---
title: 'Matlis Domain: Homological and Divisibility Aspects'
url: https://www.emergentmind.com/topics/matlis-domain
type: topic
---

# Matlis Domain: Homological and Divisibility Aspects

A **Matlis domain** is a commutative domain \(R\) whose field of fractions \(Q\) has projective dimension at most \(1\) as an \(R\)-module, i.e. \(\operatorname{pd}_R Q \le 1\). The notion arose as an extension of the homological behavior familiar from Dedekind domains and has become a central organizing condition in the study of torsion modules, cotorsion and contramodule structures, divisible modules, and homological constructions attached to \(Q\) and \(Q/R\) [1605.08018]. A later characterization identifies the same class of domains by the condition that every divisible \(R\)-module is \(h\)-divisible, and this criterion is equivalent to the existence of strongly flat covers for all divisible modules over \(R\) [2509.01045].

## 1. Definition and basic characterizations

The ring-theoretic definition is homological: if \(R\) is an integral domain with field of fractions \(Q\), then \(R\) is a Matlis domain precisely when
\[
\operatorname{pd}_R Q \le 1.
\]
This is the formulation used in the derived-categorical treatment of Harrison–Matlis equivalence, where the condition is also generalized from domains to pairs \((R,S)\) consisting of a commutative ring and a multiplicative system with \(\operatorname{pd}_R S^{-1}R \le 1\) [1605.08018].

Several standard examples and extensions are recorded in the literature. Every Dedekind domain is a Matlis domain. More generally, for a Noetherian commutative ring of Krull dimension \(1\), one has \(\operatorname{pd}_R S^{-1}R \le 1\) for any multiplicative system \(S\), so the Matlis-type homological framework extends well beyond the original domain case [1605.08018].

A second characterization is expressed in terms of divisibility. For an integral domain \(R\), an \(R\)-module \(M\) is **divisible** if \(sM=M\) for all \(s\in R\setminus\{0\}\), while \(M\) is **\(h\)-divisible** if it is a quotient of a \(Q\)-vector space. The statement that every divisible \(R\)-module is \(h\)-divisible characterizes Matlis domains [2509.01045]. This characterization is especially useful because it translates a homological condition on \(Q\) into a concrete closure property of a familiar module class.

## 2. Torsion, contramodules, and derived equivalence

The classical Harrison–Matlis correspondence relates torsion modules and reduced cotorsion modules over suitable domains. In modern form, this correspondence is lifted to a triangulated equivalence between derived categories over a Matlis domain and, more generally, over any commutative ring \(R\) with multiplicative system \(S\) satisfying \(\operatorname{pd}_R S^{-1}R \le 1\) [1605.08018].

For such a pair \((R,S)\), an \(R\)-module is **\(S\)-torsion** if every element is annihilated by some \(s\in S\). An \(R\)-module \(C\) is an **\(S\)-contramodule** if
\[
\operatorname{Hom}_R(S^{-1}R,C)=0
\quad\text{and}\quad
\operatorname{Ext}^1_R(S^{-1}R,C)=0.
\]
Under the Matlis condition, the full subcategory of \(S\)-contramodules is abelian, and one obtains equivalences
\[
D^*(R\text{-mod}_{S\text{-tors}})\simeq D^*(R\text{-mod}_{S\text{-ctra}})
\]
for \( *\in\{b,+,-,\infty\}\), as well as for absolute derived categories [1605.08018].

A central construction is the two-term complex
\[
K^\bullet=[R\to S^{-1}R],
\]
with \(R\) in degree \(-1\) and \(S^{-1}R\) in degree \(0\). Its cohomology records \(S^{-1}R/R\), and it functions as a dedualizing complex mediating the equivalence via derived \(\operatorname{Hom}\) and derived tensor operations [1605.08018]. When \(S\) consists of nonzero-divisors, or when the \(S\)-torsion in \(R\) is bounded, the equivalence takes the particularly transparent form of an equivalence between derived categories of abelian categories.

This derived viewpoint places Matlis domains in a broader categorical framework: the condition \(\operatorname{pd}_R Q\le1\) is not merely a local homological bound but the exact hypothesis ensuring that torsion and contramodule theories have parallel exact structures.

## 3. Divisible modules, \(h\)-divisibility, and strongly flat covers

The most explicit module-theoretic characterization of Matlis domains currently recorded in the supplied literature concerns strongly flat covers. Over an integral domain \(R\), every divisible module admits a strongly flat cover if and only if \(R\) is a Matlis domain [2509.01045]. In this sense, the Matlis condition marks the exact point at which a global approximation property for divisible modules becomes valid.

The same paper formulates a localization-theoretic analogue. If \(S\) is a regular multiplicative subset of a commutative ring \(R\) and \(R_S\) is semisimple, then every \(S\)-divisible module admits an \(S\)-strongly flat cover if and only if \(R\) is an \(S\)-Matlis ring [2509.01045]. In the domain case, taking \(S=R\setminus\{0\}\) recovers the classical statement.

This characterization is tied to the passage from divisibility to \(h\)-divisibility. The necessity direction proceeds by showing that strongly flat covers of \(S\)-divisible modules inherit enough divisibility to force \(S\)-\(h\)-divisibility; the sufficiency direction uses the implication from the \(S\)-Matlis condition to \(S\)-\(h\)-divisibility and then invokes existence theorems for covers [2509.01045]. The result is a precise equivalence:
\[
\text{all divisible modules have strongly flat covers}
\iff
\text{all divisible modules are \(h\)-divisible}
\iff
R\text{ is a Matlis domain.}
\]

The surrounding homological theory also emphasizes weak cotorsion phenomena. In a later extension of Matlis’ work, if \(\operatorname{id}_R(N)\le1\), then \(\operatorname{Ext}_R(M,N)\) is weakly cotorsion for every \(R\)-module \(M\), and these weakly cotorsion properties are used to derive splitting criteria and decomposability results [2212.10332]. This suggests that Matlis domains are best understood not only through \(Q\) itself but through the full cotorsion-theoretic environment generated by \(Q\).

## 4. The fraction field, completion, and the Matlis quadric

A recurring package of objects in Matlis-style homological algebra is the **Matlis quadric**
\[
(Q,\;Q/R,\;\widehat{R},\;\widetilde{R}),
\]
where \(Q\) is the fraction field of a domain \(R\), \(Q/R\) controls torsion-theoretic behavior, \(\widehat{R}\) is the completion, and \(\widetilde{R}\) is Matlis’ closed extension [2212.10332]. The study of this quadric is explicitly linked to the theory of Matlis domains and to questions about torsion, cotorsion, injective and projective dimensions, and duality.

Among the homological comparisons established in this setting are
\[
\operatorname{pd}_R(Q)=\operatorname{pd}_R(\widehat{R})\le \operatorname{pd}_R(\widetilde{R}),
\]
with explicit computations in one-dimensional regular or Gorenstein cases, where \(\operatorname{pd}_R(\widehat{R})=1\) [2212.10332]. These formulas show that the projective dimension of \(Q\) is part of a broader network of dimension comparisons involving completion and closure.

Completion phenomena are particularly transparent in one-dimensional local domains. For such a domain \(R\),
\[
Q\otimes_R \widehat{R}\cong Q\oplus \widehat{R}/R,
\]
and there is a short exact sequence
\[
0\to Q\to Q\otimes_R \widehat{R}\to \widehat{R}/R\to 0.
\]
Moreover, \(R\) is complete if and only if \(\widehat{R}/R=0\), equivalently if and only if \(Q\otimes_R\widehat{R}\cong Q\) [1306.3311]. The same paper gives a decomposition
\[
Q\otimes_R\widehat{R}\cong \bigoplus_{i=1}^r \widehat{R_{\mathfrak q_i}}
\]
indexed by the associated primes of \(\widehat{R}\) in the one-dimensional local domain case [1306.3311].

Related duality formulas reinforce the role of completion. If \((R,\mathfrak m)\) is local and \(\mathfrak p\in\operatorname{Spec}R\) with \(\dim R/\mathfrak p=1\), then
\[
\operatorname{Hom}_R(E_R(R/\mathfrak p),E)\cong \widehat{R_{\mathfrak p}}
\]
if and only if \(R/\mathfrak p\) is complete [1306.3311]. In a higher-rank direction, the indecomposability problem originally studied by Matlis is extended to show that for any torsion-free indecomposable module \(S\) of finite rank, \(Q/R\otimes_R S\) is indecomposable whenever it is nonzero [2212.10332].

## 5. Related notions and terminological distinctions

The name “Matlis” appears in several adjacent but non-equivalent constructions. The following terminology occurs in the supplied literature.

| Notion | Definition or criterion | Source |
|---|---|---|
| **Matlis domain** | Integral domain \(R\) with \(\operatorname{pd}_R Q\le1\) | [1605.08018] |
| **\(S\)-Matlis ring** | For regular \(S\), the paper states that \(R\) is \(S\)-Matlis if \({}_R R_S\le1\) | [2509.01045] |
| **Weakly Matlis domain** | Finite \(t\)-character plus independence of maximal \(t\)-ideals | [2005.10633] |
| **Matlis-reflexive module** | A module \(M\) with \(M\to M^{\circ\circ}\) an isomorphism | [1306.6820] |
| **Matlis semi-regular ring** | Every module embeds in a flat module; equivalently every injective module is flat | [1604.02795] |

The **weakly Matlis** condition belongs to multiplicative ideal theory rather than to the homological definition \(\operatorname{pd}_R Q\le1\). In the cited treatment, a domain \(D\) is weakly Matlis when it has finite \(t\)-character and satisfies independence, meaning that no two distinct maximal \(t\)-ideals contain a common nonzero prime ideal [2005.10633]. For a Prüfer \(v\)-multiplication domain, this notion enters a factorization-theoretic equivalence:
\[
D \text{ is a VFD}
\iff
D \text{ is a weakly Matlis GCD-domain}
\iff
D[X]\text{ is a VFD}
\]
[2005.10633].

The expression **Matlis-reflexive** refers instead to a module over a Noetherian local ring. If \(E\) is the injective hull of the residue field and \(M^\circ=\operatorname{Hom}_R(M,E)\), then \(M\) is Matlis-reflexive when the canonical map \(M\to M^{\circ\circ}\) is an isomorphism [1306.6820]. The cited characterization uses Bass numbers:
\[
M \text{ is reflexive}
\iff
\mu(\mathfrak p,M)=\mu(\mathfrak p,M^{\circ\circ})
\text{ for all }\mathfrak p\in\operatorname{Spec}(R)
\]
[1306.6820].

A different ring-theoretic branch of the terminology concerns **Matlis semi-regular rings** or IF-rings. In the study of trivial ring extensions \(A\ltimes E\) with \(A\) a domain, semi-regularity is characterized via coherence, divisibility of \(E\), annihilator conditions, and a module-theoretic double annihilator condition (DAC) [1604.02795]. This theory enriches the supply of semi-regular rings, but it is distinct from the definition of a Matlis domain.

## 6. Role in current research

In recent work, Matlis domains function as a boundary condition separating well-behaved homological and categorical regimes from more pathological ones. On the categorical side, the Matlis condition allows the classical torsion–cotorsion correspondence to be promoted to a triangulated equivalence of derived categories [1605.08018]. On the approximation-theoretic side, it is the exact criterion for the existence of strongly flat covers of all divisible modules [2509.01045].

The same framework supports higher-dimensional and finite-rank generalizations of Matlis’ original homological results. These include weakly cotorsion properties of \(\operatorname{Ext}\), splitting criteria for modules of finite injective dimension, computations of the projective dimension of \(\widehat{R}\), non-Noetherian versions of Grothendieck’s localization problem, and higher-rank forms of Matlis’ decomposability problem [2212.10332]. The recurrent structures are the fraction field \(Q\), the quotient \(Q/R\), completion, and the cotorsion phenomena they generate.

For that reason, the modern significance of Matlis domains is broader than the bare inequality \(\operatorname{pd}_R Q\le1\). The condition organizes a theory in which localization, divisibility, completion, torsion, contramodules, and derived duality all interact with unusually low homological complexity.

Source: https://www.emergentmind.com/topics/matlis-domain