A generalised cross-ratio and limits of local heights

Emre Can Sert\"oz; Hidde Lammert Bakker

A generalised cross-ratio and limits of local heights

T0 review · 0 major / 8 minor · reviewed 2026-07-08 · glm-5.2

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read Generalized cross-ratio unifies height pairings and intersection degrees

desk verdict Clean, correct paper giving a generalised cross-ratio equal to the augmented height pairing, with a valuation formula and degenerate extension. Proofs are sound; the one flagged concern about Theorem 2.2 does not land. read the letter →

arxiv 2607.06477 v1 pith:H26EK2U2 submitted 2026-07-07 math.AG math.NT

Hidde Lammert Bakker , Emre Can Sertöz This is my paper · ORCID

classification math.AGmath.NT MSC 14G4014C1732G2014C3014N2014C25

keywords cross-ratioheightpairingmixedHodgestructureintersectiontheoryGrassmanniandegenerationlocalheightsarithmeticgeometry

topics The Hodge Conjecture

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

The paper extends the classical cross-ratio of four points on a line to a cross-ratio of four linear subspaces in projective n-space, where two subspaces are k-dimensional and two are (n-k-1)-dimensional, with each pair from opposite groups being disjoint. The authors prove three things about this generalized cross-ratio. First, over the complex numbers, it equals the augmented height pairing of the corresponding cycles (Theorem A), a direct higher-dimensional analogue of the classical fact that the cross-ratio of four points equals their height pairing. Second, over a discretely valued field, the valuation of the cross-ratio equals the intersection degree of the cycles when spread out over the valuation ring (Theorem B). Third, when the four subspaces are allowed to intersect, a degenerate cross-ratio can be defined using infinitesimal perturbation data, and the limiting value of the height pairing in a degenerating holomorphic family equals this degenerate cross-ratio on the central fibre (Theorem C). Together, these results show that the asymptotic behavior of Archimedean height pairings in degenerating families is governed by a purely algebro-geometric intersection degree, and that the limiting constant term admits a motivic interpretation as a degenerate cross-ratio.

What carries the argument

The cross-ratio is defined as an alternating product of pairings between generators of exterior powers of subspaces and their duals. The proof of Theorem A uses an incidence correspondence between the Grassmannian G(k,n) and projective space P^n to construct an isomorphism of mixed Hodge structures. The proof of Theorem B uses Smith normal form and Fulton's intersection theory for degeneracy loci. The degenerate cross-ratio relies on a determinantal identity (det V ≅ det A ⊗ det B ⊗ hom(det S, det Q)) and a regularisation map that extracts the leading coefficient of a valued element.

What would settle it

If the incidence-correspondence map between H_P and H_G fails to be an isomorphism of mixed Hodge structures (not just at the graded-piece level), Theorem A would not follow, and the entire chain of identifications linking the cross-ratio to height pairings, valuations, and intersection degrees would break.

Watch

Extended reading notes

Core claim

The generalized cross-ratio of four complementary linear subspaces in projective n-space serves as a single object that simultaneously encodes the Archimedean height pairing (over C), the non-Archimedean local height (via valuation), and the intersection degree (over a DVR). The key mechanism is that the cross-ratio, defined as an alternating product of determinantal pairings between subspaces and their annihilators, reduces to the classical cross-ratio when n=1, and the authors prove it coincides with the extension class of a mixed Hodge structure via an incidence correspondence between the Grassmannian and projective space. This identification, combined with the valuation-intersection dier

Load-bearing premise

The proof of Theorem A depends on showing that a map between two mixed Hodge structures, constructed via an incidence correspondence between the Grassmannian and projective space, is an isomorphism. The authors verify this at the level of the weight-graded pieces (the top and bottom layers of the structure) but the full verification that the map respects the complete mixed Hodge structure is condensed; if this map fails to be a genuine isomorphism of mixed Hodge structures, a

Editorial extensions

If this is right

Corollary B' confirms Chen's conjecture (that divergence of Archimedean height pairings in degenerating families is governed by intersection degree) for the case of degenerating linear subspaces of arbitrary dimension, without requiring properness assumptions.
The degenerate cross-ratio provides a framework for computing limit periods of mixed Hodge structures in arbitrary-dimensional degenerations with arbitrarily bad singularities, offering a testbed for more general limit-period conjectures.
The arithmetic cross-ratio and its norm expression in terms of lattice indices connect Archimedean limit periods to non-Archimedean arithmetic data, suggesting a path toward explicit height computations on moduli spaces of linear subspace configurations.
The S4 symmetry of the cross-ratio in the middle-dimensional case (n=2k+1) may yield new constraints on height pairings and their behavior under duality, analogous to classical cross-ratio symmetries.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

The cross-ratio's role as a universal invariant bridging Archimedean and non-Archimedean heights suggests that other classical projective invariants might admit similar height-theoretic interpretations in higher dimensions, particularly for configurations involving more than four subspaces.
The regularisation map reg_t, which extracts the leading coefficient of a valued element, is analogous to regularisation procedures for period integrals; this connection could potentially unify algebraic and analytic approaches to limit periods.
The dependence of the limit perturbation c_lim only on t mod m^2 suggests that the leading asymptotics of height pairings are determined by first-order deformation data, which could simplify numerical computations in practice.
The fact that Theorem B requires no properness assumption, combined with Corollary B', hints that Chen's conjecture might hold for arbitrary (not just proper) degenerations of algebraically trivial cycles beyond the linear case.

Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit.

Desk Editor's Note

read the letter

The paper defines a cross-ratio for complementary configurations of planes in P^n and proves three things: it equals the augmented height pairing (Theorem A), its valuation over a DVR equals the intersection degree of spread-out cycles (Theorem B), and a degenerate version captures the limit period of a degenerating family (Theorem C). Corollary B' resolves Chen's conjecture [Che25, Conjecture 1.5] for degenerating planes of any dimension. This is a real result, independently confirmed by Goncharov [Gon26] via a different (inductive) proof and partially anticipated by Werner [Wer01]. The authors are upfront about all of this in Remarks 1.2 and 4.3. The definition itself (Definition 1.1) is natural — it is the obvious generalisation of the classical formula (1.1) — and the connection to height pairings via the incidence correspondence on the Grassmannian is the right way to see it. Proposition 2.1 computes the extension class cleanly using Carlson's theory, and the Smith normal form argument in Lemma 4.2 is careful: the telescoping sum properly yields v_R(det mu), and the use of Fulton's residual formula is correct. The reader flagged the proof of Theorem 2.2 as potentially condensed — only the weight graded pieces are checked, not the full MHS isomorphism. This concern does not land. Both H_P and H_G are extensions of Z(0) by Z(1), so they have exactly two non-trivial weight graded pieces (weights 0 and 1). By Deligne's strictness theorem, a morphism of MHS that induces isomorphisms on all graded pieces is automatically an isomorphism. The paper's statement that 'we need only show that the generators of the weight graded pieces map to one another' is a complete argument, not a gap. The Z(1) check is terse but the symmetry with the Z(0) check makes it straightforward. Theorem C is more definitional than theorem — the authors say so themselves ('the heavy lifting is done by the definitions') — but the construction of the limit perturbation c_lim in Proposition 5.1 is clean, and the arithmetic cross-ratio and its norm formula (Lemma 5.7) are sensible. The restriction to class number 1 for the arithmetic version is a real limitation but the authors note the fractional ideal generalisation in Remark 5.4. This paper is for arithmetic geometers and Hodge theorists working on height pairings, regulators, or degeneration questions. It is short, self-contained, and the proofs check out. It deserves a serious referee, primarily to verify the Fulton intersection theory details in Section 4 and the determinantal identities in Section 5, which are the parts where a careful reader would want to see every step.

Referee Report

0 major / 8 minor

Summary. This paper generalises the classical cross-ratio of four points on P^1 to a cross-ratio of four linear subspaces in P^n (two k-planes A_1, A_2 and two (n-k-1)-planes B_1, B_2, pairwise disjoint). Three main results are proved: (Theorem A) over C, the cross-ratio equals the augmented height pairing of the cycles A_1-A_2 and B_1-B_2; (Theorem B) over a DVR, the valuation of the cross-ratio equals the intersection degree of the spread-out cycles; (Theorem C) the t-regularised limit of the cross-ratio equals a degenerate cross-ratio on the special fibre. Combining Theorems A and B yields Corollary B', which proves a conjecture of Chen on the asymptotics of Archimedean height pairings for degenerating planes. The proofs combine mixed Hodge theory (Carlson, Hain), an incidence correspondence on the Grassmannian, Smith normal form lattice arguments, and Fulton's intersection theory.

Significance. The paper makes a clean contribution at the interface of Arakelov geometry, mixed Hodge theory, and intersection theory. The generalised cross-ratio is a natural and parameter-free invariant, and the identity with the augmented height pairing (Theorem A) is a satisfying higher-dimensional generalisation of the classical P^1 case. Theorem B provides a concrete, computable formula relating the non-Archimedean local height to an intersection degree, and Corollary B' verifies Chen's conjecture [Che25, Conjecture 1.5] in the setting of degenerating linear subspaces of arbitrary dimension. The degenerate cross-ratio (Definition 1.3) and arithmetic cross-ratio (Definition 5.6) are natural constructions that give a motivic interpretation of limit periods. The independent confirmation by Goncharov [Gon26] via a different method (induction rather than correspondence) strengthens confidence in the results. The Smith normal form argument in Lemma 4.2 is careful and the telescoping computation is correct.

minor comments (8)

Theorem 2.2, proof: the verification that the Z(1) generator maps correctly is dispatched with 'we argue analogously this time using the residue maps.' While the argument is indeed analogous (and sufficient by Deligne's strictness, since both sides are extensions of Z(0) by Z(1) with only two non-trivial weight pieces), a sentence or two making the residue map check explicit would improve readability for readers less familiar with the correspondence.
Proposition 2.1: the statement that 'the exponential of this integral is the extension class for H_G' is described as 'a general fact.' A reference to the specific result in Carlson's theory (e.g., [Car80]) that justifies this would be helpful, even if the result is standard.
Section 3.2: the computation a_2 ∧ b_1 = a_2 ∧ b_2 = (-1)^{k+1} e appears to contain a typo; presumably a_2 ∧ b_j should involve the graph of φ_j, and the two expressions should differ. Please verify this line.
Remark 4.3 notes that Theorem B also follows from [Wer01, Thm. 3.4]. It would be useful to briefly state what is additionally gained by the present proof (e.g., the explicit Smith normal form decomposition, or the removal of the finite-extension hypothesis), so the reader can assess the relative merits.
Definition 1.3 and surrounding text: the notation c = (c_{11}, c_{12}, c_{21}, c_{22}) with c_{ij} ∈ hom(det Ŝ_{ij}, det Q̂_{ij}) is introduced somewhat abruptly. A brief remark clarifying that each c_{ij} lives in a potentially different hom-space (since Ŝ_{ij} and Q̂_{ij} depend on the pair (i,j)) would help the reader.
Corollary C'': the statement involves log|c^⊗_{lim}/c^⊗_{ar}|_σ, but c^⊗_{ar} is defined only up to O^×_κ, so the ratio lives in κ^×/O^×_κ. The absolute value is then well-defined up to log|O^×_κ|_σ, which is consistent with the R/log|O^×_κ|_σ on the right-hand side, but this subtlety should be stated explicitly.
The reference [Bak26] is the first author's M.S. thesis. While appropriate for context, the paper should be self-contained; the proofs of Theorems A, B, and C here appear to be complete, but the authors should confirm that no step silently relies on the thesis.
Page 2, line below equation (1.5): 'the logarithms of primes are linearly independent over Q' — this should say 'over Z' or 'Q-linearly independent' for precision, since the statement is about linear combinations with integer coefficients summing to zero.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for a careful reading and a positive assessment. The referee's recommendation is minor revision, and the report does not raise specific major or minor comments requiring changes to the manuscript. We address the substance of the report below.

read point-by-point responses

Referee: The referee provides a detailed and accurate summary of the paper's three main results (Theorems A, B, C), the role of Corollary B' in verifying Chen's conjecture, and the methods used (mixed Hodge theory, incidence correspondence, Smith normal form, Fulton's intersection theory). No specific corrections or requests for revision are raised.

Authors: We are grateful for the referee's careful and accurate summary of the paper. We confirm that the referee's account of our results, methods, and the relationship to Chen's conjecture [Che25, Conjecture 1.5] is correct. We also note the referee's observation regarding the independent confirmation by Goncharov [Gon26] via a different method (induction rather than correspondence), which we discuss in Remark 1.2. Since no specific issues are raised, no revision is needed on this point. revision: no
Referee: The referee assesses the paper as making 'a clean contribution at the interface of Arakelov geometry, mixed Hodge theory, and intersection theory,' finds the generalised cross-ratio to be 'a natural and parameter-free invariant,' and evaluates the Smith normal form argument in Lemma 4.2 as 'careful' and the telescoping computation as 'correct.' The recommendation is minor_revision.

Authors: We appreciate the referee's positive assessment of the paper's significance and the verification of correctness for Lemma 4.2. The recommendation of minor revision is noted. As the referee's report does not identify specific points requiring revision, we will conduct a final careful proofreading of the manuscript to address any typographical or expositional issues before the final version. If the referee has specific minor corrections they would like to flag, we would be happy to incorporate them. revision: partial

Circularity Check

0 steps flagged · score 1.0 of 10

No significant circularity; one definitional identity in Theorem C is openly acknowledged by the authors

full rationale

The paper's three main results are independently derived. Theorem A combines Proposition 2.1 (computing the extension class H_G via an explicit dlog integral that evaluates to the cross-ratio) with Theorem 2.2 (showing H_P ≅ H_G via the incidence correspondence and strictness of MHS morphisms). The cross-ratio is defined algebraically (Definition 1.1); the height pairing is defined via extension classes (1.3). Their equality is a genuine computation, not a renaming. Theorem B derives the equality of intersection degree and valuation of the cross-ratio through a non-trivial Smith normal form argument with Fulton's localized Chern classes (Lemma 4.2), using bilinearity of the intersection product. Theorem C is the only step where the identity is close to definitional: c_lim is constructed in Proposition 5.1 as the element corresponding to reg_t(det Ψ) under the determinantal isomorphism (5.1), and then the proof of Theorem C verifies that this construction makes the identity hold for each pair (i,j), with the full identity following by multiplicativity. The paper is transparent: 'The heavy lifting in this theorem is done by the definitions, but the result is suggestive.' However, the construction of c_lim from geometric data (infinitesimal separation of meeting planes) has independent content, and the theorem's role is interpretive rather than predictive. The self-citation to [Bak26] (the first author's thesis) is not load-bearing: the proofs are reproduced and streamlined here, not merely cited. Independent confirmation by Goncharov [Gon26] via a different method (induction vs. correspondence) is noted in Remark 1.2. No circular reduction of inputs to outputs is present.

Assumptions & free parameters 0 free parameters · 3 assumptions · 2 invented entities

No free parameters are introduced. The axioms are standard results from intersection theory and Hodge theory, except for the domain-specific assumption that the incidence correspondence yields an isomorphism of MHS extensions, which is proved (if somewhat tersely) in Theorem 2.2. The invented entities (degenerate and arithmetic cross-ratios) are new constructions with falsifiable predictions.

assumptions (3)

standard math Fulton's intersection theory over regular bases [Ful98, Ch. 14, Ch. 20]
Invoked in Section 4 to define intersection products of cycles over a DVR and compute degrees via localized top Chern classes.
standard math Mixed Hodge structure theory: Ext^1_{Z-MHS}(Z(0),Z(1)) ≃ C× [Car80] and height pairing as obstruction to splitting real MHS [Hai90, Prop. 3.3.7]
Used in the introduction and Section 2 to identify the extension class HP with the augmented height pairing and to derive the Archimedean formula (1.4).
domain assumption The incidence correspondence induces a morphism of MHS that is an isomorphism of extensions
Theorem 2.2 asserts that the morphism ϕ: HP → HG induced by the incidence correspondence is an isomorphism of extensions. The proof checks that generators of graded pieces map correctly but is condensed on full MHS compatibility.

invented entities (2)

Degenerate cross-ratio independent evidence
purpose: To define a cross-ratio for configurations where Ai ∩ Bj ≠ ∅, enabling interpretation of limit heights on special fibres.
The degenerate cross-ratio is a new algebraic construction (Definition 1.3) that produces a falsifiable prediction: the limit period u(0) in a degenerating family equals the degenerate cross-ratio of the central fibre (Corollary C′). This is verified against the asymptotic expansion (1.6).
Arithmetic cross-ratio independent evidence
purpose: To define a canonical cross-ratio on special fibres defined over number fields of class number 1, with norm expressible as lattice indices.
Definition 5.6 and Lemma 5.7 show the norm equals an alternating product of torsion module cardinalities, providing an independent arithmetic check.

how reviews work

0 comments

Cite this review

Pith. "Pith review of A generalised cross-ratio and limits of local heights." pith.science (2026). https://pith.science/paper/H26EK2U2

@misc{pith2026260706477,
  author       = {Pith},
  title        = {Pith review of: A generalised cross-ratio and limits of local heights},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/H26EK2U2}},
  note         = {Machine review of arXiv:2607.06477}
}

abstract

We generalise the standard cross-ratio of four points on a projective line to a cross-ratio of a configuration of four planes in projective $n$-space, the first pair $A_1,\,A_2$ being $k$-dimensional and the second pair $B_1,\,B_2$ being $(n-k-1)$-dimensional, with $A_i \cap B_j = \emptyset$. Over the complex numbers, we show that this cross-ratio equals the augmented height pairing of the corresponding cycles $A_1-A_2, \, B_1-B_2$. Over a discretely valued field, we show that the valuation of the cross-ratio equals the intersection degree of the cycles once they are spread out over the valuation ring. Putting the two together, we conclude that the asymptotics of the Archimedean height pairing of a holomorphic family of configurations are governed by this intersection degree. We also define a degenerate cross-ratio for when $A_i \cap B_j \neq \emptyset$ and interpret the "limit height" of a degenerating holomorphic family of planes as the degenerate cross-ratio of the central plane configuration.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

16 extracted references · 16 canonical work pages

[1]

, title =

Goncharov, Alexander B. , title =. 2026 , eprint =

work page 2026
[2]

Notices of the American Mathematical Society , volume =

Vakil, Ravi , title =. Notices of the American Mathematical Society , volume =. 2003 , pages =

work page 2003
[3]

, title =

Hain, Richard M. , title =. Duke Mathematical Journal , volume =. 1990 , pages =

work page 1990
[4]

Steenbrink, J. H. M. , title =. 1999 , url =

work page 1999
[5]

Fulton, William , title =

work page
[6]

, title =

Bakker, Hidde L. , title =. 2026 , url =

work page 2026
[7]

Algebraic Geometry and Physics , volume =

Beilinson, Alexander , title =. Algebraic Geometry and Physics , volume =. 2025 , pages =. doi:10.4310/AGP.250624000316 , eprint =

work page doi:10.4310/agp.250624000316 2025
[8]

Annales de l'Institut Fourier , volume =

Werner, Annette , title =. Annales de l'Institut Fourier , volume =. 2001 , pages =

work page 2001

Show all 16 references

[9]

Mathematical Research Letters , volume =

Holmes, David and de Jong, Robin , title =. Mathematical Research Letters , volume =. 2015 , pages =

2015
[10]

2025 , eprint =

Chen, Zhelun , title =. 2025 , eprint =

2025
[11]

Heights on curves and limits of

Bloch, Spencer and de Jong, Robin and Sert. Heights on curves and limits of. Journal of the London Mathematical Society , volume =. 2023 , pages =

2023
[12]

Height pairing on higher cycles and mixed

Burgos Gil, Jos. Height pairing on higher cycles and mixed. Proceedings of the London Mathematical Society , volume =. 2022 , pages =

2022
[13]

2025 , eprint =

Holland, Jonathan and Sparling, George , title =. 2025 , eprint =

2025
[14]

Regularized integrals and manifolds with log corners , journal =

Dupont, Cl. Regularized integrals and manifolds with log corners , journal =. 2026 , pages =. doi:10.5802/jep.335 , eprint =

2026 doi
[15]

and Migliorini, Luca , title =

de Cataldo, Mark Andrea A. and Migliorini, Luca , title =. Algebraic Structures and Moduli Spaces , series =. 2004 , pages =. doi:10.1090/crmp/038/05 , eprint =

2004 doi
[16]

, title =

Carlson, James A. , title =. Journées de Géométrie Algébrique d'Angers 1979 , year =

1979

Pith tools

Reviewed July 8, 2026 · model on record in the stance chip above.

[1] [1]

, title =

Goncharov, Alexander B. , title =. 2026 , eprint =

work page 2026

[2] [2]

Notices of the American Mathematical Society , volume =

Vakil, Ravi , title =. Notices of the American Mathematical Society , volume =. 2003 , pages =

work page 2003

[3] [3]

, title =

Hain, Richard M. , title =. Duke Mathematical Journal , volume =. 1990 , pages =

work page 1990

[4] [4]

Steenbrink, J. H. M. , title =. 1999 , url =

work page 1999

[5] [5]

Fulton, William , title =

work page

[6] [6]

, title =

Bakker, Hidde L. , title =. 2026 , url =

work page 2026

[7] [7]

Algebraic Geometry and Physics , volume =

Beilinson, Alexander , title =. Algebraic Geometry and Physics , volume =. 2025 , pages =. doi:10.4310/AGP.250624000316 , eprint =

work page doi:10.4310/agp.250624000316 2025

[8] [8]

Annales de l'Institut Fourier , volume =

Werner, Annette , title =. Annales de l'Institut Fourier , volume =. 2001 , pages =

work page 2001