psi) = \begin{pmatrix} A & N \\ M & B \end{pmatrix}$.

• Condition (P) Verification: The authors prove that the recollements associated with Morita rings satisfy Condition (P) if the bimodule maps $\phi$ or $\psi$ are monomorphisms.
• New Constructions:
  • Corollary 4.6 & 4.7: Construct complete hereditary cotorsion pairs on Morita rings where $\phi$ or $\psi$ is a monomorphism.
  • Corollary 4.8 & 4.9: Handle the case where $M \otimes_A N = 0$ or $N \otimes_B M = 0$.
  • Comparison: The authors explicitly show (via Example 4.11) that their constructed pairs are distinct from those previously constructed by Zhang-Cui-Rong [6], particularly in cases where $\phi=0=\psi$. Their method works even when $i_!$ is not exact, a scenario where previous methods failed.

4. Significance

• Relaxation of Constraints: The paper significantly broadens the scope of gluing techniques in relative homological algebra. By replacing the exactness of $i_!$ or $i_*$ with the vanishing of specific derived functors ($L_1j_!, R_1j_*$) and the monomorphism condition on the counit, the authors enable the study of cotorsion pairs in more complex ring structures.
• Unification: It unifies and strengthens existing results on triangular matrix rings and Morita rings, recovering known results as special cases while providing new constructions for cases previously inaccessible.
• Model Structures: Since cotorsion pairs are in correspondence with abelian model structures (Hovey's correspondence), these results facilitate the construction of new model structures on Morita rings and related categories, which is crucial for representation theory and homotopy theory.
• Independent Interest: The introduction of "Condition (P)" and the analysis of recollements where the counit is a monomorphism on projectives are highlighted as having independent interest due to their rich homological properties.

In summary, this paper provides a robust theoretical framework for "gluing" homological structures across recollements, overcoming previous rigidity constraints and yielding new, explicit constructions of cotorsion pairs in Morita contexts.