Equivariant unirationality of toric varieties

Andrew Kresch; Yuri Tschinkel

Equivariant unirationality of toric varieties

T0 review · 0 major / 4 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read A finite group action on a smooth projective toric variety is equivariantly unirational exactly when the universal-torsor obstruction class vanishes.

desk verdict The paper delivers what it claims: vanishing of ∂(1_Pic) is necessary and sufficient for G-unirationality of toric actions, with a sound proof and honest corrections of prior work. read the letter →

arxiv 2506.07152 v1 pith:YGNKLUD2 submitted 2025-06-08 math.AG

Andrew Kresch , Yuri Tschinkel This is my paper

classification math.AG MSC 14M2514E0814L30

keywords equivariantunirationalitytoricvarietiesuniversaltorsorfinitegroupactionscohomologyAmitsurprojectivetorus

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 proves that equivariant unirationality of a smooth projective toric variety, under a finite group action that preserves the embedded torus, is equivalent to the vanishing of a single computable cohomology class. That class is the obstruction to lifting the group action to a universal torsor, denoted $\partial(1_{\mathrm{Pic}(X)})$ in $H^2(G, \mathrm{Pic}(X)^\vee \otimes k^\times)$. Necessity was already known from general torsor theory; the new content is sufficiency in the toric setting. The proof works by showing that any such action is a translation modification of a toric action, classified by a class in $H^1(G, M^\vee \otimes k^\times)$, and that the universal-torsor obstruction is the image of this class under a connecting homomorphism. When the obstruction vanishes, the lifted class produces a linear action on a torus that dominates $X$, so the criterion is complete.

What carries the argument

The machinery is the equivariant universal torsor obstruction class $\partial(1_{\mathrm{Pic}(X)}) \in H^2(G, \mathrm{Pic}(X)^\vee \otimes k^\times)$, attached to a smooth projective rational $G$-variety through the exact sequence $H^1_G(X, T_{\mathrm{NS}}) \to \mathrm{End}_G(\mathrm{Pic}(X)) \to H^2(G, \mathrm{Pic}(X)^\vee \otimes k^\times)$. The toric computation routes this class through two exact sequences: $0 \to M \to P \to \mathrm{Pic}(X) \to 0$, where $P$ is the permutation module generated by toric divisors, and $1 \to k^\times \to k[T]^\times \to M \to 0$. The extension class $\rho$ of the second sequence is shown to equal the cocycle classifying translation modifications of the toric action; Proposition 5.4 then gives $\partial(1_{\mathrm{Pic}(X)}) = -\sigma$ with $\sigma$ the image of $\rho$ under the connecting homomorphism of the dualized first sequence. This turns the geometric question into group cohomology on the fan: one reads off $M$, $P$, and the action of $G$ from the rays and the translation cocycle, and evaluates a 2-cocycle. The paper notes in Remark 5.3 that an earlier statement (Proposition 4 of [14]) used in this circle of ideas is incorrect as stated and replaces it with Proposition 5.4.

What would settle it

Take a smooth projective toric variety with a finite group action preserving the torus, compute $\partial(1_{\mathrm{Pic}(X)})$ from the fan by the recipe of Proposition 5.4, and test the predicted dominance: if the class vanishes but no $G$-equivariant dominant rational map from any $\mathbb{P}(V)$ to $X$ exists, sufficiency is false. A concrete check would be to implement (6.3) and Proposition 5.4 on a fan where the class vanishes and verify that the lifted cocycle in $H^1(G, P^\vee \otimes k^\times)$ indeed gives a linear action whose projectivization dominates $X$; the paper's Example 6.3 provides the opposite benchmark, a nonzero class and a proof of non-unirationality.

Watch

Extended reading notes

Core claim

The central claim is Theorem 6.1: for a smooth projective toric variety $X$ with dense torus $T$, and a finite group $G$ acting regularly with $T$ stable, $X$ is $G$-unirational if and only if $\partial(1_{\mathrm{Pic}(X)}) = 0$. The forward implication is Proposition 5.1, which holds for all smooth projective rational $G$-varieties. The reverse implication uses the structure of toric actions: automorphisms of $T$ as a variety form a semidirect product $1 \to T(k) \to \mathrm{Aut}(T) \to \mathrm{GL}(M) \to 1$, so a $G$-action that lifts a toric action is determined up to equivalence by a cocycle $(\lambda_g)$ in $H^1(G, M^\vee \otimes k^\times)$. The paper proves in (6.3) that this cocycle is exactly the extension class $\rho$ of $1 \to k^\times \to k[T]^\times \to M \to 0$, and Proposition 5.4 identifies $\partial(1_{\mathrm{Pic}(X)})$ with the negative of the image of $\rho$ under the connecting homomorphism of $0 \to \mathrm{Pic}(X)^\vee \otimes k^\times \to P^\vee \otimes k^\times \to M^\vee \otimes k^\times \to 0$. Vanishing therefore lifts $\rho$ to $H^1(G, P^\vee \otimes k^\times)$, and the lift gives a translation modification of the permutation action on $\mathrm{Spec}(k[P])$ that is linear and dominates $X$.

Load-bearing premise

The sufficiency direction rests on the claim that the way the group action twists the torus by translations is recorded by exactly the cohomology class that appears in the short exact sequence of units of the torus; if that claim failed, a vanishing obstruction would not yield the dominating linear action.

Editorial extensions

If this is right

For any smooth projective toric $X$ with $G$-stable torus, the $G$-unirationality question reduces to evaluating one connecting homomorphism in finite group cohomology; the input is the $G$-action on the fan and the translation cocycle.
Because the obstruction is a stable $G$-birational invariant, equivariant blow-ups along $G$-invariant centers do not change the answer, so the criterion can be checked on any equivariant resolution.
Condition (A) (fixed points for all abelian subgroups) is not sufficient: Example 6.3 satisfies Condition (A) yet has nonzero $\partial(1)$ and is not $G$-unirational.
Projective unirationality of toric $G$-actions is testable by applying Theorem 6.1 to each central cyclic extension of $G$, as indicated in Remark 7.4.

Reading between the lines

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

Editorial inference: The same torsor-lifting obstruction might be sufficient for other classes of rationally connected $G$-varieties whose equivariant birational geometry is generated by torus actions and equivariant blow-ups, though the paper only proves the toric case.
Editorial inference: The explicit identification of the translation cocycle with the extension class makes the criterion algorithmic; one could enumerate fans and cocycles for a fixed group and search for the first vanishing class, producing new examples of $G$-unirational toric varieties in higher dimensions.
Editorial inference: The correction flagged in Remark 5.3 suggests that earlier results relying on the incorrect statement should be revisited; in particular, equivariant universal torsor existence is not equivalent to the elementary obstruction, so descent arguments in other settings may need the corrected comparison.
Editorial inference: Via the known correspondence between equivariant unirationality and unirationality of twists over nonclosed fields, Theorem 6.1 predicts that for toric varieties the universal-torsor obstruction controls unirationality of every twist by a $G$-torsor, a statement that could be tested arithmetically on explicit torsors.

Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Desk Editor's Note

read the letter

Read it. The main result is real: for a smooth projective toric variety with a regular action of a finite group G that preserves the torus, G-unirationality is equivalent to vanishing of the new equivariant obstruction ∂(1_Pic(X)) in H^2(G, Pic(X)^∨ ⊗ k^×). That vanishing was known to be necessary; the sufficiency (Theorem 6.1) is the new part, and the proof is coherent. Proposition 5.4 identifies the obstruction with the negative of the class σ built from the translation cocycle, and the sufficiency step lifts that class to a linear action on an affine space over the torus. The identification (6.3) that the extension class of 1 → k^× → k[U]^× → M → 0 equals the translation cocycle checks out — I looked at the sign and the action on characters, and it is right.

The paper is also honest about its own past: Remark 5.3 states that Proposition 4 of [14] is incorrect and gives a counterexample. That is good practice and doesn't weaken the present argument. Appendix A records the homological algebra compatibilities (signs in connecting homomorphisms, d_2 in spectral sequences) that are easy to get wrong; including them is the right call and raises confidence.

Main soft spot is computational: Examples 4.1 and 6.3 give nontrivial cohomology computations in summarized form. The 5-dimensional D4 example in 6.3 is sketched with a huge fan and a single displayed check (6.4) that the class is 2-torsion and nonzero. No code or machine-checked data is provided. That is a reproducibility gap, but it is not where the main theorem lives. The central proof does not depend on those examples; they illustrate the criterion and exclude a specific action. So the gap is real but minor.

The citation pattern looks fine. The paper builds on [14], [20], [23], [27] and clearly distinguishes its new contribution. The statement about correcting [14, Prop. 4] is specific and verifiable.

Who should read it: anyone working on equivariant birational geometry, torsor methods à la Colliot-Thélène–Sansuc, or toric actions. It is a compact, well-written note that gives a computable criterion.

Send it to a serious referee. The theorem is important enough within the area, the proof is structurally sound, and the remaining computational gaps are a matter of exposition and reproducibility, not correctness.

Referee Report

0 major / 4 minor

Summary. The paper studies equivariant unirationality of smooth projective toric varieties equipped with a regular action of a finite group G that preserves the dense torus T. The authors introduce a torsor-theoretic obstruction class ∂(1_Pic(X)) ∈ H^2(G, Pic(X)∨ ⊗ k×), defined as the obstruction to lifting the G-action to a universal torsor. Their main theorem (Theorem 6.1) states that for such toric varieties, X is G-unirational if and only if ∂(1_Pic(X)) = 0. The proof classifies all G-actions preserving T as translations relative to a toric action, parameterized by H^1(G, M∨ ⊗ k×), and shows via a cohomological identification ((6.3)) that the extension class of the basic sequence 1 → k× → k[T]× → M → 0 equals this translation cocycle. When the obstruction vanishes, the cocycle lifts to H^1(G, P∨ ⊗ k×), defining a linear action on an affine space whose torus Spec(k[P]) maps dominantly to X. The paper also discusses projective unirationality and provides two substantial computational examples.

Significance. The paper gives a beautifully concrete and computable criterion for equivariant unirationality in the toric setting, reducing the geometric problem to a finite group cohomology calculation. This is a substantial advance: previously only necessary cohomological obstructions (Amitsur groups) were known in this generality, while here the same type of invariant is proved to be sufficient. The proof is coherent and self-contained, with a helpful homological algebra appendix. A notable strength is the correction of an erroneous statement in [14], demonstrating careful engagement with prior work. The main theorem is machine-checkable in principle, and the two examples, while computational, illustrate the criterion's power by excluding G-unirationality in concrete cases. If the sketched computations are independently verified, the paper provides new non-unirational equivariant actions.

minor comments (4)

[Section 6, proof of Theorem 6.1] The sentence 'By an argument similar to the one above, a cocycle representative determines a modification by translations of the right permutation action on the torus Spec(k[P])' is quite compressed; the equivariance of the induced dominant rational map Spec(k[P]) 99K X with respect to the constructed linear action is the heart of the sufficiency direction and deserves at least one additional explanatory sentence.
[Example 6.3] There is a typo: 'implemention' should be 'implementation'. More substantively, the group cohomology computation is only sketched and is difficult for a reader to verify without access to the underlying code or the full data; since the example is not used in the proof of the main theorem, this is acceptable, but a note on reproducibility would be helpful.
[Proposition 5.1] In the proof of the second assertion, the diagram chase is terse; explicitly labeling the vertical maps (precomposition and postcomposition by π*) would make the argument easier to follow.
[Example 4.1] The formulas for the group action contain several signs and fractions that are easy to misread; a brief indication of how Condition (A) is verified for this action would improve the example's self-containedness.

Circularity Check

0 steps flagged · score 1.0 of 10

No circularity: the obstruction is defined independently and the sufficiency proof constructs the dominating linear action from its vanishing.

full rationale

The paper's central claim (Theorem 6.1) is not circular. The obstruction ∂(1_Pic(X)) is defined in (5.2) purely cohomologically via the Leray spectral sequence and the universal torsor formalism, with no reference to unirationality. Necessity is Proposition 5.1. Sufficiency starts from the assumed vanishing ∂(1_Pic(X))=0. By Proposition 5.4 this is −σ, the image under the connecting homomorphism of (5.4) of the extension class ρ of 1 → k× → k[U]× → M → 0. The crucial identification (6.3), ρ=[(λ_g)], is verified directly in the text: the action sends (χ↦χ) to (χ↦λ_g(χ)χ), yielding the cocycle representative (λ_g^{-1}) of −ρ. This is a direct computation, not an assumption equivalent to the conclusion. Vanishing of σ gives a lift of ρ to H^1(G,P∨⊗k×); a cocycle representative defines a linear (monomial) action on Spec(k[P]) by g·x_i=μ_g(i)x_{σ_g(i)}, and the toric morphism Spec(k[P])→X is dominant and equivariant up to translation, which yields G-unirationality by definition. No parameter is fitted to the target property. The paper explicitly corrects an earlier proposition from [14] (Remark 5.3), showing self-citation is not used to force the result. Reliance on the authors' prior equivariant weak factorization [21] for birational invariance is a genuine external theorem and not the target claim. Large computations in Examples 4.1 and 6.3 are unverified but do not enter the logical derivation of the main theorem. Consequently there is no self-definitional collapse, no fitted-input prediction, and no load-bearing self-citation chain.

Assumptions & free parameters 0 free parameters · 5 assumptions · 0 invented entities

No fitted parameters or hand-chosen constants enter the proof; the only inputs are the toric fan, the finite group, and the character lattice. The axioms listed are established tools in algebraic geometry and homological algebra, none of which is invented for this paper. The new obstruction is a cohomology class, not a new physical or geometric entity.

assumptions (5)

standard math Equivariant weak factorization: any two smooth projective G-varieties that are G-birational are connected by a sequence of blow-ups and blow-downs along smooth G-invariant centers.
Used in Proposition 5.1 to reduce G-birational invariance of Condition (5.2) to invariance under one blow-up; cited to [21].
standard math Aut(T) ≅ T(k) ⋊ GL(M) for an algebraic torus T with character lattice M.
Section 6, equation (6.1), underlies the classification of G-actions preserving T into a linear part and a translation cocycle in H^1(G, M^∨ ⊗ k^×).
domain assumption For a torus T with G-stable dense torus in X, X arises from a smooth projective G-invariant fan.
Invoked in Section 6 to set up Theorem 6.1; cited to [7]. The fan gives the permutation module P and the exact sequence (4.1).
standard math Standard group cohomology identifications and sign conventions, including Ext^n(M,N) ≅ H^n(G, M^∨⊗N) for torsion-free M and the compatibility δ∘δ′ + δ′∘δ = 0.
The appendix states and proves the needed compatibilities (Lemmas A.1, A.2, A.3); the main text uses them in Propositions 5.4 and 6.1.
standard math Every irreducible representation of a finite group appears in a tensor power of a faithful representation (Burnside).
Used in Proposition 7.2 to pass from projective unirationality to dominance by sums of copies of a fixed representation; cited to [26].

how reviews work

0 comments

Cite this review

Pith. "Pith review of Equivariant unirationality of toric varieties." pith.science (2026). https://pith.science/paper/YGNKLUD2

@misc{pith2026250607152,
  author       = {Pith},
  title        = {Pith review of: Equivariant unirationality of toric varieties},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/YGNKLUD2}},
  note         = {Machine review of arXiv:2506.07152}
}

read the original abstract

We introduce a torsor-theoretic obstruction to equivariant unirationality and show that it is also sufficient for actions of finite groups on toric varieties arising from automorphisms of the torus.

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

27 extracted references · 24 canonical work pages

[14]

Hassett and Yu

B. Hassett and Yu. Tschinkel. Torsors and stable equivariant birational geometry. Nagoya Math. J. , 250:275–297, 2023

work page 2023
[20]

Kresch and Yu

A. Kresch and Yu. Tschinkel. Cohomology of finite subgroups of the plane Cremona group, 2022. arXiv:2203.01876, to appear in Algebraic Geom. and Physics

arXiv 2022
[23]

Kresch and Yu

A. Kresch and Yu. Tschinkel. Unramified Brauer group of quotient spaces by finite groups. J. Algebra, 664:75–100, 2025

work page 2025
[27]

Tschinkel and Zh

Yu. Tschinkel and Zh. Zhang. Cohomological obstructions to equivariant unira- tionality, 2025. arXiv:2504.10204. Institut f¨ur Mathematik, Universit¨at Z¨urich, Winterthurerstrasse 190, CH-8057 Z¨urich, Switzerland Email address : andrew.kresch@math.uzh.ch Courant Institute, 251 Mercer Street, New York, NY 10012, USA Email address : tschinkel@cims.nyu.edu...

arXiv 2025
[1]

Adem and R

A. Adem and R. J. Milgram. Cohomology of finite groups , volume 309 of Grundlehren der mathematischen Wissenschaften [Fundamental Principles of Mathematical Sciences]. Springer-Verlag, Berlin, second edition, 2004

work page 2004
[2]

Blanc, I

J. Blanc, I. Cheltsov, A. Duncan, and Yu. Prokhorov. Finite quasisimple groups acting on rationally connected threefolds. Math. Proc. Cambridge Philos. Soc. , 174(3):531–568, 2023

work page 2023
[3]

Bogomolov and Yu

F. Bogomolov and Yu. Prokhorov. On stable conjugacy of finite subgroups of the plane Cremona group, I. Cent. Eur. J. Math. , 11(12):2099–2105, 2013

work page 2013
[4]

K. S. Brown. Cohomology of groups, volume 87 ofGraduate Texts in Mathematics. Springer-Verlag, New York, 1994. Corrected reprint of the 1982 original

work page 1994

Show all 27 references

[5]

Cartan and S

H. Cartan and S. Eilenberg. Homological algebra. Princeton University Press, Princeton, NJ, 1956

1956
[6]

Cheltsov, Yu

I. Cheltsov, Yu. Tschinkel, and Zh. Zhang. Equivariant unirationality of Fano threefolds, 2025. arXiv:2502.19598

2025 arXiv
[7]

Colliot-Th´ el` ene, D

J.-L. Colliot-Th´ el` ene, D. Harari, and A. N. Skorobogatov. Compactifica- tion ´ equivariante d’un tore (d’apr` es Brylinski et K¨ unnemann).Expo. Math. , 23(2):161–170, 2005

2005
[8]

Colliot-Th´ el` ene and J.-J

J.-L. Colliot-Th´ el` ene and J.-J. Sansuc. La descente sur les vari´ et´ es rationnelles. II. Duke Math. J. , 54(2):375–492, 1987

1987
[9]

B. Conrad. Grothendieck duality and base change , volume 1750 of Lecture Notes in Mathematics . Springer-Verlag, Berlin, 2000

2000
[10]

Deligne and D

P. Deligne and D. Mumford. The irreducibility of the space of curves of given genus. Inst. Hautes ´Etudes Sci. Publ. Math. , (36):75–109, 1969

1969
[11]

A. Duncan. Equivariant unirationality of del Pezzo surfaces of degree 3 and 4. Eur. J. Math. , 2(4):897–916, 2016

2016
[12]

Duncan and Z

A. Duncan and Z. Reichstein. Versality of algebraic group actions and rational points on twisted varieties. J. Algebr. Geom. , 24(3):499–530, 2015

2015
[13]

Grothendieck

A. Grothendieck. Sur quelques points d’alg` ebre homologique. Tohoku Math. J. (2), 9:119–221, 1957

1957
[15]

Heller and I

A. Heller and I. Reiner. Representations of cyclic groups in rings of integers. I. Ann. of Math. (2) , 76:73–92, 1962

1962
[16]

P. J. Hilton and U. Stammbach. A course in homological algebra , volume 4 of Graduate Texts in Mathematics . Springer-Verlag, New York, second edition, 1997

1997
[17]

Hoshi and A

A. Hoshi and A. Yamasaki. Rationality problem for algebraic tori. Mem. Amer. Math. Soc., 248(1176):v+215, 2017

2017
[18]

I. M. Isaacs. Character theory of finite groups , volume No. 69 of Pure and Ap- plied Mathematics . Academic Press [Harcourt Brace Jovanovich, Publishers], New York-London, 1976. 22 A. KRESCH AND YU. TSCHINKEL

1976
[19]

Kresch and Yu

A. Kresch and Yu. Tschinkel. Effectivity of Brauer-Manin obstructions. Adv. Math., 218(1):1–27, 2008

2008
[21]

Kresch and Yu

A. Kresch and Yu. Tschinkel. Equivariant birational types and Burnside volume. Ann. Sc. Norm. Super. Pisa Cl. Sci. (5) , 23(2):1013–1052, 2022

2022
[22]

Kresch and Yu

A. Kresch and Yu. Tschinkel. Equivariant Burnside groups and toric varieties. Rend. Circ. Mat. Palermo (2) , 72(5):3013–3039, 2023

2023
[24]

B. `E. Kunyavski ˘ ı. Three-dimensional algebraic tori. InInvestigations in number theory (Russian), pages 90–111. Saratov. Gos. Univ., Saratov, 1987. Translated in Selecta Math. Soviet. 9 (1990), no. 1, 1–21

1990
[25]

Yu. I. Manin. Rational surfaces over perfect fields. II. Mat. Sb. (N.S.) , 72(114):161–192, 1967

1967
[26]

Steinberg

R. Steinberg. Complete sets of representations of algebras. Proc. Amer. Math. Soc., 13:746–747, 1962

1962

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.

[1] [14]

Hassett and Yu

B. Hassett and Yu. Tschinkel. Torsors and stable equivariant birational geometry. Nagoya Math. J. , 250:275–297, 2023

work page 2023

[2] [20]

Kresch and Yu

A. Kresch and Yu. Tschinkel. Cohomology of finite subgroups of the plane Cremona group, 2022. arXiv:2203.01876, to appear in Algebraic Geom. and Physics

arXiv 2022

[3] [23]

Kresch and Yu

A. Kresch and Yu. Tschinkel. Unramified Brauer group of quotient spaces by finite groups. J. Algebra, 664:75–100, 2025

work page 2025

[4] [27]

Tschinkel and Zh

Yu. Tschinkel and Zh. Zhang. Cohomological obstructions to equivariant unira- tionality, 2025. arXiv:2504.10204. Institut f¨ur Mathematik, Universit¨at Z¨urich, Winterthurerstrasse 190, CH-8057 Z¨urich, Switzerland Email address : andrew.kresch@math.uzh.ch Courant Institute, 251 Mercer Street, New York, NY 10012, USA Email address : tschinkel@cims.nyu.edu...

arXiv 2025

[5] [1]

Adem and R

A. Adem and R. J. Milgram. Cohomology of finite groups , volume 309 of Grundlehren der mathematischen Wissenschaften [Fundamental Principles of Mathematical Sciences]. Springer-Verlag, Berlin, second edition, 2004

work page 2004

[6] [2]

Blanc, I

J. Blanc, I. Cheltsov, A. Duncan, and Yu. Prokhorov. Finite quasisimple groups acting on rationally connected threefolds. Math. Proc. Cambridge Philos. Soc. , 174(3):531–568, 2023

work page 2023

[7] [3]

Bogomolov and Yu

F. Bogomolov and Yu. Prokhorov. On stable conjugacy of finite subgroups of the plane Cremona group, I. Cent. Eur. J. Math. , 11(12):2099–2105, 2013

work page 2013

[8] [4]

K. S. Brown. Cohomology of groups, volume 87 ofGraduate Texts in Mathematics. Springer-Verlag, New York, 1994. Corrected reprint of the 1982 original

work page 1994