Torsion order and irrationality of complete intersections

Basic Information

• Paper ID: 2510.24258
• Title: Torsion order and irrationality of complete intersections
• Authors: Jan Lange and Guoyun Zhang
• Classification: math.AG (Algebraic Geometry)
• Publication Date: October 28, 2025
• Paper Link: https://arxiv.org/abs/2510.24258

Abstract

This paper provides new logarithmic lower bounds on the torsion order for very general complete intersections in projective space, as well as very general hypersurfaces in products of projective spaces and Grassmann manifolds. In particular, it proves that their retracts are irrational.

Research Background and Motivation

Research Problem

This paper studies the torsion order (Tor(X)) of algebraic varieties, a central invariant in the rationality problem of algebraic geometry. The torsion order Tor(X) is defined as the smallest positive integer e such that e times the diagonal can be decomposed into a special form (formula 1.1).

Significance

1. Core tool for rationality problems: Torsion order directly relates to determining whether an algebraic variety is rational. If Tor(X) > 1, then X is neither rational nor retract rational.
2. Obstruction to unirational parametrization: Torsion order provides a lower bound on the degree of unirational parametrizations.
3. Characterization of zero-cycle groups: Torsion order is the smallest positive integer making the kernel of the zero-cycle group e-torsion.

Limitations of Existing Methods

• Chatzistamatiou-Levine CL17: For Fano complete intersections, when some degree $d_i \geq m \cdot r(X) + m$, the torsion order is divisible by m (where r(X) is the Fano index). This is a linear bound.
• Schreieder Sch19b, Sch21: Improved to a logarithmic bound $d \geq \log_2(r(X) + m) + m$ for hypersurfaces, but only in that case.

Research Motivation

This paper aims to generalize Schreieder's logarithmic bound to:

1. General complete intersections
2. Hypersurfaces in products of projective spaces
3. Hypersurfaces in Grassmann manifolds

Core Contributions

1. Logarithmic lower bound for complete intersections (Theorem 1.1): Proves that the torsion order of very general complete intersections with multidegree (d₁,...,dₛ) and positive Fano index r is divisible by m if some $d_i \geq \log_2(r + m) + m$.
2. Precise 2-divisibility bound (Theorem 1.2): For complete intersections of dimension ≥4, when $r \leq (d_i + 1)2^{d_i-4} - \lfloor(d_i+2)/2\rfloor$, the torsion order is divisible by 2, proving they are neither rational nor A¹-connected.
3. Hypersurfaces in products of projective spaces (Theorem 1.3): For hypersurfaces of multidegree (d₁,...,dₛ) in $\mathbb{P}^{M_1} \times \cdots \times \mathbb{P}^{M_s}$, when $M_1 \geq 4$ and $(d_1,...,d_s) \geq (\log_2(M_1)+m, M_2+1,...,M_s+1)$, the torsion order is divisible by m.
4. Hypersurfaces in Grassmann manifolds (Theorem 1.4): For hypersurfaces of degree d≥4 in Grassmannian Gr(l,n), when $l(n-l) \leq (d+1)2^{d-4}$, the torsion order is divisible by 2.
5. Methodological innovation: Develops a systematic framework of "affine degeneration," generalizing hypersurface results to complete intersections through iterative application.

Detailed Methods

Core Idea

The key innovation is developing an affine degeneration framework that decomposes the problem into:

1. Constructing affine complete intersections satisfying specific conditions
2. Controlling their projective closures via Gröbner basis theory
3. Using degeneration arguments to transfer divisibility of torsion order

Technical Route

1. Strongly Rational Algebras (Definition 3.1)

A k-algebra B is called strongly k-rational if B is isomorphic to a localization of a polynomial ring. This ensures that Spec B is isomorphic to an open subscheme of affine space.

2. Admissible Pairs (Definition 3.5)

For a smooth k-algebra A, a pair $f_1 \in A[z]$ and $f_2 \in A$ is called admissible if:

• $A[z]/(f_1, f_2)$ is geometrically integral
• Both $A[z]/(f_1)$ and $A/(f_2)$ are strongly k-rational

3. Affine Strictly Semistable Families (Proposition 3.6)

For an admissible pair (f₁, f₂), the constructed family $\text{Spec}(A[z]_{\partial_z f_1} \otimes_k R)/(t - f_1 f_2)$ is a strictly semistable R-scheme (where R = kt₍ₜ₎).

4. Key Construction (Theorem 3.12)

Input: Hypersurface $f \in k[x_1,...,x_{n+r}][z]$ satisfying conditions (C1)-(C4)

• (C1): $B = k[x_1,...,x_{n+r}]/(f_1,...,f_r)$ is strongly k-rational
• (C2): $\text{Spec } B[z]/(f)$ is geometrically integral
• (C3): Relative torsion order is m
• (C4): Satisfies condition (⋆⋆)

Output: For any d ≥ M ≥ 1, construct polynomials $\tilde{f}, \check{f}$ such that:

• $f_1,...,f_{r+1}, \tilde{f}$ satisfy (C1)-(C4) (adding M variables)
• $f_1,...,f_r, \check{f}$ satisfy (C1)-(C4) (degree increased by d)

Proof Strategy:

• Step 1 (d = M): Inductive construction via Example 3.10 and Corollary 3.9
• Step 2 (d > M ≥ 2): Use degeneration argument $t \to 0$ and apply Lemma 2.9

Key Lemmas

Theorem 2.10 (Degeneration Invariance of Relative Torsion Order)

Let X → Spec R be a flat family with special fiber Y = Y₀ ∪ Y₁ and intersection Z = Y₀ ∩ Y₁. If:

1. X° := X \ W_X is strictly semistable
2. Y_i \ W_ ≅ open subscheme of $\mathbb{A}^{\dim Y_i}$

Then $\text{Tor}_\Lambda(Z, W_Z) | \text{Tor}_\Lambda(\bar{X}, W_{\bar{X}})$

Proposition 2.4 (Gröbner Basis and Projective Closure)

If the leading monomials of f₁,...,fᵣ are relatively coprime, then:

• $I^h = (f_1^h,...,f_r^h)$ (homogeneous ideal)
• The projective closure is controllable

Experimental Setup

Basic Examples (Section 4)

4.1 Schreieder's Hypersurface Example

Starting from the construction in Sch21: $f_0 = \pi \left(1 + \sum_{i=1}^n x_i^{\lceil (n+1)/m \rceil}\right)^m - (-1)^n x_1 \cdots x_n + \sum_{j=1}^{N-n} c_j x_{n+j}^m + (-1)^n x_1 \cdots x_n z^m$

Verification (Lemma 4.3): f₀ satisfies condition (⋆⋆), i.e., for any field extension F/k and q ∈ F, there exists an F-algebra surjection: $F[x_1,...,x_N,z,w]/(f_0+q, w\partial_z f_0 - 1) \twoheadrightarrow F$

4.2 Special Quartic Fourfold

Based on the example in HPT18a, construct the polynomial: $f = x_1 z_1^2 + x_2 z_2^2 + x_1 x_2 z_3^2 + (1 + x_1^2 + x_2^2 - 2x_1 - 2x_2 - 2x_1 x_2)$

Key property (Example 4.6): $\text{Tor}_{\mathbb{Z}/2}(\text{Spec } k[x_1,x_2,z_1,z_2,z_3]/(f), W) = 2$

Construction Strategy

Proposition 5.1 (Construction of Complete Intersections)

For N and M satisfying $s-1 \leq M \leq \sum_{i=1}^s d_i - n - m, \quad 4 \leq N \leq n + 2^n - 1 + \sum_{j=0}^{n-1} \binom{n}{j}\lfloor j/m \rfloor$ there exist polynomials $g_1,...,g_s \in k[x_1,...,x_N,y_1,...,y_M]$ such that:

• $\text{Spec } k[x_1,...,x_N,y_1,...,y_M]/(g_1,...,g_s)$ has dimension N+M-s
• $\text{Tor}_{\mathbb{Z}/m}(X,W) = m$
• When s ≤ M, the leading monomials are relatively coprime

Proof in three cases:

• Case a (M ≥ 2s-2): Direct application of Theorem 3.12 (a)
• Case b (s ≤ M < 2s-2): Mixed use of Theorem 3.12 (a) and simple construction
• Case c (M = s-1): Reduction to hypersurface case

Experimental Results

Main Results

Theorem 5.2 (Torsion Order of Complete Intersections)

Let n, m ≥ 2, and let X be a complete intersection with multidegree (d₁,...,dₛ) and dimension D ≥ 4. If $d_1 \geq n+m$ and Fano index $r := D + s + 1 - \sum_{i=1}^s d_i \leq 2n + \sum_{j=0}^{n-1} \binom{n}{j}\lfloor j/m \rfloor - m$ then the torsion order of very general X is divisible by m.

Corollary (Proof of Theorem 1.1): Taking $n = d_1 - m \geq \lceil \log_2(r+m) \rceil$, we have $r \leq 2n - m \leq 2n + \sum_{j=0}^{n-1} \binom{n}{j}\lfloor j/m \rfloor - m$ thus the conclusion holds.

Theorem 5.4 (Small Fano Index Case)

Over a field of characteristic ≠2, a complete intersection X ⊂ $\mathbb{P}^N$ (N ≥ 4+s) with multidegree (d₁,...,dₛ) ∈ $\mathbb{Z}_{\geq 2}^s$ and Fano index r ≤ 2 has torsion order divisible by 2.

Key points of proof:

• Use Proposition 5.3 to construct affine complete intersections
• Apply special examples from Examples 4.6-4.12 (quartic fourfolds, (2,2,2)-complete intersections, (3,3)-fivefolds)
• Apply Gröbner basis theory to control projective closures

Results in Products of Projective Spaces

Theorem 5.7 (Proof of Theorem 1.3)

For hypersurfaces of multidegree (d₀,...,dₛ) in $\mathbb{P}^{M_0} \times \cdots \times \mathbb{P}^{M_s}$, if

• $d_0 \geq n+m$, $d_i \geq M_i+1$ (i=1,...,s)
• $4 \leq M_0 \leq n + 2^n - 1 + \sum_{l=0}^{n-1} \binom{n}{l}\lfloor l/m \rfloor$

then the torsion order of very general hypersurfaces is divisible by m.

Proof strategy:

1. Construct affine hypersurfaces via Proposition 5.6
2. Multihomogenize to obtain hypersurfaces in products of projective spaces
3. Use degeneration arguments to transfer torsion order properties

Results in Grassmann Manifolds

Theorem 5.9 (Generalization of Theorem 1.4)

For the Plücker embedding $\text{Gr}(l,n) \hookrightarrow \mathbb{P}^N$ ($N = \binom{n}{l}-1$), if degree d ≥ n'+m and $4 \leq l(n-l) \leq 2n' - 1 + \sum_{j=0}^{n'-1} \binom{n'}{j}\lfloor j/m \rfloor + d - m$ then the torsion order of the intersection of Gr(l,n) with very general degree d hypersurfaces is divisible by m.

Key technique (Lemma 5.8): Lift affine hypersurfaces in Gr(l,n) to hypersurfaces in $\mathbb{P}^N$, preserving torsion order properties.

Specific Numerical Comparisons

Related Work

History of Torsion Order Research

1. Bloch Blo80, Bloch-Srinivas BS83: Introduced torsion order concept
2. Rojtman Roi80: Finiteness of torsion order for hypersurfaces
3. Chatzistamatiou-Levine CL17: Linear bounds for complete intersections
4. Schreieder Sch19b, Sch21: Logarithmic bounds for hypersurfaces

Methods for Rationality Problems

1. Voisin Voi15: Decomposition of diagonal method
2. Kontsevich-Tschinkel KT19, Nicaise-Shinder NS19: Motivic methods
3. Nicaise-Ottem NO22: Tropical degeneration methods
4. Pavic-Schreieder PS23, Lange-Schreieder LS24: Cycle theory analogs

Positioning of This Paper

• Builds on: Results for hypersurfaces in Sch21, LS24
• Generalizes: From hypersurfaces to complete intersections, products of projective spaces, Grassmann manifolds
• Provides: Unified affine degeneration framework
• Complements: Motivic methods NO22 giving same bounds in positive characteristic

Conclusions and Discussion

Main Conclusions

1. Universality of logarithmic bounds: Generalizes Schreieder's logarithmic bound from hypersurfaces to complete intersections
2. New geometric settings: First treatment of hypersurfaces in products of projective spaces and Grassmann manifolds
3. Methodological contribution: Affine degeneration framework has broad applicability

Limitations

1. Base field restrictions:
   • Complete intersections and products of projective spaces: arbitrary fields
   • Grassmann manifolds: main results primarily over complex numbers (Remark 5.10 indicates possible generalization)
2. Dimension restrictions:
   • Requires dimension ≥4 (Theorem 1.2)
   • Low-dimensional cases need special treatment
3. Fano index:
   • Primarily addresses positive Fano index case
   • Non-Fano complete intersections may have infinite torsion order
4. Missing upper bounds:
   • Only provides lower bounds, not upper bounds on torsion order
   • Upper bounds for hypersurfaces in products of projective spaces unknown (Section 1.2)

Future Directions

1. Upper bound research: Seek upper bounds for torsion order, especially in products of projective spaces
2. Generalization to other rational varieties: Flag varieties, toric varieties
3. Complete positive characteristic theory: Full positive characteristic version for Grassmann manifolds
4. Computational methods: Develop algorithms for explicitly computing torsion order

In-Depth Evaluation

Strengths

1. Technical Innovation

• Affine degeneration framework: Section 3's systematic construction is an important methodological contribution
  • Algebraizes the problem (Definition 3.1, 3.5)
  • Iterative construction (Theorem 3.12) achieves dimension increase from hypersurfaces to complete intersections
  • Introduction of condition (⋆⋆) ensures geometric integrality
• Clever application of Gröbner bases:
  • Proposition 2.4 controls projective closures
  • Relative coprimality of leading monomials guarantees ideal generation

2. Depth of Results

• Breakthrough in logarithmic bounds: Improvement from linear $m \cdot r$ to logarithmic $\log_2(r+m)$
• Unifies multiple previous results:
  • Quartic fivefolds NO22, PS23
  • (2,3)-fourfolds Ska23, FL25
  • New results: (2,2,2)-fourfolds, (3,3)-fivefolds

3. Clarity of Exposition

• Clear structure:
  • Section 2: Preliminaries
  • Section 3: Theoretical framework
  • Section 4: Basic examples
  • Section 5: Applications and main theorems
• Detailed proofs: Key lemmas (e.g., Lemma 4.3, 4.7) provide complete verification

4. Universality

• Multiple geometric backgrounds: Projective spaces, products of projective spaces, Grassmann manifolds
• Arbitrary characteristic: Main results apply to arbitrary characteristic (except Grassmann manifold part)

Weaknesses

1. Method Limitations

• Dependence on special examples: Section 4's constructions depend on explicit equations from Sch21, HPT18a
• Complexity of affine-projective conversion: Requires Gröbner basis theory, not always easy to verify

2. Completeness of Results

• Missing upper bounds: No upper bounds on torsion order provided; unclear if bounds are optimal
• Very general nature: Results may not hold for special complete intersections

3. Computational Complexity

• Condition verification: Verification of conditions (C1)-(C4) requires explicit computation
• Non-uniqueness of construction: Theorem 3.12's construction involves choices; unclear how to optimize

4. Geometric Intuition

• Cost of algebraization: Section 3's algebraic framework lacks geometric intuition
• Unclear connection to motivic methods: Relationship to tropical degeneration method NO22 not sufficiently clarified

Impact

1. Contribution to Field

• Rationality problems: Provides new determination tools, especially for complete intersections
• Methodology: Affine degeneration framework potentially applicable to other problems
• Bridge role: Connects cycle theory methods with explicit constructions

2. Practical Value

• Computability: In principle, can verify specific examples
• Generalization potential: Framework potentially applicable to other classes of rational varieties

3. Reproducibility

• Detailed proofs: Key steps have complete proofs
• Explicit constructions: Section 4 provides verifiable examples
• But lacks code: No computational tools provided

Applicable Scenarios

1. Direct Applications

• Proving non-rationality: When theorem conditions are satisfied, can prove torsion order > 1
• Unirational parametrization: Provides degree lower bounds

2. Theoretical Research

• Rationality problems: New research tools
• Chow group theory: Torsion properties of zero-cycles

3. Generalization Directions

• Other homogeneous spaces: Flag varieties, Schubert varieties
• Moduli spaces: Rationality of certain moduli spaces

Technical Details Supplement

Key Inequalities

For the m=2 case (Theorem 1.2), the key is proving: $2n + \sum_{j=0}^{n-1} \binom{n}{j}\lfloor j/2 \rfloor - 2 = (n+3)2^{n-2} - \lfloor (n+4)/2 \rfloor$

This follows from the combinatorial identity: $\sum_{j=0}^{n-1} \binom{n}{j}\lfloor j/2 \rfloor = (n-1)2^{n-2} - \lfloor n/2 \rfloor$ (LS24, Lemma 7.4).

Role of Condition (⋆)

Condition (⋆) ensures that the k(t)-scheme $\text{Spec}(A[z]_{\partial_z f_1} \otimes_k k(t))/(t - f_1 f_2)$ has a k(t)-rational point, hence by Stacks, Tag 0CDW is geometrically integral. This is crucial for the iterative construction in Corollary 3.9.

Precise Application of Gröbner Bases

In Case b of Proposition 5.1, choose graded lexicographic order: $x_2 > x_3 > \cdots > x_{4+M} > x_1$ so that leading monomials are:

• $\text{LM}(q_1) = x_3^2$
• $\text{LM}(q_2) = x_2^2$
• $\text{LM}(q_i) = x_{3+i}^2$ (i≥3)

relatively coprime, thus Proposition 2.4 applies.

Key References

1. Sch21 S. Schreieder, Torsion order of Fano hypersurfaces, Algebra Number Theory 15 (2021), 241–270.
2. LS24 J. Lange and S. Schreieder, On the rationality problem for low degree hypersurfaces, arXiv:2409.12834.
3. NO22 J. Nicaise and J. C. Ottem, Tropical degenerations and stable rationality, Duke Math. J. 171 (2022), 3023–3075.
4. PS23 N. Pavic and S. Schreieder, The diagonal of quartic fivefolds, Algebr. Geom. 10 (2023), 754–778.
5. CL17 A. Chatzistamatiou and M. Levine, Torsion orders of complete intersections, Algebra Number Theory 11 (2017), 1779–1835.

Overall Assessment: This is a technically sophisticated algebraic geometry paper that achieves substantial progress on the core rationality problem. The systematic development of the affine degeneration framework represents an important methodological contribution, and the achievement of logarithmic bounds represents a significant improvement over previous linear bounds. The paper's main value lies in unifying multiple previous results and extending them to new geometric settings. Despite some limitations (such as missing upper bounds and dependence on special constructions), this is an important advance in the field with expected lasting impact on subsequent research.