{"id":"e113ca04-ad40-409f-8953-029416e82800","arxiv_id":"1908.07020","paper_version":1,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Suspension flows over transitive sub-shifts of finite type can be C0-perturbed to have either uncountably many or a unique ergodic measure of maximal entropy.","lead":"This mathematics paper shows that any suspension flow over a standard shift space can be changed by an arbitrarily small tweak to the roof function so that it has either uncountably many or exactly one measure of maximal entropy. The result marks a sharp contrast between smooth and merely continuous roof functions in thermodynamic formalism.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem 1.1 as stated fails for zero-entropy transitive SFTs such as the period-two shift; an explicit positive-entropy hypothesis is needed, though the intended construction appears sound after that fix.","rationale":"The reader's weakest assumption identifies exactly the load-bearing gap: the theorem statement is broader than the hypotheses used in the proof. The period-two shift is a legal object under the stated standing assumptions, and it is a genuine counterexample to Theorem 1.1(a), since the suspension flow has only one invariant measure. The proof's appeal to positive entropy is explicit in Proposition 3.4 and implicit in Lemma 3.6's division by h(Φτ). This is a localized, fixable flaw rather than a collapse of the main construction: for positive-entropy SFTs, Lemmas 3.5 and 3.6 correctly transfer Israel's and Ruelle's density results through the pressure equation and Abramov's formula. Remark 3.9 overclaims the sufficiency of entropy density, as the period-two shift satisfies the listed hypotheses while Israel's conclusion fails; the missing condition is that the relevant sets of ergodic measures support non-atomic measures, which fails when the set of ergodic measures is finite or a singleton. I agree with the reader's conditional verdict: the paper should be accepted only after the statement is amended to include positive entropy and Remark 3.9 is corrected.","tokens_in":9197,"tokens_out":6483,"duration_ms":69273,"concrete_test":"Take the period-two SFT A = [[0,1],[1,0]]. Compute that Σ consists of the two alternating sequences, h(σ)=0, and the only invariant measure is μ=(δ_a+δ_b)/2. For every φ∈C(Σ), P(φ)=(φ(a)+φ(b))/2, so the set L in Proposition 3.4 is empty. Trace Lemma 3.6 with any positive roof τ: h(Φτ)=0, so the definition of τ_n divides by zero, and part (a) cannot hold. This concrete verification settles that an explicit positive-entropy hypothesis is necessary.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The paper's standing assumption (topological transitivity and alphabet size at least two) does not imply positive topological entropy, yet the proof relies on it. Proposition 3.4 states 'by our standing assumption (Σ,σ) has positive topological entropy', but this is not part of the standing assumption. Lemma 3.6 then divides by h(Φτ), the flow entropy, which is zero exactly when the base entropy is zero. The period-two shift with transition matrix [[0,1],[1,0]] is topologically transitive on two symbols, has exactly two points, topological entropy zero, and a unique invariant measure. By the Ambrose–Kakutani bijection, any suspension flow over it has a unique invariant measure, hence a unique measure of maximal entropy, so Theorem 1.1(a) is false as stated. The same gap affects Remark 3.9: finite entropy, entropy density, and upper semi-continuity all hold for the period-two shift, so those conditions are not sufficient for the Israel conclusion; the set A(F,ε) cannot support a non-atomic measure when the ergodic measure set is a singleton. The intended theorem is recovered by adding h(σ)>0 (equivalently h(Φτ)>0) to the hypotheses and strengthening Remark 3.9 accordingly.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies suspension flows over sub-shifts of finite type with continuous roof functions. The main theorem (Theorem 1.1) asserts that for any topologically transitive one- or two-sided SFT with at least two symbols and any positive continuous roof function, an arbitrarily small C0-perturbation of the roof produces a suspension flow with uncountably many ergodic measures of maximal entropy, and another small perturbation produces a flow with a unique measure of maximal entropy. The proofs rely on Israel's theorem that continuous functions with uncountably many equilibrium states are dense, Ruelle's theorem on uniqueness of equilibrium states for Hölder potentials, and the Abramov–Kac formulas for suspension flows. The paper also proves an analogous density statement for equilibrium measures of continuous potentials on the suspension flow.","tokens_in":9381,"tokens_out":5933,"duration_ms":62465,"significance":"If the result holds as stated, it gives a strong dichotomy between the smooth (Hölder) and merely continuous thermodynamic formalism for suspension flows: both unique and uncountably many measures of maximal entropy are dense in the space of roof functions. The proof is transparent and uses external theorems without introducing fitted parameters, and the flow-level extension in Theorem 3.13 is a useful complement. However, the main theorem as stated is false without a positive-entropy hypothesis on the base, so the central claim needs correction before the result can be accepted.","major_comments":[{"comment":"The standing assumption is inconsistent and does not imply the positive-entropy claim used later. The definition of 'topologically transitive' as existence of m with all entries of A^m positive is actually a mixing condition, and the stated equivalence with existence of a dense orbit is false: the two-symbol shift with transition matrix [[0,1],[1,0]] has a dense orbit in the usual transitive sense but no power of A is positive. Under the dense-orbit reading, this base satisfies the standing assumption, has zero topological entropy, and has exactly one invariant measure; by the Ambrose–Kakutani correspondence every suspension flow over it has a unique invariant measure, so Theorem 1.1(a) is false for that base. Proposition 3.4 explicitly asserts 'by our standing assumption (Σ,σ) has positive topological entropy', but that assertion is not a consequence of the stated assumptions under the standard meaning of topological transitivity. The theorem requires an explicit hypothesis that h(σ)>0 (equivalently h(Φ_τ)>0), and the definition of transitivity should be corrected or the erroneous equivalence removed.","section":"§2.1 and Proposition 3.4"},{"comment":"The construction in Lemma 3.6 divides by h(Φ_τ) when defining τ_n(x) = (τ(x) + φ_n(x) - h(Φ_τ)τ(x)) / h(Φ_τ). If the base shift has zero entropy, Abramov's formula gives h(Φ_τ)=0 for every invariant measure, so this expression is ill-defined. This is the precise technical step that requires a positive-entropy hypothesis on the base; adding h(σ)>0 to the hypotheses of Lemma 3.6 and Theorem 1.1 restores the argument, since then h(Φ_τ)>0 as well.","section":"Lemma 3.6"},{"comment":"The claimed sufficient conditions for Theorem 1.1a are not sufficient. The period-two shift with transition matrix [[0,1],[1,0]] has finite topological entropy (zero), its ergodic measures are entropy dense in the space of invariant measures (trivially, since the space is a singleton), and the entropy map is upper semi-continuous. Nevertheless, the conclusion of Theorem 1.1a fails for this base because the set A(F,ε) defined in the remark contains only the unique ergodic measure and cannot support the non-atomic measure ν_ε used in Israel's construction. The remark therefore needs an additional hypothesis, such as positive topological entropy or a condition ensuring that A(F,ε) is non-atomic.","section":"Remark 3.9"}],"minor_comments":[{"comment":"The sentence 'we can conclude that the the suspension flow' contains a duplicated article; it should read 'the suspension flow'.","section":"Proof of Proposition 3.4"},{"comment":"The notation 'pYτ 1m, Φ τmq' for the first sequence should be 'pYτm, Φ τmq' to be consistent with the definition of τ_n in Lemma 3.6.","section":"Proof of Theorem 1.1, after Lemma 3.6"},{"comment":"Several reference entries contain typographical errors, for example 'Addis on-W esley' and 'F acultad'; these should be corrected.","section":"References"}],"recommendation":"major_revision","confidential_remarks":"The core perturbation construction is sound for positive-entropy bases, and I believe the paper can be made correct with a modest revision that adds an explicit positive-entropy hypothesis to Theorem 1.1 and repairs the standing-assumption definition. The false 'equivalence' in the definition of topological transitivity is likely to confuse readers and should be fixed. I do not see any circularity or ad-hoc fitting; the reliance on Israel's and Ruelle's theorems is appropriate."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The main results here are new and worth knowing: for a fixed transitive SFT base with a continuous positive roof, both uniqueness and uncountably many measures of maximal entropy can be achieved by arbitrarily small C0 perturbations of the roof. That density statement is not in Kucherenko–Thompson, who construct specific examples; this paper gives a general perturbation argument. The proof is clean and short: use Israel's theorem to get many equilibrium states and Ruelle's theorem to get uniqueness, then transfer to the suspension flow via the Abramov–Kac formula and the pressure equation for flows. Theorem 3.13, an Israel-type density result for functions on the suspension flow, is a nice bonus. No circularity, no fitted parameters, standard external theorems only. The exposition is careful about one-sided versus two-sided shifts and time changes.\n\nThe soft spot is real but narrow. The standing assumption — topologically transitive and alphabet of size at least two — does not imply positive topological entropy. The proof of Proposition 3.4 says \"by our standing assumption (Σ,σ) has positive topological entropy,\" which is simply false. The period-two shift with transition matrix [[0,1],[1,0]] is transitive, has exactly two points, zero entropy, and a unique invariant measure. By Ambrose–Kakutani, every suspension flow over it has a unique invariant measure, so Theorem 1.1(a) cannot hold as stated. The proof also divides by h(Φτ), which is zero exactly in this case. The fix is straightforward: add h(σ)>0 (equivalently h(Φτ)>0) to the hypotheses. This also affects Remark 3.9: the listed conditions (finite entropy, entropy density, upper semi-continuity) are all satisfied by the period-two shift, yet Israel's conclusion fails because the set of ergodic measures is a singleton. Those conditions are not sufficient.\n\nEverything else checks out. The perturbation argument itself is sound for positive-entropy bases; I don't see any other load-bearing issues. The citations are appropriate, and the authors are honest about what comes from Israel, Ruelle, and Kucherenko–Thompson.\n\nThis deserves a serious referee. A competent referee will catch the missing hypothesis, the authors will correct it, and the paper will be a solid contribution to thermodynamic formalism. I would send it out with a request to add the positive-entropy condition and to fix Remark 3.9.","headline":"Smart, short density result for MME of suspension flows, but Theorem 1.1 as stated is false for zero-entropy transitive SFTs — an easy positive-entropy hypothesis fixes it.","tokens_in":9950,"tokens_out":3366,"would_cite":true,"duration_ms":34848,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["37D35","37A10","37A35"],"pacs":[],"model":"deepseek-v4-flash","headline":"An arbitrarily small uniform perturbation of the roof of a suspension flow over a sub-shift of finite type can produce either uncountably many ergodic measures of maximal entropy or exactly one.","keywords":["measures of maximal entropy","suspension flows","sub-shifts of finite type","thermodynamic formalism","equilibrium states","time reparametrization","continuous roof functions","topological entropy"],"falsifier":"Take the period-two shift on two symbols, with the transition matrix that has zeros on the diagonal and ones off the diagonal. It is topologically transitive and has alphabet size two, but it consists of exactly two points, has zero topological entropy, and has a unique invariant measure; the corresponding suspension flow therefore has a unique ergodic invariant measure for every continuous roof. This directly contradicts Theorem 1.1(a) as written, showing the theorem requires an explicit positive-entropy hypothesis.","tokens_in":8946,"feed_emoji":"","tokens_out":10438,"duration_ms":104462,"temperature":0.7,"pith_summary":"The paper asks whether a suspension flow over a sub-shift of finite type can be changed by an arbitrarily small uniform perturbation of its roof function—a tiny time reparametrization—so that its measures of maximal entropy (the invariant probability measures whose entropy equals the topological entropy) become either uncountably many or exactly one. It claims the answer is yes in both directions, making both behaviors dense in the space of continuous roof functions. This matters because the classical theory for Hölder roof functions gives a unique measure of maximal entropy, and the new result shows that uniqueness is not stable once the roof is merely continuous. The mechanism transfers the question to the base shift: flow measures of maximal entropy correspond to equilibrium states of a certain potential on the base, so the cardinality of equilibrium states is carried back to the flow.","feed_headline":"Small roof changes make maximal-entropy measures many or unique","feed_subtitle":"A small roof change yields either uncountably many or exactly one maximal-entropy measure.","key_machinery":"The carrying object is the dictionary between a suspension flow and its base: the Ambrose–Kakutani bijection $R(\\mu)=(\\mu\\times \\mathrm{Leb})|_Y/(\\mu\\times \\mathrm{Leb})(Y)$ sends base invariant measures to flow invariant measures; Abramov's formula gives flow entropy $h(R(\\mu))=h(\\mu)/\\int\\tau\\,d\\mu$; and the flow's topological entropy is the unique root of $P(-t\\tau)=0$, with flow measures of maximal entropy exactly $R(\\mu)$ for $\\mu$ an equilibrium state of $-h(\\Phi)\\tau$. The perturbation step takes a potential $\\psi$ with $P(-\\psi)=0$ and approximates it uniformly by potentials $\\phi_n$ with $P(-\\phi_n)=0$ whose equilibrium sets are, respectively, uncountable or a singleton; the roof is then changed to $\\tau_n=\\tau+(\\phi_n-h\\tau)/h$, preserving the original topological entropy while transferring the equilibrium cardinality to the flow.","core_discovery":"The paper's central claim is that, under its stated standing assumptions, any suspension flow over a one- or two-sided sub-shift of finite type with a positive continuous roof can have its roof perturbed by an arbitrarily small amount in the uniform norm so that the resulting flow has uncountably many ergodic measures of maximal entropy, and can also be perturbed so that it has a unique measure of maximal entropy. The perturbed flows are obtained by time reparametrization and have the same topological entropy as the original flow. The many-measures half is shown to extend to any base transformation whose thermodynamic formalism has finite entropy, entropy-dense ergodic measures, and an upper semi-continuous entropy map, while the uniqueness half extends to any expansive base map with the specification property.","pith_inferences":["Beyond the paper's claims: the stated standing assumptions are missing a positive-entropy hypothesis; the two-point period-two shift satisfies the stated assumptions but has zero entropy and a unique invariant measure, so Theorem 1.1(a) fails as written.","Beyond the paper's claims: the same dictionary suggests a general principle—any dense class of base potentials with prescribed equilibrium-state cardinality should yield an analogous density statement for measures of maximal entropy of suspension flows over that base.","Beyond the paper's claims: one testable extension is whether the same $C^0$-density dichotomy holds for suspension flows over countable Markov shifts or for roof regularities intermediate between continuous and Hölder.","Beyond the paper's claims: the uncountably many measures in the dense examples are produced by a purely existential theorem; identifying them explicitly, or showing they can form a Cantor set, is left open."],"forward_implications":["Both sets—flows with uncountably many ergodic measures of maximal entropy and flows with a unique such measure—are dense in the space of suspension flows over a fixed sub-shift of finite type.","Each perturbation can be made arbitrarily small in the uniform norm and preserves the topological entropy of the original flow, so the dichotomy is a property of nearby time changes rather than a special construction.","The dichotomy transfers from one-sided semi-flows to two-sided sub-shifts and genuine flows.","The uniqueness half extends to any expansive base map with specification, and the many-measures half to any compact base with finite entropy, entropy-dense ergodic measures, and an upper semi-continuous entropy map."],"supporting_citations":[{"why":"Supplies the dense set of continuous potentials with uncountably many ergodic equilibrium states; this is the engine for part (a) and the many-measures examples.","marker":"[Is]"},{"why":"Gives uniqueness of equilibrium states for Hölder potentials, which yields the unique-measure perturbation in part (b).","marker":"[Ru]"},{"why":"Provides the pressure equation $P(\\Delta_g-t\\tau)=0$ and the Abramov/Kac formulae linking flow and base thermodynamics.","marker":"[PP]"},{"why":"Provides Abramov's formula for flow entropy in terms of base entropy and roof integral, used throughout the identification of measures of maximal entropy.","marker":"[Ab]"},{"why":"Establishes the bijection between base invariant measures and flow invariant measures used to transfer equilibrium states to flow measures.","marker":"[AK]"},{"why":"Supplies the standard thermodynamic formalism facts about pressure, equilibrium states, and upper semi-continuity used in the perturbation lemma.","marker":"[Wa1]"},{"why":"Gives the cohomology reduction from two-sided shifts to one-sided shifts used to extend the results to genuine flows.","marker":"[Wa2]"}],"fun_headline_variants":["Tiny roof tweaks multiply or annihilate maximal-entropy measures","Roof shifts: uncountably many or one maximal-entropy measure","Small roof changes yield either a crowd or a single maximal-entropy measure","Perturb roof slightly, get many or one measure of maximal entropy","Roof perturbations flip maximal-entropy measure count to many or one"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The proof needs the base sub-shift to have positive topological entropy, but the paper's standing assumption only says topologically transitive with at least two symbols; the period-two shift satisfies that assumption and has zero entropy, so Theorem 1.1(a) as stated is false for it.","fun_headline_variants_meta":{"raw":{"variants":["Tiny roof tweaks multiply or annihilate maximal-entropy measures","Roof shifts: uncountably many or one maximal-entropy measure","Small roof changes yield either a crowd or a single maximal-entropy measure","Perturb roof slightly, get many or one measure of maximal entropy","Roof perturbations flip maximal-entropy measure count to many or one"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00074,"raw_usage":{"total_tokens":3212,"prompt_tokens":762,"completion_tokens":2450,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":378,"completion_tokens_details":{"reasoning_tokens":2355}},"tokens_in":378,"tokens_out":2450,"duration_ms":16687,"temperature":1.0,"reasoning_tokens":2355,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T12:30:12.487412+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take the period-two shift on two symbols, with the transition matrix that has zeros on the diagonal and ones off the diagonal. It is topologically transitive and has alphabet size two, but it consists of exactly two points, has zero topological entropy, and has a unique invariant measure; the corresponding suspension flow therefore has a unique ergodic invariant measure for every continuous roof. This directly contradicts Theorem 1.1(a) as written, showing the theorem requires an explicit positive-entropy hypothesis.","supporting_citations":[],"review_version":1}