REVIEW 3 major objections 4 minor 2 cited by
Free actions on products of real projective spaces
T0 review · 3 major / 4 minor · reviewed 2026-08-07 · deepseek-v4-flash
Pith's one-line read If a finite-dimensional complex homotopy equivalent to a product of real projective spaces admits a free cellular (Z/2)^r-action with trivial mod-2 cohomology action, then r ≤ mu(n_1)+⋯+mu(n_k).
desk verdict Solid, substantial paper that proves the homotopy-theoretic Cusick conjecture in full, corrects a genuine error in the author's earlier work, and introduces a usable Bockstein-transgression comparison; deserves serious refereeing. read the letter →
The pith
A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.
The reading
What carries the argument
The load-bearing object is the Borel fibration X → X_G → BG and its Serre spectral sequence in both mod-2 and integral coefficients. The k-invariants are the classes alpha_i = d_2(1 ⊗ t_i) in $H^{2}$(G;F_2) recording how the degree-one generators t_i of H^*(X;F_2) are hit by the first differential. The new mechanism is the transgression–Bockstein comparison (Proposition 9 and Corollary 11): if a class in the fiber is transgressive in the F_2 spectral sequence, then its integral Bockstein beta_0 is transgressive in the integral spectral sequence and tau_Z(beta_0(x)) = beta_0(tau_{F_2}(x)). This lets the proof pass from d_2(t_i) = alpha_i to d_3(beta_0(t_i)) = beta_0(alpha_i), and the product structure of the spectral sequence extracts the factorization of alpha_i.
What would settle it
Run the computation of the integral d_3 differential in a concrete case where both sides of the transgression–Bockstein identity are known, such as the free D_8/(Z/2) action on $RP^{{4m+1}}$ described in Example 18, and check whether d_3(beta_0(t)) equals beta_0(d_2(t)) in the E_3-term; a mismatch would disprove the comparison. Alternatively, construct a free cellular (Z/2)^r-action on a product of real projective spaces with trivial mod-2 cohomology action and r exceeding mu(n_1)+⋯+mu(n_k), directly falsifying Theorem 2.
Extended reading notes
Core claim
The central claim, Theorem 2, is that for a free cellular (Z/2)^r-action on a finite-dimensional CW complex homotopy equivalent to a product of real projective spaces with trivial action on mod-2 cohomology, the rank satisfies r ≤ mu(n_1)+⋯+mu(n_k). The discovery that carries the proof is Proposition 3: for each factor with n_i ≡ 1 mod 4, the k-invariant alpha_i = d_2(t_i) in $H^{2}$(G;F_2) must factor as l_i(l_i+l'_i) for some one-dimensional classes l_i, l'_i, and it is a square when the action on integral cohomology is trivial. This is reached through the transgression–Bockstein comparison, which translates the mod-2 differential into an integral differential d_3(beta_0(t_i)) = beta_0(alpha_i) and then shows a relevant differential vanishes. Combined with the no-common-zeros obstruction for the k-invariants, this forces the rank inequality. The paper also demonstrates that the bound is sharp and corrects an earlier argument that required the action on integral cohomology to be trivial.
Load-bearing premise
The load-bearing premise is the transgression–Bockstein comparison (Proposition 9, Corollary 11), that the integral Bockstein of a transgressive class is transgressive with target the Bockstein of the original target, which could fail due to a hidden sign or convergence condition.
Editorial extensions
If this is right
- The conjecture holds for spaces homotopy equivalent to products of real projective spaces, removing earlier restrictions on the factors and on triviality of the action in integral cohomology.
- The upper bound is sharp: the maximum free rank equals mu(n_1)+⋯+mu(n_k), realized by products of the quaternion and rotation actions described in the paper.
- For factors with n_i ≡ 1 mod 4, the k-invariants are constrained to factor as l_i(l_i+l'_i), and they are squares when the integral action is trivial; a nonzero beta_0(alpha_i) always equals a nonzero multiple l_i alpha_i.
- A gap in an earlier proof is identified and corrected: the relevant earlier lemma and theorem require the action on integral cohomology to be trivial, not merely the action on mod-2 cohomology.
- As a corollary, the rank bound is equivalent to the algebraic statement that the k-invariants, as quadratic forms, have no common zeros, limiting the dimension of any subspace on which they vanish simultaneously.
Reading between the lines
- The transgression–Bockstein comparison is a general identity for Serre spectral sequences, so the same strategy may apply to free-action rank problems on other spaces with truncated polynomial cohomology, such as products of spheres, Dold spaces, or Milnor manifolds.
- The factorization of alpha_i as a rank-at-most-2 quadratic form suggests the bound b+2c is really a dimension count on the zero locus of the k-invariants; one could try to realize each possible degeneracy pattern with explicit actions to test whether the b+2c estimate is optimal factor-by-factor.
- Because the proof uses only the spectral-sequence structure and the cohomology ring, the theorem may extend to finite CW complexes with the same mod-2 cohomology ring as a product of real projective spaces, which would bring it closer to the original conjecture's wording.
Editorial analysis
A structured set of objections, weighed in public.
Referee Report
Summary. The paper proves a homotopy-theoretic version of Cusick's conjecture: if G=(Z/2)^r acts freely and cellularly on a finite-dimensional CW-complex X homotopy equivalent to a product RP^{n_1} × ... × RP^{n_k}, with trivial action on mod-2 cohomology, then r ≤ μ(n_1)+...+μ(n_k), where μ(n)=0,1,2 according as n is even, ≡1 mod 4, or ≡3 mod 4. The proof uses the Serre spectral sequence of the Borel construction, with the new input being a comparison (Proposition 9 and Corollary 11) between transgression in the mod-2 and integral spectral sequences via the integral Bockstein operator. The k-invariants of the action are shown to have a specific quadratic form, and a no-common-zeros argument yields the rank bound.
Significance. If correct, the main theorem settles a conjecture of Cusick in the homotopy-theoretic setting, extending earlier results that required additional assumptions such as triviality of the action on integral cohomology. The paper's new transgression–Bockstein comparison is a natural and useful tool, and the proof includes self-contained derivations of several earlier lemmas. The paper also supplies examples showing sharpness of the bound, and it explicitly corrects an error in the author's previous work [18], which is a valuable contribution to the record. The main spectral sequence arguments are carefully written and the central idea is sound, but two proof gaps described below need to be addressed.
major comments (3)
- [Section 2, Lemma 4] Lemma 4 is false as stated and its proof is invalid: a finite-dimensional CW-complex need not have finite-dimensional mod-2 cohomology (for example, an infinite wedge of circles is 1-dimensional but has infinite H^1). The assertion that E∞ is a finite-dimensional F2-vector space does not follow from the fact that X/G is finite-dimensional. In the application in Proposition 7, the needed contradiction can be obtained without Lemma 4: if the spectral sequence collapsed for a cyclic subgroup C, then E∞^{p,0}=H^p(C;F2) is nonzero for every p, whereas convergence and the finite-dimensionality of X/C force E∞^{p,q}=0 for p+q>dim X. The lemma should be restated with the additional hypothesis that H^*(X;F2) is finite-dimensional and the proof should use the boundedness of total degree.
- [Section 4, Lemma 15] In the proof of Lemma 15, after deriving m_2(θ_i)=Σ_{p=1}^k l_p ⊗ t_i^{4m_i} t_p, the text claims that applying the Bockstein operator gives 0=Σ_p l_p ⊗ t_i^{4m_i} t_p^2, which implies l_p=0 for all p. This implication is not valid: for p=i, t_i^{4m_i}t_i^2 = t_i^{4m_i+2}=0, so the equation only forces l_p=0 for p≠i. To conclude m_2(θ_i)=0 one must additionally use that θ_i lies in the torsion part H^1(G;T^{4m_i+1}) and that the monomial t_i^{4m_i+1} is the reduction of the free generator v_i, hence is not in the image of the torsion subgroup under mod-2 reduction. This step should be rewritten, for example by separating the free and torsion components of θ_i as is done later in the proof of Proposition 3.
- [Section 3 and Section 5] Proposition 9 and Corollary 11 are stated for fibrations with constant coefficients, but in Section 5 they are applied to the Borel fibration where the integral cohomology of the fiber carries a possibly nontrivial local coefficient system. The proof of Proposition 9 is a chain-level diagram chase and can be adapted to coefficient modules, and the classes s_i=β_0(t_i) are invariant under the G-action, so the transgression comparison is expected to remain valid. However, the paper does not state this extension or justify it in detail, and the mismatch between the constant-coefficient statement and the local-coefficient application should be addressed explicitly.
minor comments (4)
- [Title page] There is a typo in the title display: 'SP ACES' should be 'SPACES'.
- [Section 5, proof of Proposition 3] The first sentence of the proof says 'finite CW-complex', but the theorem assumes only a finite-dimensional G-CW-complex; the wording should be made consistent.
- [Section 4, Lemma 13] The diagram in the proof uses a short exact sequence for H^n(Bπ;Z) that is stated without proof; it would be helpful to note that this follows from the universal coefficient theorem and the fact that H^n(Bπ;Z) is 2-torsion for n≥1.
- [Section 4, Proposition 14] The proof of Proposition 14 is quite dense; a few more details on why the listed relations are complete would improve readability, especially for the verification that no additional relations occur.
Circularity Check
No significant circularity: the rank bound is derived from k-invariants defined by spectral-sequence differentials, with the new transgression–Bockstein comparison proved inside the paper.
full rationale
The main claim, Theorem 2, is not circular. The k-invariants alpha_i are defined directly as the F2 Serre spectral sequence differentials d2(t_i) = alpha_i (Definition 5), and the rank bound is derived from their algebraic form together with the no-common-zeros condition, not assumed in advance. The key new input, Proposition 9 and Corollary 11, is proved within the paper by an explicit chain-level diagram chase (Lemma 10), giving tau_Z(beta_0(x)) = beta_0(tau_F2(x)). Lemma 15 and Proposition 3 then use this identity to show that when n_i ≡ 1 mod 4, beta_0(alpha_i) = 0 (under integral-trivial action) or beta(alpha_i) = l_i alpha_i (in the general case), from which alpha_i = l_i(l_i + l_i') follows by the algebraic Lemma 21. These conclusions are forced by the differentials and injectivity arguments, not by the target inequality. The no-common-zeros input is cited to Carlsson's external theorem and Cusick's Proposition B/D, not to the present author's prior work. The author's earlier paper [18] is explicitly corrected in Remark 20 rather than used as a load-bearing citation. The sharpness examples (Examples 16–19) are independent constructions. The only in-text caveat is that Lemma 4's proof overstates that finite-dimensionality alone implies finite-dimensional F2-cohomology; in the actual applications H*(X;F2) is finite and the free action setting supplies the needed finite-dimensional E-infinity by standard arguments, so this is a presentation issue, not a circular step. Overall, no fitted quantity is renamed as a prediction and no conclusion reduces by construction to its own input.
Assumptions & free parameters
assumptions (6)
- standard math Serre spectral sequence for the Borel fibration, including product structure, transgression, and naturality under subgroup inclusions
- standard math Integral cohomology ring of (Z/2)^r is generated by the classes u_I with the Benson-Carlson relations
- domain assumption Cusick's Prop D: for c quadratic forms in r variables over F2 with no common zeros, r ≤ 2c
- domain assumption Jo-Lee Theorem 1.3: rk_2 G ≤ rk_2(pi/F) + l for the extension associated to the covering in Proposition 22
- standard math Free action implies X_G ≃ X/G, so E_∞ is finite-dimensional when X is a finite-dimensional free G-CW-complex
- standard math Bockstein-transgression comparison (Corollary 11) derived from diagram chase in Proposition 9
Cite this review
Pith. "Pith review of Free actions on products of real projective spaces." pith.science (2026). https://pith.science/paper/PDC5ROMQ
@misc{pith2026250604067,
author = {Pith},
title = {Pith review of: Free actions on products of real projective spaces},
year = {2026},
howpublished = {\url{https://pith.science/paper/PDC5ROMQ}},
note = {Machine review of arXiv:2506.04067}
}
abstract
We prove that if $G=(\mathbb{Z}/2)^r$ acts freely and cellularly on a finite-dimensional CW-complex $X$ homotopy equivalent to $\mathbb{R}P ^{n_1} \times \cdots \times \mathbb{R} P ^{n_k}$ with trivial action on the mod-$2$ cohomology, then $r \leq \mu (n_1)+ \cdots + \mu(n_k )$ where for each integer $n\geq 0$, $\mu (n)=0$ if $n$ is even, $\mu(n)=1$ if $n\equiv 1$ mod 4, and $\mu(n)=2$ if $n\equiv 3$ mod 4. This proves a homotopy-theoretic version of a conjecture of Cusick.
Forward citations
Cited by 2 Pith papers
-
Carlsson's Conjecture and the Generalized Total Rank Conjecture in Characteristic Two
Over regular rings of characteristic 2, every bounded free complex with nonzero homology has rank at least 2^{codim}; Carlsson's conjecture in every rank and the sphere rank problem follow.
-
Finite abelian group actions on weakly Lefschetz cohomologically symplectic manifolds
Free actions of large finite abelian groups on weakly Lefschetz cohomologically symplectic manifolds force total Betti number at least 2^k, with a torus-like structure theorem for elementary abelian p-groups.
Reference graph
Works this paper leans on
-
[18]
Yal¸ cın,Group actions and group extensions, Trans
E. Yal¸ cın,Group actions and group extensions, Trans. Amer. Math. Soc.352(2000), 2689–2700. Department of Mathematics, Bilkent University, 06800 Bilkent, Ankara, Turkey Email address:yalcine@fen.bilkent.edu.tr
work page 2000
-
[1]
A. Adem and W. Browder,The free rank of symmetry on(S n)k, Invent. Math.92(1988), 431-440
work page 1988
-
[2]
A. Adem and E. Yal¸ cın,On some examples of group actions and group extensions, J. Group Theory 2 (1999), 69-79
work page 1999
-
[3]
D. Benson and J. Carlson,The cohomology of extraspecial groups, Bull. London Math. Soc. 24(1992), 209-235
work page 1992
-
[4]
G. E. Bredon,Introduction to Compact Transformation Groups, Academic Press, New York, 1972
work page 1972
-
[5]
G. Carlsson,On the non-existence of free actions of elementary abelian groups on products of spheres, Amer. J. Math.102(1980), 1147-1157
work page 1980
-
[6]
Carlsson,On the rank of abelian groups acting freely on(S n)k, Invent
G. Carlsson,On the rank of abelian groups acting freely on(S n)k, Invent. Math.69(1982), 393-400
work page 1982
-
[7]
Cartan and S
H. Cartan and S. Eilenberg,Homological Algebra, Princeton University Press, 1956
1956
Show all 18 references
-
[8]
L. W. Cusick,Elementary abelian 2-groups that can act freely on products of real projective spaces, Proc. Amer. Math. Soc.87(1983), 728-730
1983
-
[9]
L. W. Cusick, Free actions on products of even-dimensional spheres, Proc. Amer. Math. Soc. 99(3)(1987), 573–574
1987
-
[10]
Dey,Free rank of symmetry of products of Dold manifolds, Proc
P. Dey,Free rank of symmetry of products of Dold manifolds, Proc. of the Edinburgh Math. Soc.66(2023), 117–132
2023
-
[11]
Hanke,The stable free rank of symmetry of products of spheres, Invent
B. Hanke,The stable free rank of symmetry of products of spheres, Invent. Math.178(2009), 265–298
2009
-
[12]
Hatcher,Algebraic Topology, Cambridge: Cambridge University Press, 2002
A. Hatcher,Algebraic Topology, Cambridge: Cambridge University Press, 2002
2002
-
[13]
J. H. Jo and J. B. Lee,A note on free actions of groups on products of spheres, Bull. Aust. Math. Soc.88(2013), 340-344
2013
-
[14]
McCleary,A User’s Guide to Spectral Sequences, Cambridge Studies in Advanced Mathe- matics 58, Second Edition, Cambridge University Press, Cam- bridge, 2001
J. McCleary,A User’s Guide to Spectral Sequences, Cambridge Studies in Advanced Mathe- matics 58, Second Edition, Cambridge University Press, Cam- bridge, 2001
2001
-
[15]
O. B. Okutan and E. Yal¸ cın,Free actions on products of spheres at high dimensions, Algebr. Geom. Topol.13(4)(2013), 2087–2099
2013
-
[16]
Singh,Free 2-rank of symmetry of products of Milnor manifolds, Homology Homotopy Appl.16(1)(2014), 65-81
M. Singh,Free 2-rank of symmetry of products of Milnor manifolds, Homology Homotopy Appl.16(1)(2014), 65-81
2014
-
[17]
J. H. C. Whitehead,Combinatorial homotopy I, Bull. Amer. Math. Soc.55(1949), 213-245
1949
Reviewed August 7, 2026 · model on record in the stance chip above.
Discussion (0). Sign in to comment.