{"id":"b583088c-288d-4935-974a-44e9071755cb","arxiv_id":"1908.09107","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Irreducible linear involutions of Roth type admit bounded solutions to the cohomological equation up to a constant, and such maps form a full-measure set.","lead":"This paper extends the Marmi-Moussa-Yoccoz theorem on solving cohomological equations from interval exchange maps to irreducible linear involutions, which govern foliations from quadratic differentials. It proves that Roth-type linear involutions form a full-measure set and that the equation has a bounded solution up to a constant outside two exceptional strata.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Proposition 6, the full-measure core of Theorem 2, is asserted rather than proved, and Lemma 7's count estimates are inconsistent as printed.","rationale":"The paper is an honest extension of a substantial prior theorem, and the high-level strategy is plausible: condition (b) is backed by Gutiérrez-Romo's spectral simplicity result, and condition (c) follows from Oseledets theory as in the IET case. The load-bearing weakness is exactly where the reader placed it: the full-measure proof of condition (a) is delegated to a 'completely similar' case analysis that is not present, and Lemma 7's proof contains numerical inconsistencies that prevent verification. I agree with the reader's conditional verdict: the central claims are not disproved, but the paper does not currently supply enough detail to certify Proposition 6 and Theorem 2. The proposed concrete test would either expose a counterexample to Lemma 7 or force the authors to supply the missing combinatorial estimates; either outcome settles whether the concern lands.","tokens_in":8284,"tokens_out":5121,"duration_ms":53761,"concrete_test":"Enumerate all finite paths of length up to roughly 20 in the Rauzy diagram of a small irreducible linear involution (for instance, a d=3 generalized permutation with A' of size 2) and compute Qext and Q' for every prefix whose arrow names stay in A'. If any prefix violates Qext ≤ (2d−5)Q', Lemma 7 is false. If no violation appears, the remaining test is to write out the omitted Case A/B and Type I/II/III estimates for Proposition 6 and verify each step against the corresponding IET argument in [MaMoYo1, §4.8], checking that no step uses orientability or one-sidedness unavailable for linear involutions.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Theorem 2, and hence the full-measure status of the Roth-type condition used in Theorem 1, rests on Proposition 6 (§4.1). The proof of Proposition 6 (§4.2) stops after defining balanced paths and says that the unbalanced case should be handled by extending the path with cases 'completely similar to [MaMoYo1]'; none of the Case A/B and Type I/II/III estimates is written out for linear involutions. This is not a cosmetic omission: the Rauzy diagram of a linear involution is non-orientable and has two-to-one arrows, so counting arguments proved for IETs require independent verification. Lemma 7, used to control external columns, is also hard to check as printed. The proof allows a bad/good ratio up to 2d−4 for double letters, while the asserted conclusion is Qext ≤ (2d−5)Q'; and the initial-segment bound writes Qext(0)+m ≤ (d−D)+(n'_1−1) and then claims ≤ (d−2)Q', which appears to require n'_1−1 ≤ d−2 rather than the stated d0−1 ≤ 2d−5. Thus the constants do not line up, and since Proposition 6 depends on this estimate, the central full-measure claim is not established by the text as written.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper extends the Marmi–Moussa–Yoccoz theorem on cohomological equations for Roth-type interval exchange transformations to irreducible linear involutions, i.e., first-return maps of measured foliations defined by quadratic differentials. Theorem 1 states that for a minimal linear involution of Roth type not belonging to the strata Q(4g-4) (minimal) or Q(2a,2b,2c,...,2z), every anti-invariant function Phi in BV^1_* admits a decomposition Phi = chi + Psi - Psi circ T with chi constant on each interval and agreeing with the involution, and Psi bounded. Theorem 2 states that Roth-type linear involutions form a full-measure set in the space of irreducible linear involutions. The proof follows the strategy of [MaMoYo1]: Roth-type conditions (growth-rate, spectral gap, coherence) are introduced in Section 2; condition (b) is derived from Gutiérrez-Romo's simplicity theorem and condition (c) from Oseledets; Section 3 adapts the Birkhoff-sum argument, with the Denjoy step and the matrix-length equivalence (Proposition 5) addressed; Section 4 reduces the full-measure statement to Proposition 6, a combinatorial estimate on Rauzy diagrams. However, the proof of Proposition 6 and of the auxiliary Lemma 7 is not actually written out, and Lemma 7 contains internal inconsistencies in its constants as printed.","tokens_in":8539,"tokens_out":14613,"duration_ms":136757,"significance":"If the results hold, this is a valuable and natural extension of a landmark result in the theory of cohomological equations: it moves from orientable (Abelian) to non-orientable (quadratic) situations and singles out a full-measure Diophantine class for which a bounded solution with the correct anti-invariance is obtained. The paper is honest about its scope: it explicitly labels which arguments are verbatim, which are 'completely similar', and which are genuinely new, notably Proposition 5 and the anti-invariance remark after Theorem 1, which correctly explains why the double-cover reduction to the IET case cannot prove the stated theorem. The precise identification of the exceptional strata inherited from [Gu], together with the remark that a forthcoming veering-triangulation argument should remove it, is useful to the community. The main weakness is that the announced structure is not matched by written proofs: the central combinatorial estimate for the full-measure claim is asserted rather than demonstrated, and the constants in Lemma 7 as printed do not close.","major_comments":[{"comment":"Proposition 6 is the load-bearing step for the full-measure claim of Theorem 2, but its proof is not written out. The text after Lemma 8 states that for a path which is only (D-tilde, N, C-tilde)-balanced with D-tilde < D, the strategy is to extend gamma 'without losing volume', that 'There are several cases to distinguish (as in [MaMoYo1], Case A, Case B, Type I, II, III)', and that 'an argument completely similar to [MaMoYo1] leads to the desired estimate.' No volume estimate for the extension is carried out. This is not a cosmetic omission: the Rauzy diagram of a linear involution is non-orientable and has two-to-one arrows, so the counting arguments proved in [MaMoYo1, §4.8] for orientable IET diagrams do not automatically transfer and must be verified independently. The same reservation applies to Lemma 8, which is justified solely by 'a verbatim copy of [MaMoYo1, Lemma 2]' with no check that its hypotheses hold in the present setting. This is aggravated by the paper's own Remark after Theorem 2, which states that the simplified full-measure proof of [MaMaMo]/[AvGoYo] is not applicable for quadratic differentials; the missing case analysis is therefore exactly the part that cannot be imported wholesale. As printed, Theorem 2 is not established, and with it the full-measure status of the Roth-type condition used in Theorem 1.","section":"§4.1–4.2, Proposition 6 and Lemma 8"},{"comment":"Lemma 7 is asserted with conclusion Qext(n,T) ≤ (2d−5)Q′(n,T), but the estimates in its proof do not imply this bound as printed. First, for segments where the winning letter is double, the proof allows a bad/good ratio up to 2d−4, which is strictly larger than the constant 2d−5 of the conclusion; no averaging argument is supplied to reconcile the two. Second, the initial-segment estimate in the last three lines of the proof writes Qext(m,T) ≤ (d−D)+(n′₁−1) ≤ (d−2)Q′(m,T); the middle inequality requires n′₁−1 ≤ d−2, whereas the only stated bound is n′₁ ≤ d₀ with d₀ ≤ 2d−4, so the argument conflates the period d₀ with the alphabet size d. Third, the lower bound Q′(m,T) ≥ D ≥ 2 for m ∈ [0,n′₁) is asserted without justification: within the initial segment only the letter α₀ has appeared as a primary name before m, and nothing in the text shows that all D letters of A′ have already contributed to Q′(m,T). Since Lemma 7 controls the growth of external columns inside the proof of Proposition 6, the constant mismatch and the unjustified bounds must be repaired before the argument goes through.","section":"§4.2, Lemma 7"},{"comment":"The adaptation of the Gottschalk–Hedlund step to linear involutions is delegated in a single sentence: after defining D₀ and D₁ in (4), the text says 'The rest of the proof repeats verbatim the construction described in [MaMoYo1]'. In [MaMoYo1] the Denjoy-type construction for the discontinuous interval exchange, the control of the Birkhoff sums near the singularities, and the passage back to the original map form a delicate chain; in the present setting it is further required that the bounded function Ψ and the constant function χ agree with the linear involution, a property that is specific to the statement of Theorem 1 and is not a by-product of the double-cover reduction (as the paper itself notes in the Remark after Theorem 1). The authors should state the analogue of the corresponding lemma of [MaMoYo1, §2.1.2] and verify that its hypotheses hold for the anti-invariant class BV^1_∗; as written, Theorem 1 is a plausible but unverifiable claim.","section":"§3.2"}],"minor_comments":[{"comment":"The constant bookkeeping in the proof of the direction (2) ⇒ (a) is sloppy: the printed chain contains '2dCε 2' (presumably 2dCε²) and the final step silently absorbs a factor 2dCε² into Cε and renames 2ε as ε; please rewrite the chain with explicit constants.","section":"§3.1, Proposition 5"},{"comment":"The terms '1-segment', 'secondary name', and the period d₀ are used without definition; readers not already familiar with [MaMoYo1, §4.8] cannot follow the proof of Lemma 7 as it stands.","section":"§4.2, Lemma 7"},{"comment":"In the definitions Q′(n,T) = Σ_{α∈A′} Qα(n,t) and Qext(n,T) = Σ_{α∈A\\A′} Qα(n,t), the argument is written as a lowercase 't' in Qα(n,t), inconsistent with the uppercase T used throughout; this should be fixed.","section":"§4.2, display (5)"},{"comment":"There are several typos: 'Gottshalk–Hedlund' in the first bullet of Section 3 should read 'Gottschalk–Hedlund', and 'To be consistant' at the start of §4.2 should read 'To be consistent'.","section":"§3 and §4.2"},{"comment":"Reference [Gu] is cited as a preprint in both the text and the bibliography; if a published version exists, the reference should be updated.","section":"References"}],"recommendation":"major_revision","confidential_remarks":"This is a research-announcement-style manuscript: the introduction, the statements of the results, and the novelty claims are clear and honest, and the paper explicitly identifies the parts that are imported verbatim or by analogy. The decisive question for acceptance is whether the journal's standards allow Theorem 2 to rest on an unwritten case analysis. My reading is that the missing part is essential and must be supplied: the constants in Lemma 7 do not close as printed, which suggests that the adaptation from the IET counting arguments to the non-orientable diagram is not a purely mechanical task. I would recommend requesting a major revision in which the authors either write out the case analysis for Proposition 6 and a corrected Lemma 7, or expand the note into a full paper; if the missing details are supplied, the result is likely correct and publishable. The fit with the journal is otherwise good, and the authors are appropriately careful in attributing the input theorems."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The two theorems are new and the obstruction they flag is real: the double-cover route to the IET result would not produce a solution respecting the involution, so the anti-invariant setting genuinely needs its own argument. The paper is also honest about where it relies on external results — Gutiérrez-Romo's Lyapunov simplicity, Avila–Resende, the Boissy–Lanneau dictionary — and those citations look appropriate. No fitted parameters, no circularity, and the claimed full-measure statement for Roth-type linear involutions is a natural and significant target.\n\nThat said, the proof as written stops exactly where the hard work is. Proposition 6 is the engine behind Theorem 2, and its proof in §4.2 ends with a sentence: the unbalanced case is handled by extending the path with cases \"completely similar\" to Cases A, B, Type I, II, III in MaMoYo1. None of those estimates are written down for linear involutions. This is not a cosmetic omission: the Rauzy diagram for a linear involution is non-orientable and has two-to-one arrows, so the counting arguments for IETs do not automatically transfer. Lemma 8 is quoted as a verbatim copy of Lemma 2 of MaMoYo1 — that is fine if the copy is faithful, but it leaves no way for a reader to check the adaptation.\n\nLemma 7 also has a constants problem. The text allows a bad/good ratio up to 2d−4 for double letters and then concludes Qext ≤ (2d−5)Q′, and the initial-segment estimate appears to require n′1−1 ≤ d−2 while n′1 can be as large as d0, which earlier is only bounded by 2d−4. Either I am misreading the notation or the inequality as printed is false. Since Proposition 6 depends on this estimate, the central full-measure claim is not established by the text as written.\n\nWhere credit is due: the paper is clearly written by people who know the area, the conceptual framework is sensible, and the two theorems would be valuable if the missing details check out. The note is short but that is not itself a flaw — MaMoYo1 itself is long and the authors are explicit about what they are copying. The issue is that the parts they defer are exactly the parts that need independent verification for linear involutions.\n\nWho this is for: specialists in Teichmüller dynamics and cohomological equations. It deserves a serious referee, but the referee should demand either a full write-up of the Proposition 6 case analysis or a clear statement that it is an assumption rather than a theorem, and a corrected Lemma 7 with uniform constants. As it stands, I would not cite it for the full-measure result, though I would cite it as evidence that the MaMoYo1 framework is being extended.","headline":"New extension of the Marmi-Moussa-Yoccoz cohomological equation to irreducible linear involutions, but the full-measure core is sketched rather than proved.","tokens_in":9063,"tokens_out":2083,"would_cite":false,"duration_ms":20319,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["37E05","37E35","37A05","30F30"],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper proves that every minimal irreducible linear involution of Roth type outside two exceptional strata solves the cohomological equation with a bounded solution and a piecewise-constant correction, and that Roth-type involutions…","keywords":["cohomological equation","linear involutions","Roth type","Rauzy-Veech induction","quadratic differentials","interval exchange transformations","bounded variation","Lyapunov exponents"],"falsifier":"A direct test is to carry out the omitted case analysis of Proposition 6 for a concrete generalized permutation: check whether, for every finite path γ whose arrow names miss at least one letter, the claimed proportion η·vol_{d−1}(Δ(γ)) of initial data admits the required extension M with no more than l(D−1) covering segments. If a single path violates the estimate, Theorem 2 loses its support; if the analysis goes through for all small alphabets, the full-measure step is confirmed.","tokens_in":8083,"feed_emoji":"📐","tokens_out":6937,"duration_ms":66463,"temperature":0.7,"pith_summary":"This paper proves that the cohomological equation Ψ − Ψ∘T = Φ − χ is solvable for irreducible linear involutions, the natural non-orientable counterpart of interval exchange transformations and the first-return maps of foliations defined by quadratic differentials. For any minimal Roth-type linear involution outside two exceptional strata, and for any function Φ of class $C^{1}$ with derivative of bounded variation and zero mean that is compatible with the involution, there is a bounded function Ψ and a function χ constant on each continuity interval such that the equation holds. The paper also shows that the Roth-type condition has full Lebesgue measure among all irreducible linear involutions. This extends to the quadratic-differential setting the sharp, explicitly Diophantine solvability result that was known for interval exchange transformations.","feed_headline":"Nearly every linear involution solves the cohomological equation","feed_subtitle":"The exceptions are two sparse strata; the Roth-type full-measure class gains bounded solutions.","key_machinery":"The argument is carried by Rauzy–Veech induction with Zorich acceleration and the further acceleration from the interval-exchange paper, which packages the dynamics into products of matrices Z(k) and Q(k). Roth type is defined by three conditions: a growth-rate condition on these matrices, a spectral gap for the associated Birkhoff-sum cocycle on the kernel of the invariant mean, and a coherence condition on stable subspaces. The proof uses a Denjoy-type construction to pass from a minimal but discontinuous involution to a continuous setting, then reduces general Birkhoff sums to special return-time sums and estimates those using the three conditions. The anti-invariance required in the solution is handled by working on the oriented double cover with the 'minus' construction.","core_discovery":"Theorem 1 states that every minimal irreducible linear involution of Roth type, outside the strata Q(4g−4) and Q(2a,2b,2c,…,2z), admits a bounded solution Ψ and a step function χ constant on each interval A_i to Ψ − Ψ∘T = Φ − χ, for every datum Φ in the space of $C^{1}$ functions with derivative of bounded variation, zero mean, and agreement with the involution. The solution has the exact expected shape: both Ψ and χ agree with the linear involution, which rules out obtaining the result by a direct pullback from the interval-exchange theorem. Theorem 2 adds that the Roth-type condition defines a full-measure subset of the space of irreducible linear involutions, so the solvability statement applies to almost every such involution outside the exceptional strata.","pith_inferences":["The unstated part of the proof of Proposition 6 is the true test of the paper: writing out the case analysis for linear involutions would either confirm the full-measure step or produce a counterexample.","A likely next step is Hölder regularity of Ψ for C^r data with r > 1 under a restricted Roth-type condition, mirroring the improvement made for interval exchange transformations.","Because the same cocycle appears in the study of time changes of quadratic differential flows, the theorem may transfer to smoothness and rigidity statements for those flows, a direction the paper does not explore."],"forward_implications":["Every minimal Roth-type irreducible linear involution outside the exceptional strata is uniquely ergodic, since condition (b) holds almost everywhere and implies unique ergodicity.","Because Roth type has full measure and conditions (b) and (c) hold almost everywhere, the main theorem applies to almost every irreducible linear involution outside Q(4g−4) and Q(2a,2b,2c,…,2z).","If the cited simplicity-of-spectrum result is extended to all strata, as the paper notes is plausible, Theorem 1 extends automatically to every minimal irreducible linear involution.","The agreement of Ψ and χ with the involution is essential: it prevents a naive reduction to the interval-exchange case and makes the theorem a genuinely quadratic-differential statement."],"supporting_citations":[{"why":"Supplies the full proof architecture for the cohomological equation, including the combinatorial Proposition 6 whose adaptation to linear involutions is asserted but not written out.","marker":"[MaMoYo1]"},{"why":"Provides the simplicity of Lyapunov spectra for specific quadratic-differential strata used in Lemma 3 to establish condition (b) almost everywhere.","marker":"[Gu]"},{"why":"Supplies exponential mixing for the Teichmüller flow in the space of quadratic differentials, referenced as the basis for the log-integrability and stronger estimates behind conditions (b) and (c).","marker":"[AvRe]"},{"why":"Defines irreducible generalized permutations and the Rauzy–Veech induction for linear involutions, which form the combinatorial framework of the theorems.","marker":"[BoLa]"},{"why":"Introduces linear involutions and the initial Rauzy induction for them, fixing the object of study.","marker":"[DaNo]"}],"fun_headline_variants":["Almost all linear involutions solve the cohomological equation","Cohomological equation solved for almost every linear involution","Roth-type involutions: bounded solutions except two sparse strata","Linear involutions: a.e. solve cohomological equation, sparse exceptions","Exceptional strata only: cohomological equation solved a.e. for involutions"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The full-measure claim for condition (a) stands on Proposition 6, a volume estimate for extending paths in the Rauzy diagram, and the paper does not actually prove it for linear involutions: it says the extension strategy and case analysis are 'completely similar' to the interval-exchange situation without presenting those details.","fun_headline_variants_meta":{"raw":{"variants":["Almost all linear involutions solve the cohomological equation","Cohomological equation solved for almost every linear involution","Roth-type involutions: bounded solutions except two sparse strata","Linear involutions: a.e. solve cohomological equation, sparse exceptions","Exceptional strata only: cohomological equation solved a.e. for involutions"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000725,"raw_usage":{"total_tokens":3132,"prompt_tokens":708,"completion_tokens":2424,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":324,"completion_tokens_details":{"reasoning_tokens":2332}},"tokens_in":324,"tokens_out":2424,"duration_ms":16379,"temperature":1.0,"reasoning_tokens":2332,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T11:20:47.896830+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A direct test is to carry out the omitted case analysis of Proposition 6 for a concrete generalized permutation: check whether, for every finite path γ whose arrow names miss at least one letter, the claimed proportion η·vol_{d−1}(Δ(γ)) of initial data admits the required extension M with no more than l(D−1) covering segments. If a single path violates the estimate, Theorem 2 loses its support; if the analysis goes through for all small alphabets, the full-measure step is confirmed.","supporting_citations":[],"review_version":1}