{"id":"e8d6a1fe-fc1e-4129-8480-280f81f94b7e","arxiv_id":"2607.29368","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":8.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"For 2-step nilpotent actions, polynomial averages with distinct-degree iterates converge to the product of the integrals, and the joint ergodicity conjecture for Z^D polynomial actions is fully resolved.","lead":"This paper proves explicit limit formulas for multiple ergodic averages of 2-step nilpotent group actions with distinct-degree polynomial iterates, and resolves the joint ergodicity conjecture for multidimensional polynomial actions of Z^D. It also constructs counterexamples showing the natural 2-step nilpotent analog of that conjecture fails.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Proposition 8.1's bridge is the load-bearing risk: its proof of Proposition 8.2 invokes Parry's affine-isomorphism theorem without verifying that the conjugated system is a nilsystem on the original nilmanifold.","rationale":"The reader's weakest assumption identifies exactly the same bridge: Proposition 8.1, with its reliance on Parry's theorem and noncommutative factor-invariance. I read the surrounding proof carefully: the induction in Section 8.3 and the use of Candela-Szegedy are structurally coherent, and the issue concentrates in Proposition 8.2's invocation of Parry. This does not make me doubt the theorem's truth, so I do not recommend rejection; but the paper should either supply the missing verification that the conjugated transformations are nilrotations on the original nilmanifold, or cite the precise affine/normalizer theorem that applies. Since Theorem 1.5's proof collapses without this step, CONDITIONAL remains the appropriate verdict, so the reader's verdict does not change.","tokens_in":73217,"tokens_out":36420,"duration_ms":337096,"concrete_test":"Take the Heisenberg nilmanifold Y=H_3(Z)\\H_3(R), fix a totally ergodic nilrotation S_1=L_a, and let S_2 be a measure-preserving transformation such that ⟨S_1,S_2⟩ is 2-step nilpotent. Compute \\hat S_1=S_2S_1S_2^{-1} and check explicitly whether it is again a left translation on Y. Equivalently, re-derive Proposition 8.2 with the target nilsystem written as G'/Γ' instead of G/Γ and verify whether Parry's theorem forces G'=G and σ(Γ)=Γ. If the conjugates need not be nilrotations on the original nilmanifold, then the proof of Proposition 8.1 is incomplete and must be replaced.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The route from the seminorm estimates (Theorem 1.6) to the limiting formula (Theorem 1.5) passes through Proposition 8.1, i.e. the inclusion Z_s(T_j) ⊆ Z_{s'}(H). Its proof in Section 8.1 depends on Proposition 8.2, where Parry's theorem is used to conclude that S_{ℓ+1} is affine: S_{ℓ+1}(gΓ) = bσ(g)Γ with b∈G, σ∈Aut(G), σ(Γ)=Γ. But Parry's theorem is an affinity statement between nilsystem structures. The target system here is generated by the conjugates \\hat S_j = S_{ℓ+1}S_jS_{ℓ+1}^{-1}; the authors do not verify that these conjugates are nilrotations on the original nilmanifold Y=G/Γ, nor that the target nilsystem structure is the same G/Γ. If the structures differ, the conclusion should be b∈G', σ:G→G' for some other nilpotent Lie group G', and the subsequent construction of the ks²-step nilsystem, as well as the factor-invariance chain Z_s(H)=T_ℓZ_s(H) in Propositions 8.4/8.6, does not follow. Theorem 1.5 is reduced to equidistribution on an H-nilsystem exactly by this inclusion, so a failure here separates the new seminorm estimates from the claimed limit. This is not a known counterexample, but it is a missing verification at the most noncommutative, load-bearing step.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper develops seminorm estimates and limiting formulas for multiple ergodic averages of the form N^{-1}\\sum_{n=1}^N T_1^{p_1(n)}f_1\\cdots T_\\ell^{p_\\ell(n)}f_\\ell for measure-preserving actions generated by 2-step nilpotent groups and nonconstant polynomials of distinct degrees. The central results are Theorem 1.5 (total ergodicity implies L^2 convergence to the product of integrals), Theorem 1.6 (Host-Kra seminorm estimates), Theorem 1.8/Corollary 1.9 (popular common differences), Theorem 1.13 (full resolution of the joint ergodicity conjecture for polynomial Z^D-actions), Theorem 1.15 (a joint ergodicity criterion in the 2-step nilpotent setting), Theorems 1.14 and 1.16 (counterexamples to naive nilpotent analogues), and Theorem 1.18 (new abelian box-seminorm estimates). The proof combines a 2-step nilpotent PET induction, relative concatenation and seminorm smoothing, Host-Kra factor comparisons via affine-isomorphism theorems, and equidistribution results on nilsystems.","tokens_in":73580,"tokens_out":10696,"duration_ms":107772,"significance":"If the proofs are correct, this is a substantial advance: it gives the first explicit limiting formulas for genuinely noncommutative polynomial multiple averages, closes the abelian joint ergodicity conjecture, and supplies new seminorm control that is likely to be useful beyond the present applications. The paper is largely proof-heavy and self-contained, and the model computation in §6 convincingly illustrates the PET mechanism. The counterexamples to the nilpotent analogue of joint ergodicity are valuable and clearly presented. The main caveats are a load-bearing gap in the use of Parry's theorem in §8.1 and an explicitly omitted variant of Proposition 9.3 that is invoked in the discussion around Theorem 1.15; both need to be resolved before the central claims can be regarded as fully proven.","major_comments":[{"comment":"The proof uses Parry's theorem to conclude that S_{\\ell+1} is affine, but the hypotheses of Parry's theorem are not verified. The conjugated transformations \\hat S_j = S_{\\ell+1}S_jS_{\\ell+1}^{-1} are not shown to be nilrotations on the original nilmanifold G/\\Gamma, so the target system (Y,\\hat S_1,\\dots,\\hat S_\\ell) is not known to be a nilsystem before applying Parry. As written, the proof even contains an apparent typo: the conjugating map is called S_\\ell, not S_{\\ell+1}. This is not a cosmetic issue: Proposition 8.2 is the bridge that lets the authors identify Z_{s'}(H) factors and pass from the seminorm estimates of Theorem 1.6 to the limiting formula of Theorem 1.5. The authors should either prove a normalizer-type theorem for ergodic nilsystems or give an independent argument that S_{\\ell+1} is affine with respect to the original nilmanifold structure.","section":"§8.1, Proposition 8.2"},{"comment":"The text explicitly states that Proposition 9.3 does not extend to integer-valued polynomials and that 'a more complicated version' sufficient for Theorem 1.15 is omitted. This is a flagged missing proof. If Theorem 1.15 is intended only for p_j\\in Z[n], the omission is not load-bearing and the remark should be clarified. But if the theorem or its applications cover integer-valued polynomials, the proof of Theorem 1.15 is incomplete as it stands. The manuscript should either include the promised variant or explicitly restrict the statement and remove the implication that the omitted result is needed.","section":"§9.1, after Proposition 9.3"},{"comment":"The final reduction from the box-seminorm control in Theorem 1.18 to control by a single Host-Kra seminorm |||f_j|||_{s,T_j} is compressed. In a general Z^D-system the notation T_j is not defined, and the passage through 'only ergodic subgroups' followed by (26) hides several nontrivial steps: one must justify why the subgroups H_{j,j'} can be replaced by a single ergodic subgroup and why that subgroup gives the same seminorm as the one generated by the relevant coordinate action. This is a central step in resolving Conjecture 1.12, so the argument should be written out or the relevant known lemma should be quoted precisely.","section":"§10, proof of Theorem 1.13"}],"minor_comments":[{"comment":"Section 2.1 says that Section 6 is dedicated to the proof of 'Theorem 1.18 for the model average', but Theorem 6.1 is a special case of Theorem 1.6, not of Theorem 1.18. Please correct the cross-reference.","section":"§6, Section 2 overview"},{"comment":"The proof has several apparent typos: the isomorphism map should be S_{\\ell+1}, not S_\\ell, in the first paragraph, and later 'S_\\ell agrees m_Y-a.e. with an affine map' should presumably refer to S_{\\ell+1}. These typos make the already delicate argument harder to verify.","section":"§8.1, Proposition 8.2"},{"comment":"The footnote contains a duplicated word: 'applies applies'. Please correct.","section":"§1.2, footnote 4"}],"recommendation":"major_revision","confidential_remarks":"The paper is innovative and likely correct in its overall strategy, but the gap in Proposition 8.2 is at the most noncommutative, load-bearing point of the proof of Theorem 1.5. I would be willing to review a revision in which this gap is repaired and the omitted Proposition 9.3 variant is either supplied or explicitly removed from the claims."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Let me give you the short version: this is a serious, high-quality paper that delivers on its main promises—Theorem 1.13 resolves the abelian joint ergodicity conjecture in full, and Theorem 1.5 gives the first explicit limit for genuinely 2-step nilpotent polynomial averages, including the model case {n, n^2}. The seminorm estimates in Theorem 1.18 are new and genuinely useful. The paper is well written, with detailed proof sketches and honest discussion of where methods break.\n\nThe main soft spots are two. First, the proof of Theorem 1.15 explicitly omits a 'more complicated version' of Proposition 9.3 that is needed to handle integer-valued polynomials. The authors state this flat out; the theorem is asserted without the missing support. For a paper of this scope, that's acceptable only if the omitted argument is genuinely routine—and it may be—but it needs to be supplied or Theorem 1.15 downgraded to a conditional statement.\n\nSecond, the bridge from the seminorm estimates to the limit in Theorem 1.5 goes through Proposition 8.1, whose proof rests on Proposition 8.2. That argument invokes Parry's theorem to conclude a conjugate transformation is affine on the same nilmanifold, but the proof doesn't verify that the conjugated system is a nilsystem on the original G/Γ with the same structure. If the structures differ, the parity argument shifts to a different nilmanifold and the factor-invariance chain doesn't follow. I don't have a counterexample, and the authors' proof may be fixable with a short verification, but this is exactly the kind of noncommutative step that deserves a referee's eye.\n\nThere's also a moderate self-citation pattern: Theorem 1.18 is proven via Theorem 1.17 and the relative concatenation result [19, Theorem 3.1], both from the same group of authors. That's not damning—these are published results—but it does mean the paper's independence from its own prior work is overstated.\n\nNet: the central claims are probably correct, the writing is clear, and the paper is important. I would send it to a serious referee, with the explicit instruction to check Proposition 8.2 and demand the missing Proposition 9.3 variant. If those two items check out, this becomes a landmark paper.","headline":"Major paper that resolves the multidimensional joint ergodicity conjecture and gives the first explicit limits for 2-step nilpotent polynomial averages; the proofs are plausible but two gaps need filling before I'd trust the central theorem unconditionally.","tokens_in":74080,"tokens_out":1686,"would_cite":true,"duration_ms":18614,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["37A30","11B30","28D05","37A44"],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper proves explicit L2 limits for multiple ergodic averages of totally ergodic 2-step nilpotent systems with distinct-degree polynomial iterates, and resolves the abelian joint ergodicity conjecture.","keywords":["ergodic averages","2-step nilpotent group actions","box seminorms","joint ergodicity","polynomial iterates","nilsystems","popular common differences","multiple recurrence"],"falsifier":"A concrete check is to take the 2-step nilpotent skew rotations on the two-dimensional torus given by T(x,y)=(x+a, y+2x+a) and S(x,y)=(x+b, y+2x+b) with 1,a,b rationally independent, and compute the L2 limit of (1/N)∑_{n=1}^N f1(T^n z) f2(S^{n^2} z) for continuous f1,f2. Theorem 1.5 predicts the product of the two integrals. A Fourier/spectral computation can settle this directly in the simplest noncommuting case. A failure would refute the theorem; a success would exercise the claimed mechanism and also test the factor inclusion Z_2(T) ⊆ Z_8(H).","tokens_in":73104,"feed_emoji":"🔁","tokens_out":10619,"duration_ms":107412,"temperature":0.7,"pith_summary":"This paper aims to describe, rather than only prove the existence of, the limit of multiple ergodic averages in which transformations T1,...,Tℓ generate a 2-step nilpotent group and the iterates are nonconstant polynomials of distinct degrees. The main theorem says that when each Tj is totally ergodic, the L2 limit of (1/N) Σ_{n≤N} f1(T1^{p1(n)}x)···fℓ(Tℓ^{pℓ(n)}x) equals the product of the integrals of f1,...,fℓ. Along the way the paper establishes box-seminorm estimates for such averages, derives popular-common-difference versions of the polynomial multiple recurrence theorem for 2-step nilpotent groups, and proves the joint ergodicity conjecture for multidimensional polynomials in Z^D-systems. It also gives a counterexample showing that the analogous joint ergodicity criterion fails in the 2-step nilpotent world, and it states that its methods do not extend beyond 2-step nilpotency or distinct-degree polynomials.","feed_headline":"Explicit limits for 2-step nilpotent averages","feed_subtitle":"Distinct-degree polynomial iterates converge to the product of integrals, and the abelian joint ergodicity conjecture is resolved.","key_machinery":"The argument is carried by four interlocking components. First, a 2-step nilpotent version of the PET induction scheme: by repeatedly applying a standard difference (Cauchy-Schwarz-type) inequality, it eliminates the transformations T1,...,Tℓ-1 one by one, using a property called distinguishability to keep the Tℓ-coordinate independent of the commutator coordinates until only commuting transformations remain. Second, new abelian seminorm estimates (Theorem 1.18) control the resulting multiparameter averages by box seminorms attached to subgroups generated by all coefficient differences of the polynomials. Third, a comparison of box factors (Proposition 8.1) embeds each individual factor Z_s(","core_discovery":"On the paper's own terms, the central discovery is that 2-step nilpotent polynomial averages have a completely described limit: for totally ergodic T1,...,Tℓ and nonconstant polynomials p1,...,pℓ of distinct degrees, the average E_{n≤N} T1^{p1(n)}f1···Tℓ^{pℓ(n)}fℓ converges in L2 to ∫f1 dμ···∫fℓ dμ (Theorem 1.5). The accompanying seminorm estimates (Theorem 1.6) say the average tends to zero as soon as any one function has vanishing box seminorm of a suitable degree along its own transformation; this gives a characteristic factor for the average and supplies the technical engine for the paper's other results. The same machinery proves the full joint ergodicity conjecture for abelian Z^D-syst","pith_inferences":["If the factor-inclusion exponent can be improved from O(s^2) to s, the approach might extend to higher-step nilpotent groups; the quadratic overhead looks like an artifact of the comparison argument, and the paper's open problems highlight this as a natural target.","The counterexample to the naive nilpotent joint ergodicity criterion suggests that joint ergodicity for nilpotent actions will need additional conditions on the action of the commutator subgroup, not just product and difference ergodicity.","The new abelian seminorm estimates, being uniform across functions and expressed in terms of full coefficient subgroups, may be the right input for quantitative finite-N bounds; the paper itself notes that quantification is open.","The distinguishability trick for keeping Tℓ separate from commutators is likely the key noncommutative innovation; testing it on the model average T^n S^{n^2} for a 3-step nilpotent pair would reveal whether the method can go further."],"forward_implications":["For every totally ergodic 2-step nilpotent system, any distinct-degree polynomial average has an explicit L2 limit equal to the product of the integrals.","The seminorm estimates make the characteristic factor explicit: if one function is orthogonal to the degree-s box factor of its transformation, the whole average vanishes.","In any finitely generated 2-step nilpotent group, every positive-density set contains syndetically many common differences of the form x, h1^{n^{d1}}x, ..., hℓ^{n^{dℓ}}x, with intersection density arbitrarily close to the natural density bound.","For Z^D-actions, the joint ergodicity conjecture is settled: product ergodicity plus difference ergodicity is necessary and sufficient for multidimensional polynomial sequences.","Totally ergodic nilrotations along independent polynomials produce jointly equidistributed orbits almost everywhere on nilsystems."],"fun_headline_variants":[],"cache_read_input_tokens":2304,"weakest_assumption_plain":"The load-bearing premise is Proposition 8.1: for a 2-step nilpotent system, every box factor of an individual generator T_j is contained in a sufficiently high-degree box factor of the whole group H; the proof of this inclusion uses a structural theorem for nilsystems (that measurable isomorphisms are affine) and does not obviously extend to higher-step groups. If this inclusion fails for some 2-step system, the explicit limit formula would not follow from the seminorm estima","fun_headline_variants_meta":{"error":"'NoneType' object is not subscriptable"},"cache_creation_input_tokens":0},"created_at":"2026-08-03T08:23:15.558777+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A concrete check is to take the 2-step nilpotent skew rotations on the two-dimensional torus given by T(x,y)=(x+a, y+2x+a) and S(x,y)=(x+b, y+2x+b) with 1,a,b rationally independent, and compute the L2 limit of (1/N)∑_{n=1}^N f1(T^n z) f2(S^{n^2} z) for continuous f1,f2. Theorem 1.5 predicts the product of the two integrals. A Fourier/spectral computation can settle this directly in the simplest noncommuting case. A failure would refute the theorem; a success would exercise the claimed mechanism and also test the factor inclusion Z_2(T) ⊆ Z_8(H).","supporting_citations":[],"review_version":1}