Higher Divisorial Ideal in Rings

• Higher divisorial ideal is defined as D(I) = Ann_R(Ext^g_R(R/I, R)), capturing key homological interactions in a Noetherian ring.
• It lies between I and rad(I), with proven cases where D(I) ⊆ overline{I}, especially in Cohen–Macaulay and quasi-Gorenstein settings.
• The study of D(I) informs applications in symbolic powers, local cohomology, and the triviality of vector bundles, while prompting open research questions.

The higher divisorial ideal, denoted $D(I) := \Ann_R(\Ext^g_R(R/I, R))$, is a homological ideal-theoretic invariant attached to any ideal $I$ of grade $g$ in a Noetherian ring $R$. Unlike classical closures such as the integral, reflexive, and symbolic closures, $D(I)$ is defined via the annihilator of the first nonvanishing Ext module, allowing it to encode deep interactions between the homological, module-theoretic, and closure-theoretic structures of $R$ and $I$. Recent work has placed particular emphasis on the containment $D(I) \subseteq \overline{I}$ and its structural, algebraic, and geometric implications (Asgharzadeh, 28 Jan 2026).

1. Foundational Definitions and Key Properties

Let $R$ be a Noetherian ring and $I \subseteq R$ an ideal of grade $g$. The grade of a finitely generated $R$-module $M$ is given by

$\grade_R(M) = \min\{\,i \mid \Ext^i_R(M,R) \ne 0 \,\},$

which for $M = R/I$ corresponds to the maximal length of an $R$-regular sequence in $I$, satisfying $0 \le \grade_R(R/I) \le \mathrm{height}(I)$.

The higher divisorial ideal is

$D(I) := \Ann_R\left(\Ext^g_R(R/I, R)\right).$

Elementary Koszul-homology yields

$I \subseteq D(I) \subseteq \rad(I),$

and, in the Cohen–Macaulay case, also $I^{unm} \subseteq D(I)$, where $I^{unm}$ is the unmixed part of $I$.

2. Containment Theorems for $D(I)$

The central structural question is: under which hypotheses does one have $D(I) \subseteq \overline{I}$ (the integral closure)? The following positive results delineate the main cases:

• Unmixed Ideals of Finite Projective Dimension in Quasi-Normal 3-Folds: If $(R, \m)$ is a 3-dimensional quasi-normal local ring and $I$ is unmixed with $\pd_R(R/I) < \infty$, then $D(I) \subseteq \overline{I}$. The proof reduces by heights of $I$, employing localizations and homological classification (e.g., perfectness and m-primary status).
• Parameter Ideals in Universally Catenary, Quasi-Gorenstein Local Rings: If $I$ is a parameter ideal of height $g$, $D(I) \subseteq \overline{I}$. When $g = \dim R$, this is immediate as $D(I) = I$.
• Powers of Perfect Ideals in Positive Characteristic: For $R$ of characteristic $p > 0$, if $P \subseteq R$ is a perfect prime of grade $t$ and certain Betti number conditions are satisfied (e.g., $\beta_t(R/P) = 1$), then for every $q=p^e$, $D(P^{[q]}) = P^{[q]} = \overline{P^{[q]}}$.
• Ideals of Analytic Spread One: For $P$ of height one and analytic spread one, for all $n \geq 1$, $D(P^n) = (P^n)^{**} = \overline{P^n}$.

These results rest on the interplay between homological dimension, analytic spread, and normality conditions, utilizing Koszul homology, Fitting invariants, and connections to parameter test ideals.

3. Examples, Counterexamples, and Limitations

The hypotheses above are necessary: several cases demonstrate the failure of the containment or monotonicity properties if the assumptions are relaxed.

• Lack of Unmixedness: In $R = k[[X,Y]]$, the ideal $I = (X^2, XY)$ yields $D(I) = (X)$, but a larger ideal with the same radical may have nonmonotonic behavior.
• Distinctions Between Symbolic and Integral Closures: For $R = k[[t^3, t^4, t^5]]$, the unique height-2 prime $P$ satisfies $P^2 = \overline{P^2} \ne P^{(2)}$, and erroneous identification of $D(P^2) \subseteq \overline{P^2}$ leads to inconsistencies.
• Non-Cohen–Macaulay Base: In $R = k[[X,Y]]/(X^2, XY)$, $D(\m^n) = \m$ for $n \geq 2$, but $\overline{\m^n} = \m^n$.

This suggests the necessity of purity or finite projective dimension for favorable behavior of $D(I)$.

4. Structural Relationships with Other Ideal Operations

The higher divisorial ideal interfaces with several classical closures and operations:

• Unmixed Part: Over Cohen–Macaulay rings, always $I^{unm} \subseteq D(I) \subseteq \rad(I)$.
• Reflexive Closure: If $\grade(I) = 1$, then $D(I) = (I^{*})^{*} = I^{**}$.
• Symbolic Powers: For prime $P$ of height one in a normal ring, $$D(P^n) = (P^n)^{**} = P^{(n)} \subseteq \overline{P^n}$$.
• Frobenius and Tight Closure: In one-dimensional complete domains of char $p > 0$, $D(I) \subseteq \overline{I} = I^F = I^\star$.
• Trace Ideals: For trace ideals containing a nonzerodivisor, $\tr(I) \subseteq D(I)$, and sometimes $\tr(I) \subseteq \overline{I}$.
• Iterated Closures: The sequence $I \subseteq D(I) \subseteq D^2(I) \subseteq \ldots$ stabilizes when $R/I$ is Noetherian.

A comparative table of these invariants:

Ideal Operation	Expression	Containment Hierarchy
Unmixed part	$I^{unm}$	$I \subseteq I^{unm} \subseteq D(I)$
Reflexive closure	$I^{**}$	$D(I) \subseteq I^{**}$
Integral closure	$\overline{I}$	$I^{**} \subseteq \overline{I}$

This organization highlights the place of $D(I)$ in the spectrum of closure operations and the extent to which it is detected by homological, rather than only valuation-theoretic, invariants.

5. Applications to Module Theory and Local Cohomology

The structure of $D(I)$ carries implications for the theory of reflexive modules, vector bundles, and conductor ideals.

• Triviality of Vector Bundles: Over regular local $R$ of $\dim R > 4$, if a reflexive module $M$ is free on the punctured spectrum and certain second cohomology vanishes, then $M$ is free. The argument employs $D(I)$ to detect properties of the associated Ext modules.
• Reflexive Module Criteria: If for a reflexive module $M$, the equality $\Ann(\Ext^{d-1}_R(M,R)) = \Ann(M/\Gamma_\m(M))$ holds, vanishing behavior of local cohomology can be deduced.
• Conductor and Local Cohomology: The annihilator $\Ann(\Ext^1_R(R/I, R))$ is connected to the conductor ideal $\mathfrak{C}_R$ of the integral closure of $R$, as well as to the higher local cohomology $H^i_I(R)$.

These applications underscore that $D(I)$ serves as a diagnostic for both algebraic and geometric phenomena, particularly in the behavior of vector bundles and the structure of singularities.

6. Open Problems and Research Directions

The theory surrounding $D(I)$ prompts several open questions:

• Characterization of the Containment: For which classes of Cohen–Macaulay or Gorenstein rings does $D(I) \subseteq \overline{I}$ universally hold?
• Valuation and Rees-Algebra Descriptions: Can one describe $D(I)$ in terms of valuation-theoretic or Rees-algebra data?
• Behavior under Ideal Operations: How does $D(I)$ respond to sum, intersection, and (co)restrictions under flat or étale maps?
• Geometric Interpretation: Are there geometric or sheaf-theoretic analogues of $D(I)$ relating to adjoint divisors or multiplier ideals in birational geometry?

The homological origin of $D(I)$ differentiates it sharply from algebraic closure notions, promoting further research at the interface of commutative algebra, algebraic geometry, and homological methods (Asgharzadeh, 28 Jan 2026).