Constant coefficient and intersection complex $L$-classes of projective varieties

Irma Pallar\'es; Javier Fern\'andez de Bobadilla; Morihiko Saito

arxiv: 2407.11769 · v4 · submitted 2024-07-16 · 🧮 math.AG

Constant coefficient and intersection complex L-classes of projective varieties

Javier Fern\'andez de Bobadilla , Irma Pallar\'es , Morihiko Saito This is my paper

Pith reviewed 2026-05-23 22:47 UTC · model grok-4.3

classification 🧮 math.AG

keywords L-classesintersection complexconstant coefficientsprojective varietiesHodge signaturesisolated singularitieshyperplane sectionsvirtual L-class

0 comments

The pith

The intersection complex L-class and constant coefficient L-class of a projective variety differ after reduction to Q-homologically isolated singularities precisely when the Hodge signatures of the intersection complex stalks are nonzero.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

This paper examines when two different definitions of L-classes for projective varieties agree. One comes from the intersection complex using cohomotopy, the other from applying an L-class transformation to a cubic hyperresolution. They always agree on Q-homology manifolds. The authors prove that any difference between them can be detected after intersecting with general hyperplanes to reach a variety with only Q-homologically isolated singularities. This reduces the problem of finding when they differ to the isolated singularity case, where a necessary and sufficient condition already exists from prior work in terms of Hodge signatures.

Core claim

We show that the two L-classes L_*(X) and L^c_*(X) differ if they do by replacing X with an intersection of general hyperplane sections which has only Q-homologically isolated singularities. Finding a good sufficient condition for the non-coincidence of L_*(X) and L^c_*(X) is thus reduced to the latter case, where a necessary and sufficient condition has been obtained in terms of the Hodge signatures of stalks of intersection complex in our previous paper. In the case of projective hypersurfaces having only isolated singularities, the difference between L_*(X) and L^c_*(X) is given by the Hodge signatures of the link cohomologies at singular points, and the Hodge signatures of the vanishing

What carries the argument

Reduction via successive general hyperplane sections to a variety with only Q-homologically isolated singularities, allowing the difference between the two L-classes to be read off from Hodge signatures of stalks of the intersection complex.

Load-bearing premise

The difference between L_*(X) and L^c_*(X) remains unchanged in a detectable way when passing to a general hyperplane section that isolates the singularities in the Q-homological sense.

What would settle it

A projective variety with Q-homologically isolated singularities where the Hodge signatures of the intersection complex stalks are all zero, yet L_*(X) still differs from L^c_*(X).

read the original abstract

For a projective variety $X$, we have the intersection complex $L$-classes $L_*(X)$ defined by Goresky-MacPerson using cohomotopy and also the constant coefficient $L$-class $L^c_*(X)$ defined by applying an $L$-class transformation (or $T_{1*}$) to a cubic hyperresolution of $X$. These coincide if $X$ is a $\mathbb Q$-homology manifold. We show that the two $L$-classes $L_*(X)$ and $L^c_*(X)$ differ if they do by replacing $X$ with an intersection of general hyperplane sections which has only $\mathbb Q$-homologically isolated singularities. Finding a good sufficient condition for the non-coincidence of $L_*(X)$ and $L^c_*(X)$ is thus reduced to the latter case, where a necessary and sufficient condition has been obtained in terms of the Hodge signatures of stalks of intersection complex in our previous paper. In the case of projective hypersurfaces having only isolated singularities, the difference between $L_*(X)$ and $L^c_*(X)$ is given by the Hodge signatures of the link cohomologies at singular points, and the Hodge signatures of the vanishing cohomologies give the difference between $L^c_*(X)$ and the virtual $L$-class of $X$, that is, the image by a retraction map of the $L$-class of a smooth deformation of $X$ in an ambient smooth projective variety $Y$ in the very ample case.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

The paper reduces comparison of the two L-classes to the isolated-singularity case and supplies explicit Hodge-signature difference formulas for hypersurfaces, but the reduction step is the part that needs the most scrutiny.

read the letter

The main point is that the authors reduce the question of when L_*(X) and L^c_*(X) differ to the case of Q-homologically isolated singularities by cutting with general hyperplanes, then give concrete formulas for the difference on projective hypersurfaces in terms of link and vanishing Hodge signatures. This extends their earlier necessary-and-sufficient condition and makes the comparison more computable in the isolated case. The reduction itself and the explicit formulas for hypersurfaces are the genuinely new pieces; the rest is careful bookkeeping with existing definitions from Goresky-MacPherson and the hyperresolution approach. The logical outline is coherent and the citations to prior work are appropriate rather than excessive. The authors have a solid track record in this area, so the claims are at least worth checking in detail. The soft spot is exactly the reduction step flagged in the stress-test note. It is not obvious from the abstract that the difference class commutes with the restriction to a general complete intersection in the way needed for the stalk signatures to capture it exactly; the cohomotopy and hyperresolution constructions do not automatically play well together under restriction. If the full proofs close this gap cleanly, the formulas become reliable; if not, the reduction may only work under extra assumptions. Minor point: the virtual L-class is defined via retraction from a smooth deformation, which is standard but should be spelled out so readers see the precise map. This is narrow specialist material in characteristic classes for singular varieties. A reader already working on L-classes or mixed Hodge modules on singular spaces will get direct value from the formulas and the reduction technique. It is not broad enough to change how most algebraic geometers think about L-classes, but the explicit statements are the sort of thing that can be tested on examples or used in follow-up calculations. The paper deserves a serious referee because the claims are specific and falsifiable once the proofs are examined; it is not obviously flawed on its face and the authors have the background to make the arguments hold up or be corrected. I would send it to review rather than desk-reject.

Referee Report

2 major / 2 minor

Summary. The manuscript claims that for a projective variety X, the intersection complex L-classes L_*(X) (defined via Goresky-MacPherson cohomotopy) and constant coefficient L-classes L^c_*(X) (defined via L-class transformation on a cubic hyperresolution) differ, if at all, after replacing X by a general complete intersection of hyperplane sections that has only Q-homologically isolated singularities. This reduces the search for sufficient conditions on non-coincidence to the isolated-singularity case, where a necessary-and-sufficient condition in terms of Hodge signatures of intersection-complex stalks is available from the authors' prior work. For projective hypersurfaces with isolated singularities, the difference L_*(X) - L^c_*(X) is expressed via Hodge signatures of link cohomologies at singular points, while the difference between L^c_*(X) and the virtual L-class is given by Hodge signatures of vanishing cohomologies.

Significance. If the reduction is valid, the work supplies a practical method for deciding agreement of the two L-class constructions on projective varieties by reducing to a case already characterized by Hodge-theoretic data. The explicit formulas for hypersurfaces furnish concrete, computable expressions in terms of link and vanishing cohomology, which is a clear strength for applications in singularity theory and characteristic-class computations. The paper thereby bridges the cohomotopy and hyperresolution approaches to L-classes while extending prior results on isolated singularities.

major comments (2)

[Reduction argument (as stated in the abstract and developed in the main body)] The central reduction claim (abstract and the argument that replaces X by a general complete intersection Y = X ∩ H_1 ∩ … ∩ H_k with only Q-homologically isolated singularities) requires an explicit compatibility statement showing that the difference class L_*(X) − L^c_*(X) is unchanged under this operation and remains supported exactly at the isolated singularities in a form that matches the stalk signatures of the previous paper. The cohomotopy definition of L_* and the hyperresolution definition of L^c_* need not a priori commute with restriction in the required way; without a push-forward relation or diagram for the difference, the reduction to the isolated-singularity criterion is not yet load-bearing.
[Hypersurface formulas (isolated-singularity case)] In the hypersurface case with isolated singularities, the statements that the difference L_*(X) − L^c_*(X) equals the Hodge signatures of the link cohomologies and that L^c_*(X) minus the virtual L-class equals the Hodge signatures of the vanishing cohomologies must be derived from the stalk condition of the prior paper; if these formulas are obtained by specializing the general reduction, the specialization step (including any retraction map to the virtual class) needs to be written out explicitly so that the reader can verify it follows directly from the isolated-singularity criterion.

minor comments (2)

[Abstract] Typo in the abstract: 'Goresky-MacPerson' should read 'Goresky-MacPherson'.
[Abstract] The abstract introduces L_*(X) and L^c_*(X) but does not immediately recall the precise definitions (cohomotopy versus hyperresolution); a one-sentence reminder at the first use would improve readability.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the careful reading and for identifying points where the reduction argument and hypersurface formulas require greater explicitness. We will revise the manuscript to address both major comments.

read point-by-point responses

Referee: [Reduction argument (as stated in the abstract and developed in the main body)] The central reduction claim (abstract and the argument that replaces X by a general complete intersection Y = X ∩ H_1 ∩ … ∩ H_k with only Q-homologically isolated singularities) requires an explicit compatibility statement showing that the difference class L_*(X) − L^c_*(X) is unchanged under this operation and remains supported exactly at the isolated singularities in a form that matches the stalk signatures of the previous paper. The cohomotopy definition of L_* and the hyperresolution definition of L^c_* need not a priori commute with restriction in the required way; without a push-forward relation or diagram for the difference, the reduction to the isolated-singularity criterion is not yet load-bearing.

Authors: We agree that an explicit compatibility statement is needed to make the reduction load-bearing. In the revised manuscript we will insert a new subsection that establishes a push-forward relation for the difference L_*(X) − L^c_*(X) under general hyperplane sections. The argument will show that the difference class is preserved (up to the expected degree shift) and remains supported precisely at the Q-homologically isolated singularities, with its local contributions matching the stalk signatures from our prior work. A commutative diagram relating the cohomotopy and hyperresolution constructions under restriction will be included. revision: yes
Referee: [Hypersurface formulas (isolated-singularity case)] In the hypersurface case with isolated singularities, the statements that the difference L_*(X) − L^c_*(X) equals the Hodge signatures of the link cohomologies and that L^c_*(X) minus the virtual L-class equals the Hodge signatures of the vanishing cohomologies must be derived from the stalk condition of the prior paper; if these formulas are obtained by specializing the general reduction, the specialization step (including any retraction map to the virtual class) needs to be written out explicitly so that the reader can verify it follows directly from the isolated-singularity criterion.

Authors: We will expand the hypersurface section to derive the stated formulas directly from the stalk-signature criterion of the previous paper. The specialization step will be written out in full, including the explicit retraction map from the L-class of a smooth deformation to the virtual L-class and the identification of the resulting local contributions with the Hodge signatures of link and vanishing cohomologies. This will make the derivation self-contained and verifiable from the isolated-singularity case. revision: yes

Circularity Check

1 steps flagged

Self-citation supplies the N&S condition after independent reduction to isolated singularities

specific steps

self citation load bearing [Abstract]
"Finding a good sufficient condition for the non-coincidence of L_*(X) and L^c_*(X) is thus reduced to the latter case, where a necessary and sufficient condition has been obtained in terms of the Hodge signatures of stalks of intersection complex in our previous paper."

The load-bearing sufficient condition for detecting non-coincidence is supplied entirely by the authors' own prior result rather than by an independent argument or external verification; the current paper's reduction therefore inherits its decisive criterion from that self-citation.

full rationale

The paper's main contribution is a reduction argument showing that any difference between L_*(X) and L^c_*(X) can be detected after replacing X by a general complete intersection with only Q-homologically isolated singularities. This reduction step is presented as new and does not reduce to prior inputs by construction. The necessary-and-sufficient condition that decides coincidence in the reduced case is, however, taken directly from the authors' previous paper. This produces moderate self-citation dependence on the final criterion without rendering the derivation chain circular by definition or by fitted-parameter renaming. No self-definitional, ansatz-smuggling, or uniqueness-imported patterns appear in the given text.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 0 invented entities

The paper relies on established results in intersection cohomology, Hodge theory, and resolution of singularities without introducing new free parameters or postulated entities.

axioms (2)

standard math Standard properties of intersection cohomology and the Goresky-MacPherson L-class construction
Invoked in the definition of L_*(X)
domain assumption Existence and properties of cubic hyperresolutions and the L-class transformation T_{1*}
Used to define L^c_*(X)

pith-pipeline@v0.9.0 · 5824 in / 1427 out tokens · 45700 ms · 2026-05-23T22:47:13.690549+00:00 · methodology

discussion (0)

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

IndisputableMonolith/Foundation/RealityFromDistinction.lean reality_from_one_distinction unclear

?

unclear
Relation between the paper passage and the cited Recognition theorem.

We show that the two L-classes L_*(X) and L^c_*(X) differ if they do by replacing X with an intersection of general hyperplane sections which has only Q-homologically isolated singularities... the difference ... is given by the Hodge signatures of the link cohomologies at singular points
IndisputableMonolith/Cost/FunctionalEquation.lean washburn_uniqueness_aczel unclear

?

unclear
Relation between the paper passage and the cited Recognition theorem.

the Hodge signatures of the vanishing cohomologies give the difference between L^c_*(X) and the virtual L-class of X

What do these tags mean?

matches: The paper's claim is directly supported by a theorem in the formal canon.
supports: The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends: The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses: The paper appears to rely on the theorem as machinery.
contradicts: The paper's claim conflicts with a theorem or certificate in the canon.
unclear: Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.