Failure of the invariant cycle theorem over mathbb Z
Pith reviewed 2026-05-21 13:52 UTC · model grok-4.3
The pith
A family of surfaces with nontrivial Albanese varieties gives the first counterexample to the invariant cycle theorem over the integers.
A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.
Core claim
We construct the first example of a semistable family which fails the local (and global) invariant cycle theorems with integral coefficients. The family has constant period map associated to H^2, and its smooth fibers are algebraic surfaces with p_g=q=1; in particular, they have non-trivial Albanese varieties. The surfaces in the family have maximal Picard rank and minimal discriminant, and they are closely related to Vinberg's most algebraic K3 surface. Our construction also generalizes the Shioda--Inose construction for rational double covers of K3 surfaces.
What carries the argument
The explicit construction of a semistable family of surfaces with p_g=q=1 and nontrivial Albanese variety, obtained by generalizing the Shioda-Inose construction and related to Vinberg's K3 surface.
Load-bearing premise
The family must be semistable, have constant period map for H^2, and consist of surfaces with nontrivial Albanese varieties for the failure over Z to occur in this setup.
What would settle it
Direct computation of the specialization map on integral cohomology for this family, checking whether the invariant cycles are exactly the image from the total space.
read the original abstract
We initiate a study of the local invariant cycle theorem with integral coefficients for 1-parameter semistable families of varieties. We show that it always holds for $H^1$, and it holds for $H^2$ if the general fiber has trivial Albanese variety. The latter generalizes results of Friedman, Griffiths, and Scattone on K3 surfaces and I-surfaces. We construct the first example of a semistable family which fails the local (and global) invariant cycle theorems with integral coefficients. The family has constant period map associated to $H^2$, and its smooth fibers are algebraic surfaces with $p_g=q=1$; in particular, they have non-trivial Albanese varieties. The surfaces in the family have maximal Picard rank and minimal discriminant, and they are closely related to Vinberg's most algebraic K3 surface. Our construction also generalizes the Shioda--Inose construction for rational double covers of K3 surfaces.
Editorial analysis
A structured set of objections, weighed in public.
Referee Report
Summary. The paper studies the local invariant cycle theorem with integral coefficients for one-parameter semistable families. It proves that the theorem holds unconditionally for H^1 and holds for H^2 whenever the general fiber has trivial Albanese variety, generalizing results of Friedman-Griffiths-Scattone for K3 surfaces and I-surfaces. The central contribution is an explicit construction of a semistable family of algebraic surfaces with p_g = q = 1 (hence non-trivial Albanese varieties) whose H^2 period map is constant; the authors show that this family violates both the local and global invariant cycle theorems over ℤ. The surfaces have maximal Picard rank and minimal discriminant and arise from a generalization of the Shioda-Inose construction relating them to Vinberg's most algebraic K3 surface.
Significance. If the geometric construction and lattice-theoretic verification are correct, the result is significant: it supplies the first concrete counterexample to the integral invariant cycle theorem in the semistable setting and identifies the non-trivial Albanese variety as the obstruction. The explicit, verifiable nature of the construction (via a concrete geometric recipe and direct comparison of integral lattices) provides a falsifiable example that can be checked independently, strengthening its value for the study of integral Hodge theory and degenerations.
major comments (1)
- [§4.3] §4.3, construction of the family: the relation to Vinberg's K3 surface is used to detect the failure of the integral invariant cycle map, but the precise statement of how the monodromy action on the integral cohomology of the family differs from the invariant part (including the explicit cokernel) is only sketched; a short table or diagram comparing the relevant lattices would make the failure load-bearing and verifiable.
minor comments (3)
- [§2] The notation for the period map and the Albanese variety in §2 should be made uniform with the lattice notation used later in §5.
- [Figure 1] Figure 1 (the degeneration diagram) would benefit from labeling the special fiber and indicating which cohomology classes are invariant.
- [Introduction] A brief remark on why the construction does not contradict the known rational invariant cycle theorem would clarify the scope for readers.
Simulated Author's Rebuttal
We thank the referee for their positive assessment and constructive suggestion. We address the major comment below and will incorporate the recommended clarification in the revised version.
read point-by-point responses
-
Referee: [§4.3] §4.3, construction of the family: the relation to Vinberg's K3 surface is used to detect the failure of the integral invariant cycle map, but the precise statement of how the monodromy action on the integral cohomology of the family differs from the invariant part (including the explicit cokernel) is only sketched; a short table or diagram comparing the relevant lattices would make the failure load-bearing and verifiable.
Authors: We agree that an explicit comparison of the lattices would make the argument more transparent and independently verifiable. In the revised manuscript we will insert a short table (or diagram) in §4.3 that displays the full integral cohomology lattice of a general fiber, the monodromy-invariant sublattice, the image of the specialization map from the total space, and the resulting cokernel. The table will also record the relation to the lattice of Vinberg’s most algebraic K3 surface, thereby exhibiting precisely how the constant period map produces a non-trivial cokernel and hence violates the integral invariant cycle theorem. This addition clarifies the existing lattice-theoretic verification without changing any statements or proofs. revision: yes
Circularity Check
No significant circularity; derivation self-contained via explicit construction
full rationale
The paper proves the invariant cycle theorem holds for H^1 in any semistable family and for H^2 when the general fiber has trivial Albanese variety, using direct arguments that do not reduce to fitted parameters or prior self-citations of the main result. The central negative claim is established by an explicit geometric construction of a semistable family of surfaces with p_g = q = 1, constant H^2 period map, and non-trivial Albanese varieties; this family is obtained by generalizing the Shioda-Inose construction and is related to Vinberg's most algebraic K3 surface. Failure of the integral invariant-cycle map is detected by direct comparison of integral lattices rather than by any self-definitional step, fitted-input prediction, or load-bearing self-citation chain. The derivation therefore remains independent of its own outputs and is supported by verifiable geometric data.
Axiom & Free-Parameter Ledger
axioms (1)
- domain assumption Standard properties of semistable degenerations and Hodge structures in algebraic geometry
discussion (0)
Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.