{"id":"2d1b7801-72a8-4343-a87e-097d697ad8ba","arxiv_id":"2507.14113","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Density of finitely supported invariant measures holds for all automorphisms of compact abelian groups with the descending chain condition.","lead":"This paper proves that for automorphisms of compact abelian groups satisfying a finiteness condition, every invariant measure can be approximated by measures supported on finitely many periodic orbits. It settles an open question and yields new results on Hilbert-Schmidt stability and a Livshitz-type theorem.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Ergodic half of Theorem 1.1 rests on a proof sketch: Theorem 5.13 omits the measure estimates needed to derive dense ergodic periodic measures from partial specification.","rationale":"The central claim of the paper is Theorem 1.1, whose ergodic half depends entirely on Theorem 5.13. That theorem is presented as a sketch with 'technical details' omitted, and the missing details are precisely the quantitative measure estimates needed to make the partial specification property imply density of ergodic periodic measures. This is a genuine gap in the written proof and is load-bearing for the main theorem. I do not think the Schmidt-Miles-Thomas structure theorem (Proposition 3.1) is the weakest point: it is a standard, deep result from Schmidt's book, and the paper's proof of the proposition appears consistent with that source. The non-ergodic part (Theorem 7.8) is proved in detail and seems sound, as does Theorem 1.2. Separately, the reader's second issue is correct: Theorem 1.3's application of Lemma 9.1 fails because the least periods in X_2 and X_3 are, respectively, {2n} and {3n}, so pairs such as 4 and 6 are not coprime; this is a real error but peripheral to the central claim. I therefore agree with the reader's conditional verdict, but for a different reason than the stated weakest assumption: the unresolved proof gap in Theorem 5.13. A full proof of the measure estimates, or a counterexample showing they fail, would settle the concern.","tokens_in":33707,"tokens_out":25783,"duration_ms":696370,"concrete_test":"Write out the missing proof of Theorem 5.13. Concretely, fix μ and ε, choose ergodic components μ_i with weights t_i and generic points z_i; set n_i = ⌊t_i N⌋, insert M-gaps, let b_r = Σ(n_i+M), choose n ∈ P minimal with n ≥ (1+ε)b_r, and let y be the periodic point from Definition 4.2 for ε/4. Prove that d_M(m_{T,y,n}, Σ t_i μ_i) ≤ C ε for all sufficiently large N, using Lemma 7.3 to bound the measure distance by the fraction of indices where the trace fails (≤ ε + rM/n + (n−b_r)/n). If this estimate cannot be derived, then partial specification as defined is insufficient and Theorem 5.13 needs a stronger hypothesis; if it can, the ergodic half is established. Additionally, verify that the minimal n∈P satisfies n−b_r = O(ε b_r + G) where G is the gap constant of P.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Theorem 5.13, which proves the ergodic half of Theorem 1.1, is not actually proved: §5.3 says 'we only sketch the proof and omit the technical details'. The sketch asserts that partial specification implies density of ergodic periodic measures by analogy with [Mar80, Sig70, GK18]. A full proof must show that for any μ ∈ M_T(X) and ε>0 one can choose ergodic components μ_i, μ_i-generic points z_i, segment lengths n_i ≥ N, gaps M from Definition 4.2 for ε/4, and n∈P with n ≥ (1+ε)b_r, so that the periodic point y of period n from partial specification satisfies d_M(m_{T,y,n}, μ) < ε. The missing estimates are: (i) the proportion of bad indices in each segment, which the partial tracing bounds only by ε, must be converted into a bound on the measure distance using Lemma 7.3 as in Proposition 7.4; (ii) the forced intervening gaps contribute rM points whose orbit is unconstrained, so rM/n must be small; (iii) the tail n−b_r must be o(b_r), which requires choosing n∈P minimal above (1+ε)b_r and using the bounded-gap property of P. None of these estimates appear; the assertion that the argument is 'no different' from the periodic-specification case does not address the weaker nature of partial specification. If, for example, the minimal n∈P exceeds b_r by a non-negligible factor, the unconstrained tail could dominate the empirical measure and the approximation fails. Thus Theorem 5.13 is unproved as written.","agreement_with_reader":"disagree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This paper studies continuous automorphisms of compact metrizable abelian groups satisfying the descending chain condition (dcc). The main theorem (Theorem 1.1) asserts that finitely supported invariant measures are weak-* dense in the space of all invariant probability measures, and that under Haar-ergodicity the finitely supported ergodic invariant measures are dense. The proof introduces a notion of partial specification (Definition 4.2), proves it for solenoids and shows that it is preserved under group extensions, and combines it with a structure theorem for dcc abelian group automorphisms and an analysis of unipotent toral automorphisms. The paper also derives a Hilbert-Schmidt stability result for Z⋉G, a Livshitz-type coboundary characterization, and a counterexample showing that dense periodic measures do not pass to products in general dynamical systems.","tokens_in":33959,"tokens_out":10126,"duration_ms":123650,"significance":"If the main theorem were fully established, it would resolve an open question of Levit and Vigdorovich and of Eckhardt, and the partial specification property would be a useful new tool for algebraic dynamical systems. The paper's detailed contributions are substantial: the solenoid partial specification (Propositions 5.7-5.9), the extension result (Proposition 4.5), the unipotent measure classification (Proposition 6.1), and the product counterexample (Section 9) are concrete and largely self-contained. However, the proof of the ergodic density theorem is only sketched, and since that theorem is the headline result, the significance cannot be assessed as fully realized in the present version.","major_comments":[{"comment":"The proof of the ergodic half of Theorem 1.1 is not actually supplied. The text states that the argument is only sketched and technical details are omitted, and it appeals by analogy to [Mar80, Sig70, GK18]. This is load-bearing: Theorem 1.1's second assertion and the Y=X case of Proposition 7.4 both depend on it. A complete proof must convert the partial tracing guarantees of Definition 4.2 into a bound on d_M(m_{T,y,n}, μ), using Lemma 7.3 in the manner of Proposition 7.4. Specifically, one must control the ε-fraction of bad indices in each traced segment, the M-point gaps between segments, and the tail n−b_r, using the bounded-gap property of P and the freedom to choose n∈P with n≥(1+ε)b_r. None of these estimates appears. The assertion that partial specification is 'no different' from full periodic specification does not address the fact that Definition 4.2 only traces (1−ε) of each segment and imposes no upper bound on n−b_r. Without these estimates, the ergodic density theorem is unproved as written.","section":"§5.3, Theorem 5.13"},{"comment":"The case Y=X is dismissed with the sentence 'the density of periodic measures follows from the partial specification (as in the proof of Theorem 5.13).' Since Theorem 5.13 is not proved, this gap also propagates to the proof of Theorem 7.8 in the case X2=X1. The author should either supply the missing estimates once and use them in both places, or prove the Y=X case directly.","section":"§7, Proposition 7.4, first paragraph"}],"minor_comments":[{"comment":"The sentence 'Since q can be any arbitrarily large number in cN, by enlarging q if necessary we may assume that q ≥ N and that q ∈ P⊆cN' is logically imprecise, because P is only a bounded-gap subset of cN. One should explicitly choose q∈P∩[Q,∞) and then apply Proposition 6.7 to that q.","section":"§7, proof of Proposition 7.4, after Eq. (7.3)"},{"comment":"The choice of δ that makes Eq. (5.15) hold would be clearer if stated explicitly, for example δ < (1/2) min_{i,m} d_G(g_i^{cm}, 1), rather than left as an implicit sufficiently small choice.","section":"§5.1, Lemma 5.10"},{"comment":"The phrase 'since Y is compact, the closures coincide in Y and Xi' is unclear; the argument only needs that the chosen open neighborhoods in Y are also open in Xi, which follows from the subspace topology.","section":"Section 9, Lemma 9.2"}],"recommendation":"major_revision","confidential_remarks":"The manuscript's detailed parts are strong, but the central ergodic density theorem is only sketched. I would not accept the paper in its present form. If the author supplies a complete proof of Theorem 5.13, with the missing measure estimates, the paper is likely suitable for publication."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Rotem Yaari's paper makes a real advance: it introduces partial specification, proves it for all ergodic dcc abelian group automorphisms, and uses it to prove density of finitely supported invariant measures for the whole class. The solenoid proof is careful, the extension lemma is clean, and the unipotent analysis in Section 6 is substantial. The non-ergodic half (Theorem 7.8) is essentially complete, and the corollaries on Hilbert-Schmidt stability and Livshitz theory are nice applications.\n\nThe soft spots are concentrated in two places. The ergodic half of the main theorem rests on Theorem 5.13, which is explicitly a sketch: the proof says the details are 'no different' from the classical specification case and omits them. The stress-test note is right that this is not a routine omission. To get dense ergodic periodic measures from partial specification you need to control three things: the proportion of bad indices in each traced segment, the contribution of the forced gaps between segments, and the unconstrained tail when the chosen period n exceeds the last segment by a small factor. None of these estimates appear. It may well be fixable, but as written the central claim of the paper is not proved.\n\nThe second problem is Theorem 1.3. Lemma 9.1 requires that every pair of periodic points in the two factors have coprime least periods. But X_2 contains a point of least period 2 and X_3 contains a point of least period 6, and gcd(2,6) is not 1. So the lemma's hypothesis is false, and the proof of the product counterexample collapses. The counterexample might be true—the idea of using Thue-Morse with inserted separators is plausible—but this argument does not establish it.\n\nNo circularity, no fitted parameters, and the external theorems are used appropriately. The paper deserves a serious referee, but the referee should require a complete proof of Theorem 5.13 and a corrected argument for Theorem 1.3 before acceptance. I would not cite the main theorem in its current form, but I would bring the paper to a reading group: the definition of partial specification and the structure-theoretic proof for solenoids are worth discussing on their own.","headline":"Strong result with a genuinely useful new specification variant, but the ergodic half and the product counterexample are not proved as written; both gaps look fixable.","tokens_in":34569,"tokens_out":3772,"would_cite":false,"duration_ms":42695,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["37A05","37B05","37A45"],"pacs":[],"model":"deepseek-v4-flash","headline":"Every invariant measure of a compact abelian group automorphism satisfying the descending chain condition is a weak-* limit of finitely supported invariant measures, with ergodic approximants when the automorphism is Haar-ergodic.","keywords":["abelian group dynamical systems","descending chain condition","invariant measures","periodic measures","partial specification","solenoids","Hilbert-Schmidt stability","Livshitz theorem"],"falsifier":"Find a dcc abelian group automorphism, for example an S-adic solenoid with irreducible and non-cyclotomic characteristic polynomial, and an ergodic invariant measure μ such that some open set U has μ(U) > 0 while every sufficiently long periodic orbit spends negligible time in U; that would contradict the predicted partial specification and the density of periodic measures.","tokens_in":33422,"feed_emoji":"🔄","tokens_out":8145,"duration_ms":90780,"temperature":0.7,"pith_summary":"The paper proves that for a compact metrizable abelian group with a continuous automorphism satisfying the descending chain condition, every invariant probability measure is a weak-* limit of finitely supported invariant measures; if the automorphism is ergodic with respect to Haar measure, the approximants can be chosen ergodic and finitely supported. This settles a question posed by Levit and Vigdorovich and by Eckhardt, and it is new even for automorphisms of the torus with block-diagonal hyperbolic and unipotent blocks. The proof's engine is a new 'partial specification' property, weaker than classical specification, that still yields dense periodic measures and is preserved under group extensions. Two corollaries follow: Hilbert-Schmidt stability for finitely generated semidirect products Z⋉G with G countable abelian, and a Livshitz-type theorem describing the uniform closure of coboundaries by vanishing on periodic orbits. A separate construction shows that, for general dynamical systems, the property of having dense periodic measures is not preserved under products.","feed_headline":"All invariant measures are limits of periodic orbit measures","feed_subtitle":"On compact abelian automorphisms with the descending chain condition, ergodic approximants and stability results follow.","key_machinery":"Partial specification (Definition 4.2) is the central object: a system satisfies partial specification with periods P, a bounded-gaps subset of N, if every sufficiently spaced finite collection of orbit segments can be ε-partially traced—approximated on all but an ε-fraction of each segment—by a point of prescribed period n ∈ P, for all sufficiently large n. Unlike full specification, partial specification is preserved under abelian group extensions (Proposition 4.5). For ergodic dcc systems, the paper establishes partial specification by first decomposing the system via the Schmidt-Miles-Thomas structure theorem (Proposition 3.1) into a finite chain whose quotients are Bernoulli shifts or S-adic solenoids with irreducible non-cyclotomic characteristic polynomial, then proving partial specification for each solenoid via adapted p-adic norms and bounded-below sets of periods, and finally lifting through extensions. For the non-ergodic part, a dynamically unipotent toral factor is handled by the measure classification of unipotent toral automorphisms.","core_discovery":"The central claim is Theorem 1.1: if (X,T) is an abelian group dynamical system—a compact metrizable abelian group with a continuous automorphism—and T satisfies the descending chain condition (every decreasing chain of T-invariant closed subgroups stabilizes), then the finitely supported invariant probability measures are weak-* dense in the space of all invariant probability measures. In the Haar-ergodic case, the dense subset can be taken to consist of finitely supported ergodic measures. The result answers a question posed by Levit and Vigdorovich and by Eckhardt, and it is new even for toral automorphisms combining hyperbolic and unipotent blocks, which had previously been treated only in separate cases.","pith_inferences":["The partial specification property is likely to deliver other specification-type consequences for dcc abelian group automorphisms, such as periodic orbit counting laws, exponential recurrence, or large-deviation estimates, since the property is strong enough to substitute for specification in several standard arguments.","Because the proof routes through solenoids and adapted p-adic norms, the same technique may extend to algebraic actions of higher-rank abelian groups on compact abelian groups under a dcc-type hypothesis, provided the structure theorem has an analogue in that setting.","The X2 × X3 counterexample suggests that 'dense periodic measures' is a delicate property for general dynamical systems; the extension-friendly object is the stronger partial specification, so systems with partial specification may be better suited to stability and cohomological questions than systems with merely dense periodic measures.","The Livshitz corollary might admit a sharper, non-uniform version: with partial specification replacing manifold regularity assumptions, the uniform closure in Corollary 1.6 could perhaps be replaced by a genuine coboundary statement for continuous functions on ergodic systems; the paper hints at such a version but leaves the details aside."],"forward_implications":["Every invariant probability measure of a dcc abelian group automorphism, including non-ergodic ones, is a weak-* limit of measures supported on finite orbits; in the Haar-ergodic case, the approximating finite-orbit measures can themselves be made ergodic.","The result answers the question posed by Levit and Vigdorovich and by Eckhardt, and extends known dense-periodic-measures results from hyperbolic and ergodic toral automorphisms to all dcc abelian group automorphisms, including mixed hyperbolic-unipotent block diagonal toral automorphisms.","Corollary 1.5 follows: every finitely generated group of the form Z⋉G, with G countable abelian, is Hilbert-Schmidt stable, via the established equivalence between dense periodic measures and Hilbert-Schmidt stability.","Corollary 1.6 follows: a continuous function on a dcc abelian group automorphism is a uniform limit of coboundaries exactly when it sums to zero along every periodic orbit.","Theorem 1.3 shows that, in general topological dynamics, the property of dense periodic measures is not closed under products, so the paper's group-extension stability of partial specification is essential for the main theorem."],"supporting_citations":[{"why":"Supplies the structure theorem (Corollary 6.3 and Theorem 6.5(1)) decomposing ergodic dcc abelian group automorphisms into finite chains of Bernoulli shifts and solenoids, the foundation of the partial specification proof.","marker":"[Sch95]"},{"why":"Poses the motivating question about dense periodic measures and provides the equivalence (Proposition 10.2) between dense periodic measures and Hilbert-Schmidt stability used in Corollary 1.5.","marker":"[L V24]"},{"why":"Provides the technique of dense ergodic periodic measures for ergodic toral automorphisms and the bounded-below set construction for unimodular roots, adapted here to solenoids.","marker":"[Mar80]"},{"why":"Together with [KS89], establishes density of periodic points in dcc abelian group dynamical systems, used to reduce the non-ergodic case.","marker":"[LP67]"},{"why":"Establishes density of periodic points for dcc abelian group automorphisms and supports the algebraic duality arguments for chains of subgroups.","marker":"[KS89]"},{"why":"Provides specification results for solenoidal automorphisms, the model for the tracing arguments that yield partial specification for S-adic solenoids.","marker":"[ADK82]"},{"why":"Classifies invariant measures for actions of subgroups generated by unipotent elements, used in Proposition 6.1 to classify ergodic measures of unipotent toral automorphisms.","marker":"[Sha98]"},{"why":"Establishes unique ergodicity of ergodic affine transformations with unipotent linear part on the torus, used in Proposition 6.3 to show unique ergodicity of unipotent toral restrictions.","marker":"[Hah63]"},{"why":"Provides the general identity (Proposition 10.13) identifying the uniform closure of coboundaries with functions vanishing on all invariant measures, used in Corollary 1.6.","marker":"[Kat03]"}],"fun_headline_variants":["Finite orbit measures approximate all invariant measures","Dense periodic orbit measures for abelian automorphisms","Finitely supported measures dense for compact abelian automorphisms","Approximating invariant measures by finite orbit measures"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The argument rests on the external structure theorem that every ergodic dcc automorphism is a continuous factor of a finite chain of Bernoulli shifts and solenoids with irreducible non-cyclotomic characteristic polynomials; if that decomposition failed, the partial specification proof and with it the ergodic density statement would collapse.","fun_headline_variants_meta":{"raw":{"variants":["Finite orbit measures approximate all invariant measures","Dense periodic orbit measures for abelian automorphisms","Finitely supported measures dense for compact abelian automorphisms","Approximating invariant measures by finite orbit measures"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00069,"raw_usage":{"total_tokens":3087,"prompt_tokens":866,"completion_tokens":2221,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":482,"completion_tokens_details":{"reasoning_tokens":2158}},"tokens_in":482,"tokens_out":2221,"duration_ms":16316,"temperature":1.0,"reasoning_tokens":2158,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-06T16:01:50.006052+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Find a dcc abelian group automorphism, for example an S-adic solenoid with irreducible and non-cyclotomic characteristic polynomial, and an ergodic invariant measure μ such that some open set U has μ(U) > 0 while every sufficiently long periodic orbit spends negligible time in U; that would contradict the predicted partial specification and the density of periodic measures.","supporting_citations":[],"review_version":1}