🔢 mathematics

Local Cohomological Defect and a Conjecture of Mustata-Popa

This paper establishes a general result on the depth of Du Bois complexes for singular varieties, which is then used to prove the Mustata-Popa conjecture and extend the study of local cohomological defect over the complex numbers.

Original authors: Andrew Burke

Published 2026-02-06

📖 5 min read🧠 Deep dive

CC BY 4.0

Original authors: Andrew Burke

Original paper licensed under CC BY 4.0 (http://creativecommons.org/licenses/by/4.0/). ✨ This is an AI-generated explanation of the paper below. It is not written or endorsed by the authors. For technical accuracy, refer to the original paper. Read full disclaimer

Imagine you are an architect trying to understand the structural integrity of a building that has some cracks and imperfections. In the world of mathematics, specifically algebraic geometry, these "buildings" are shapes called varieties, and the "cracks" are called singularities.

This paper, written by Andrew Burke, is like a new set of blueprints and a diagnostic tool for figuring out exactly how "broken" these shapes are, and how to measure that brokenness more accurately than ever before.

Here is a breakdown of the paper's main ideas using simple analogies:

1. The Problem: Measuring the "Brokenness"

Mathematicians have a specific way of measuring how singular (cracked) a shape is. They call this the Local Cohomological Defect (or lcdef for short).

The Analogy: Think of a smooth, perfect sphere. It has a defect of 0. Now, imagine a sphere with a sharp point or a tear. The "defect" number tells you how far the shape is from being perfect.
The Old Way: To calculate this number, mathematicians used to have to check every single layer of the shape's internal structure, from the bottom up to the very top. It was like checking every single brick in a skyscraper to see if the building was safe. This was slow and tedious.

2. The Big Discovery: The "Shortcut" Rule

The main achievement of this paper is a shortcut. Burke proves that you don't need to check every single layer.

The Analogy: Imagine you want to know if a multi-story building is safe. Previously, you had to inspect the foundation, the 1st floor, the 2nd floor, all the way to the roof. Burke discovered that if you check only the lower floors (specifically, up to a certain height), you can mathematically guarantee the safety of the entire building.
The Result: If the lower layers are strong enough, the upper layers must also be strong enough in a specific way. This allows mathematicians to stop checking much earlier, saving a huge amount of work.

3. Solving a Famous Mystery (The Mustață-Popa Conjecture)

There was a long-standing guess (a conjecture) made by two other mathematicians, Mustață and Popa. They wondered if a specific type of "weakness" in the foundation of a shape would force a specific type of "weakness" in the upper floors.

The Analogy: They asked, "If the basement is solid, does that guarantee the attic is also solid?"
The Solution: Using his new shortcut, Burke proved that yes, it does. He confirmed their guess, showing that a strong foundation (measured by a concept called depth) forces the upper levels to be strong as well. This connects two different ways of looking at the problem that were previously thought to be separate.

4. The "Perversity" Concept: Looking at the Shape from Different Angles

The paper introduces a concept called Perversity Defect.

The Analogy: Imagine looking at a sculpture. If you look at it from the front, you see one set of cracks. If you look from the side, you see another. The "Perversity Defect" is like a tool that lets you measure the cracks from different angles (or "perversities").
Why it matters: Sometimes, a shape looks fine from the front but is actually broken from the side. Burke's work shows how to combine these different angles to get a complete picture of the damage. He proves that if you know the damage is small from a few specific angles, you can predict the total damage of the whole object.

5. The "Du Bois Complex": The X-Ray Machine

To make these discoveries, the paper uses a sophisticated mathematical tool called the Du Bois complex.

The Analogy: Think of this as an X-ray machine for the shape. It doesn't just show the surface; it reveals the hidden internal layers (the "holomorphic invariants").
The Breakthrough: Burke showed that this X-ray machine has a hidden symmetry. If the image looks clear in the lower layers, the symmetry of the machine guarantees the upper layers will also look clear. This is the engine that powers all the shortcuts and proofs in the paper.

Summary of the Impact

Before: To measure the "brokenness" of a complex shape, you had to do a massive amount of work, checking every single part.
After: You can now do a much smaller amount of work (checking only the lower parts) and be mathematically certain of the result for the whole shape.
The Conjecture: It settles a debate by proving that a strong foundation guarantees a strong structure, confirming a guess made by other experts.

In short, this paper gives mathematicians a smarter, faster way to diagnose the health of complex geometric shapes, proving that you don't need to inspect the whole building to know if it's standing tall.

Technical Summary: Local Cohomological Defect and a Conjecture of Mustat¸˘a-Popa

Problem Statement
The paper investigates the local cohomological defect ($lcdef$), a singularity invariant for complex algebraic varieties that has recently attracted interest due to its topological and holomorphic characterizations and its role in Lefschetz-type theorems. For a complex algebraic variety $X$ of dimension $n$ embedded in a smooth variety $Y$ of dimension $d$ , $lcdef(X)$ is defined as $\max\{q : H^q_X(\mathcal{O}_Y) \neq 0\} - c$ , where $c = d-n$ .

A central formula established by Mustat¸˘a and Popa [MP22] expresses $lcdef(X)$ in terms of the depths of Du Bois complexes $\Omega^p_X$ :
$lcdef(X) = n - \min_{0 \le p \le n} \{ \text{depth } \Omega^p_X + p \}.$
While this formula is independent of the embedding, computing it a priori requires determining $\text{depth } \Omega^p_X$ for all $p \le n$ . Previous results by Ogus [Og73] and Dao-Takagi [DT16] established that for small values of $k$ (specifically $k \le 3$ ), the condition $\text{depth } \mathcal{O}_X \ge k$ implies $lcdef(X) \le n-k$ . However, this pattern fails for $k > 3$ . The paper aims to:

Determine the minimal range of $p$ required to compute $lcdef(X)$.
Prove a conjecture by Mustat¸˘a-Popa regarding the vanishing of higher direct images of logarithmic differential forms.
Provide a refined criterion for bounding $lcdef(X)$ that extends the known results of Dao-Takagi and Park-Popa [PP25].

Methodology
The paper relies fundamentally on the theory of mixed Hodge modules (Saito [Sa88, Sa90]). The core technical approach involves identifying Du Bois complexes with graded de Rham pieces of the trivial Hodge module $\mathbb{Q}^H_X[n]$ .

Key methodological steps include:

Identification of Complexes: Utilizing Saito's result that $\Omega^p_X \cong \text{gr}^F_{-p} DR(\mathbb{Q}^H_X[n])[p-n]$ . This allows the translation of depth questions about Du Bois complexes into vanishing questions about the graded de Rham pieces of mixed Hodge modules.
Weight Filtration Analysis: The proof leverages the weight filtration $W_\bullet$ on mixed Hodge modules. Specifically, the paper analyzes the "RHM-defect" object $K^\bullet_X$ , defined via the distinguished triangle involving the intersection cohomology Hodge module $IC^H_X$ . A crucial observation is that while $\mathbb{Q}^H_X[n]$ does not have the necessary weight bounds, the defect object $K^\bullet_X$ (and its dual) satisfies specific weight constraints ( $\le n-1$ and $\ge -n+1$ respectively).
Spectral Sequence Arguments: The authors employ spectral sequences relating the cohomology of graded de Rham pieces to the cohomology of the underlying modules. A key technical lemma (Proposition 6.5) establishes an abstract vanishing statement: if graded de Rham pieces vanish for low indices $p \le m$ and high cohomological degrees $j \ge c$ , they must also vanish for a symmetric range of high indices $p \ge -c - m - w$ , provided the module has weights $\ge w$ .
Perversity Functions: The paper introduces a family of "perversity defects" $pdef(X, d)$ parameterized by integers $0 \le d \le n$ . These generalize $lcdef(X)$ (which corresponds to the middle perversity $d=n$ ) and the constructible defect (zero perversity $d=0$ ). These are interpreted via the cohomology of small punctured neighborhoods of points in a Whitney stratification.

Key Contributions and Results

Reduction of Computation (Theorem A):
The paper proves that to determine whether $lcdef(X) \le n-k$ , it is not necessary to check all $p$ . It suffices to check the condition $\text{depth } \Omega^p_X + p \ge k$ only for $p \le \lceil (k-3)/2 \rceil$ .
- Significance: This recovers and slightly extends the results of Ogus and Dao-Takagi for $k \le 3$ and explains why the simple pattern involving only $\text{depth } \mathcal{O}_X$ fails for $k > 3$ .
General Depth Propagation (Theorem B):
A more general result states that if $\text{depth } \Omega^p_X + p \ge k$ for all $p \le m$ , then the same inequality holds for all $p \ge k - m - 2$ . This is the engine driving the reduction in Theorem A.
Proof of Mustat¸˘a-Popa Conjecture (Corollary D):
The paper proves Conjecture 14.1 from [MP22]: If $\text{depth } \mathcal{O}_X \ge j+2$ , then $R^{d-2}f_* \Omega^{d-j}_Y(\log E) = 0$ for a log resolution $f: Z \to Y$ .
- Mechanism: Using Theorem B with $m=0$ , the authors show that $\text{depth } \Omega^0_X \ge j+2$ implies $\text{depth } \Omega^j_X \ge 2$ . Since $\text{depth } \Omega^0_X \ge \text{depth } \mathcal{O}_X$ , the conjecture follows. This extends previous knowledge which was limited to isolated singularities or rational singularities outside a finite set.
Refined Bounds via Perversity Defects (Theorem F):
The paper provides a necessary and sufficient condition for $lcdef(X) \le n-k$ that generalizes the Dao-Takagi and Park-Popa criteria. Specifically, $lcdef(X) \le n-k$ if and only if:
- $\text{depth } \Omega^p_X + p \ge k$ for all $p \le m$ (for a fixed $m$ ), AND
- $pdef(X, k - 2m - 4) \le n - k$ .
- Significance: This formulation captures the "failure" of the simple depth pattern for $k > 3$ by introducing the perversity defect term, which measures the cohomology of punctured neighborhoods relative to specific perversity functions.
Explicit Formulas for Specific Cases:
- Corollary 7.11 (Theorem E): If $X$ satisfies $\text{depth } \Omega^p_X + p \ge n$ for $p \le m-1$ and $\text{depth } \Omega^m_X \le m+3$ , then $lcdef(X)$ is explicitly determined by $\max\{0, n - \text{depth } \Omega^m_X - m\}$ .
- Corollary 8.5: For even $k$ , $lcdef(X) \le n-k$ is equivalent to $\text{depth } \Omega^p_X + p \ge k$ for $p \le (k-4)/2$ and the constructible defect $cdef(X) \le n-k$ .

Significance and Claims
The paper claims to provide a "complete picture" of the local cohomological defect by:

Explaining the failure of the pattern $\text{depth } \mathcal{O}_X \ge k \implies lcdef(X) \le n-k$ for $k > 3$ .
Proving the Mustat¸˘a-Popa conjecture in full generality, removing previous restrictions on the singularities of $X$ .
Offering a unified framework using mixed Hodge modules and perversity defects that subsumes and improves upon existing bounds (specifically those of Ogus, Dao-Takagi, and Park-Popa).
Providing a direct explanation for the equivalence between the topological and holomorphic characterizations of $lcdef(X)$, a question raised in [MP22].

The authors emphasize that their results are derived strictly from the theory of mixed Hodge modules and do not rely on new experimental data or unproven future implications. The work is presented as a theoretical resolution to specific open problems in the study of singularities and Hodge theory.