{"id":"4a1bff3d-9b76-4637-a30f-cb91d70e2645","arxiv_id":"2505.07277","paper_version":1,"verdict":"ACCEPT","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"Multiplicative Toeplitz sequences are exactly Dirichlet characters away from a finite prime set, and unique Furstenberg systems for pretentious functions coincide with rational almost periodicity.","lead":"This paper classifies multiplicative sequences that repeat at every position with some period, showing they agree with a Dirichlet character outside a finite set of primes. It also gives an exact criterion for a pretentious sequence to have a unique statistical limit, and shows the corrected Elliott conjecture implies a major conjecture on these limits.","discovery_kind":"unification","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Theorem 0.5's t=0 step cites an unstated Halász-theorem consequence that appears to be wrong as written; this is the least secure link in the uniqueness characterization.","rationale":"The paper's central classification claims are attractive and the overall structure is coherent, but Theorem 0.5's proof has a specific unsupported step that is closer to the core than the [13] dependency highlighted by the reader. The claim that existence of the mean forces f^{φ(q)}(2^k) = -2^{kiφ(q)t} is not a standard Halász consequence as stated; standard asymptotics suggest no mean can exist for t≠0 under the pretentiousness condition, so the argument needs a different justification. If the step is repairable by a direct citation of the correct Halász theorem, the theorem may stand; if not, the dichotomy in Theorem 0.5 lacks proof. I therefore recommend a conditional acceptance pending verification of this step, while acknowledging the reader's accurate identification of the reliance on [13] as another genuine weak point.","tokens_in":29029,"tokens_out":40126,"duration_ms":403233,"concrete_test":"Verify the cited Halász consequence by computing the Dirichlet series for a model g with g(p)=p^{iu}e^{iθ_p}, Σθ_p^2/p<∞, and showing that the pole at s=1+iu is not canceled when u≠0, so the mean fails for all such g. Then re-derive the necessary condition for mean existence from Halász's theorem and compare it with the displayed condition in §2.2; if the condition differs, recompute the contradiction step with the correct condition and check whether the conclusion t=0 still follows.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"In the proof of Theorem 0.5, direction (i)⇒(ii), after assuming a unique Furstenberg system and t≠0, the paper states: \"Since the mean of f^{φ(q)} must exist ... f^{φ(q)}(2^k) = -2^{kiφ(q)t} for all k≥1 in view of Halász theorem.\" No theorem number or statement is supplied. Standard Halász asymptotics for g∈M with D(g,n^{iu})<∞ and u≠0 give (1/N)Σ_{n≤N}g(n) ≍ N^{iu}/(1+iu) up to slowly varying factors, so no Cesàro mean exists; a special value at 2^k cannot remove the pole at s=1+iu in the Dirichlet series because the pretentiousness condition forces local factors to be close to 1. If the intended condition is g(2^k)=(-2^{iu})^k, the displayed formula lacks parentheses and is not the stated Halász theorem. This is the only argument excluding t≠0, so the dichotomy between a unique ergodic odometer and uncountably many non-ergodic systems rests on an unsupported citation. The reader's concern about [13] is legitimate, but this internal step is more immediate and equally load-bearing.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper gives a full classification of multiplicative functions that are Toeplitz (Theorem 0.1: they coincide with a Dirichlet character outside a finite set of primes) and a characterization of pretentious multiplicative functions with a unique Furstenberg system (Theorem 0.5: uniqueness holds iff t=0 and the prime series (11) converges, iff the function is rationally almost periodic). It also proves that all Toeplitz multiplicative functions are regular, that real-valued pretentious functions have a unique Furstenberg system, that the corrected Elliott conjecture implies the Frantzikinakis–Host conjecture, and it clarifies several relations between aperiodicity notions.","tokens_in":29239,"tokens_out":47935,"duration_ms":418264,"significance":"If correct, Theorem 0.1 settles the classification of multiplicative Toeplitz sequences, and Theorem 0.5 provides a clean dichotomy for Furstenberg systems of pretentious functions: one ergodic odometer versus uncountably many isomorphic non-ergodic systems. The combination of pretentious number theory and ergodic theory is elegant, and the paper gives detailed proofs for most steps. The main caveat is a missing justification of a critical pointwise identity in the proof of Theorem 0.5; once that is supplied, the central claims are likely sound.","major_comments":[{"comment":"The step \"Since the mean of f^{φ(q)} must exist ... f^{φ(q)}(2^k) = -2^{kiφ(q)t} for all k≥1 in view of the Halász theorem\" is not justified. Halász's theorem provides asymptotic formulas for the partial sums of a multiplicative function in terms of the minimum of D(g, n^{it}); it does not directly yield pointwise identities for f(2^k). What is needed is a lemma: if g∈M, D(g, n^{iu})<∞ with u≠0, and the Cesàro mean of g exists, then the Dirichlet series of g must be holomorphic at s=1+iu, which forces the local factor at the prime 2 to satisfy G_2(1+iu)=0; since G_2(z)=∑_{k≥0} g(2^k) z^k has |z|=1/2 and coefficients bounded by 1, the only way this can happen is g(2^k)=-2^{iku} for all k≥1. This argument is absent, and without it the exclusion of t≠0 in (i)⇒(ii) is unsupported. Please either prove this lemma or provide a precise reference stating it.","section":"Section 2.2 (proof of Theorem 0.5, (i)⇒(ii))"}],"minor_comments":[{"comment":"The sentence \"f^2 is pretentious (in fact, it is RAP, see Remark 2.12 in [5])\" is inaccurate: for an aperiodic real-valued f, f^2 need not be pretentious (for example, if f(p)=0 on a set of primes with divergent reciprocal sum, then f^2 is not pretentious), even though it is RAP. Since the subsequent argument only uses the RAP property, the claim should be rephrased to state that f^2 is RAP and hence generic.","section":"Proof of Corollary 0.6, Case 2"},{"comment":"There are numerous typographical and spacing issues in the abstract and text (e.g., \"M oreover\" in the abstract, \"pretencious\" in the introduction). A careful proofread is recommended.","section":"Throughout"}],"recommendation":"major_revision","confidential_remarks":"The paper is a strong contribution and the results are likely correct, but the proof of Theorem 0.5 currently has a load-bearing gap at equation (17). The other main results (Theorem 0.1 and the periodic/automatic classifications) appear sound. The reliance on the structure theory of [13] is appropriate given its recent acceptance in ETDS. I would be willing to review a revised version."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Core: Theorem 0.1 and Theorem 0.5 are the new substance. The Toeplitz characterization is a clean, unconditional statement that genuinely improves on the automatic-sequence analogy, and the uniqueness criterion for Furstenberg systems gives the field what [13] left open: a number-theoretic way to see genericity. The conditional Corollary 0.6 is honestly labeled and does what it claims. The examples and the aperiodic-class clarifications are useful.\n\nThe paper does well to keep the proofs detailed. The structure arguments in Section 1 (Lemma 1.2, Lemma 1.9, Theorem 1.11) are careful and fill a gap that earlier survey treatments left informal. The use of [13]'s structure theorem is legitimate: it is a published recent result, and the paper does not try to reprove it; the new criterion is a separate, genuinely new result.\n\nNow the stress-test. The flagged step in Theorem 0.5 (i)⇒(ii) — the claim that mean existence plus pretentiousness forces f^{\\phi(q)}(2^k) = -2^{ki\\phi(q)t} — is terse but correct. Halász's theorem says the mean can exist only if the singularity at s=1+i\\phi(q)t is removable. Since D(f^{\\phi(q)}, n^{i\\phi(q)t}) < ∞, the Dirichlet series has a pole at that point coming from the primes where the function pretends to the Archimedean character. The only way to cancel it is a zero of a local factor. For any prime p≥3, the local factor at s=1+iu has modulus at least 1/2 (by the triangle inequality), so it cannot vanish. The 2-local factor can vanish, and equality in the triangle inequality forces g(2^k) = -2^{iku} for every k. So the contradiction argument is sound. The paper would be easier to referee if that step cited a specific lemma (or proved it in a sentence), but it is not wrong.\n\nThe weakest point is simply the dependence on the heavy machinery of [13]; readers who don't know that paper will have to take a lot on faith. That is standard for this area, and not a defect of the present work. There are minor typos (the abstract's 'M oreover') and a small clash of notation (F the set vs F the function), but nothing that affects the math.\n\nVerdict: this deserves a serious referee. I'd send it out; the main theorems will be citable. My own take is that it's a solid paper, likely publishable in a good journal without major revision.","headline":"The main classification theorems are real and the proofs hold up; the one flagged step in Theorem 0.5 is terse but correct.","tokens_in":29812,"tokens_out":21324,"would_cite":true,"duration_ms":184960,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["11A25","11N37","37A45"],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper proves that a nonzero multiplicative function is Toeplitz exactly when it agrees with a Dirichlet character off a finite set of primes, and that a pretentious multiplicative function has a unique Furstenberg system exactly when…","keywords":["multiplicative functions","Toeplitz sequences","Dirichlet characters","pretentious functions","Furstenberg systems","rationally almost periodic","automatic sequences","correlation conjecture"],"falsifier":"To refute Theorem 0.1, exhibit a multiplicative function that is Toeplitz but, for every Dirichlet character and every finite set of primes, differs from the character at some integer coprime to all primes in the set. To refute Theorem 0.5, produce a bounded pretentious $f$ with $D(f,\\chi n^{it})<\\infty$ whose correlation limits all exist while either $t\\neq 0$ or the series $\\sum_p \\frac1p(1-f(p)\\overline{\\chi(p)})$ diverges; the paper predicts no such function exists.","tokens_in":28800,"feed_emoji":"🔢","tokens_out":9131,"duration_ms":84329,"temperature":0.7,"pith_summary":"The paper gives two classification theorems for bounded multiplicative functions. Theorem 0.1 says a nonzero multiplicative function is Toeplitz precisely when, outside a finite set of primes, it equals a Dirichlet character; such functions are necessarily regular, so they determine a unique Furstenberg system. Theorem 0.5 says that among pretentious functions, having exactly one Furstenberg system is equivalent to the twist parameter being zero and the prime series $\\sum_p \\frac1p(1-f(p)\\overline{\\chi(p)})$ converging, equivalently to being rationally almost periodic. A corollary is that every real-valued pretentious multiplicative function is generic, and that the corrected correlation conjecture for strongly non-pretentious functions would imply the uniqueness conjecture for Furstenberg systems. The paper also sorts out how automatic, Toeplitz, periodic, and aperiodic classes sit inside the pretentious world.","feed_headline":"Toeplitz multiplicative = Dirichlet character off finite primes","feed_subtitle":"Pretentious functions have one Furstenberg system exactly when they are rationally almost periodic.","key_machinery":"The load-bearing objects are the multiplicative distance $D(f,g)^2=\\sum_p \\frac1p(1-\\mathrm{Re}(f(p)\\overline{g(p)}))$, Toeplitz period structures with essential periods and $p$-valuations, and Furstenberg systems obtained as weak* limits of shifts of $f$. For Theorem 0.1, multiplicativity forces the set of primes dividing the periods of the position 1 to be finite, so $f$ agrees with a Dirichlet character on all integers coprime to that finite set. For Theorem 0.5, the proof converts uniqueness of the Furstenberg system into convergence of a correlation product $\\prod_p M_p$ whose local factors satisfy $M_p=1-\\frac{2}{p}(1-f(p)\\overline{\\chi(p)})+O(p^{-2})$ for large $p$, so convergence of the product is exactly convergence of the displayed prime series; the odometer/discrete-spectrum structure theorem from the theory of Furstenberg systems of pretentious functions supplies the dynamical scaffolding.","core_discovery":"The central discovery is that two apparently dynamical notions, being Toeplitz and having a unique Furstenberg system, are, for multiplicative functions, purely number-theoretic conditions. Theorem 0.1 states that a nonzero multiplicative $f:\\mathbb{N}\\to\\mathbb{C}$ is Toeplitz iff there exist a Dirichlet character $\\chi$ and a finite set $F$ of primes such that $f(n)=\\chi(n)$ for every $n$ coprime to every element of $F$, and moreover every such $f$ is regular, hence uniquely ergodic. Theorem 0.5 states that for $f\\in\\mathcal{M}$ with $D(f,\\chi n^{it})<\\infty$, the function $f$ is generic (has exactly one Furstenberg system) iff $t=0$ and $\\sum_p \\frac1p(1-f(p)\\overline{\\chi(p)})$ converges, iff $f$ is Besicovitch rationally almost periodic, with the unique system being an ergodic odometer. The proof combines the period-structure machinery of Toeplitz sequences with correlation formulas and the structure theorem that Furstenberg systems of pretentious functions are discrete-spectrum odometer systems.","pith_inferences":["The criterion turns the one-or-uncountably-many dichotomy for Furstenberg systems of pretentious functions into a decision problem: compute $t$ and evaluate the prime series, so uniqueness becomes testable rather than a purely dynamical fact.","The paper's Corollary 2.8 extends the Toeplitz rigidity: even an infinite exceptional set with $\\sum_{p\\in F}1/p<\\infty$ cannot destroy uniqueness; this suggests that any counterexample to uniqueness for non-pretentious functions must be strongly spread out in the multiplicative sense.","One testable extension: if the rate of convergence in the prime series is controlled, one should be able to bound the rate at which correlations of $f$ converge; the paper does not quantify this.","The equivalence between Toeplitz and 'Dirichlet character off a finite set' also explains why the BBC condition is closed under multiplication, whereas automaticity is not; the two notions diverge exactly by the finite exceptional primes."],"forward_implications":["Every Toeplitz multiplicative function is regular, so its subshift is uniquely ergodic and its unique Furstenberg system is an odometer.","For pretentious functions, uniqueness of the Furstenberg system is checkable from prime values: $t=0$ and $\\sum_p \\frac1p(1-f(p)\\overline{\\chi(p)})<\\infty$.","All real-valued pretentious multiplicative functions and all finitely valued pretentious multiplicative functions are rationally almost periodic and generic, with an ergodic odometer as the unique Furstenberg system.","If a pretentious function has $t\\neq 0$, it has uncountably many pairwise isomorphic non-ergodic Furstenberg systems; uniqueness and ergodicity always go together.","The corrected correlation conjecture for strongly non-pretentious functions implies the uniqueness conjecture for every real-valued multiplicative function bounded by 1."],"supporting_citations":[{"why":"Provides the structure theorem that every Furstenberg system of a pretentious function is an odometer system with discrete spectrum; this underpins the (i) implies (ii) step and Corollary 2.3.","marker":"[13]"},{"why":"Shows that for multiplicative functions, RAP and Besicovitch almost periodicity coincide; this is the bridge in the (iii) iff (ii) step.","marker":"[5]"},{"why":"Gives the result that a multiplicative function is Besicovitch almost periodic with nonempty spectrum exactly when the series in (11) converges.","marker":"[7]"},{"why":"Supplies the correlation formulas whose local factors $M_p$ are expanded to obtain the prime series from convergence of the correlation product.","marker":"[25]"},{"why":"Proves that rationally almost periodic sequences are generic, yielding (iii) implies (i).","marker":"[4]"},{"why":"Introduces strongly non-pretentious functions and the corrected correlation conjecture used in the proof of Corollary 0.6.","marker":"[32]"},{"why":"Classifies automatic multiplicative sequences, used in Proposition 0.4 to locate automatic functions within the Toeplitz picture.","marker":"[30]"}],"fun_headline_variants":["Toeplitz iff Dirichlet character off finite prime set","Unique Furstenberg system iff rationally almost periodic","Toeplitz multiplicative functions: Dirichlet with finite exceptions","Corrected Elliott conjecture forces unique Furstenberg system"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The classification rests on the earlier structure theorem that every Furstenberg system of a pretentious function has discrete spectrum whose ergodic components are odometers; if that theorem were false or incomplete, the equivalence in Theorem 0.5 would lose its foundation.","fun_headline_variants_meta":{"raw":{"variants":["Toeplitz iff Dirichlet character off finite prime set","Unique Furstenberg system iff rationally almost periodic","Toeplitz multiplicative functions: Dirichlet with finite exceptions","Corrected Elliott conjecture forces unique Furstenberg system"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000805,"raw_usage":{"total_tokens":3524,"prompt_tokens":924,"completion_tokens":2600,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":540,"completion_tokens_details":{"reasoning_tokens":2534}},"tokens_in":540,"tokens_out":2600,"duration_ms":17877,"temperature":1.0,"reasoning_tokens":2534,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-15T22:21:52.341344+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"To refute Theorem 0.1, exhibit a multiplicative function that is Toeplitz but, for every Dirichlet character and every finite set of primes, differs from the character at some integer coprime to all primes in the set. To refute Theorem 0.5, produce a bounded pretentious $f$ with $D(f,\\chi n^{it})<\\infty$ whose correlation limits all exist while either $t\\neq 0$ or the series $\\sum_p \\frac1p(1-f(p)\\overline{\\chi(p)})$ diverges; the paper predicts no such function exists.","supporting_citations":[{"cited_title":"Furstenberg systems of pretentious and MRT multiplicative functions","cited_arxiv_id":"2304.03121","evidence_quote":"Provides the structure theorem that every Furstenberg system of a pretentious function is an odometer system with discrete spectrum; this underpins the (i) implies (ii) step and Corollary 2.3."},{"cited_title":"Bergelson, J","cited_arxiv_id":null,"evidence_quote":"Shows that for multiplicative functions, RAP and Besicovitch almost periodicity coincide; this is the bridge in the (iii) iff (ii) step."},{"cited_title":"Daboussi, H","cited_arxiv_id":null,"evidence_quote":"Gives the result that a multiplicative function is Besicovitch almost periodic with nonempty spectrum exactly when the series in (11) converges."},{"cited_title":"Klurman, Correlations of multiplicative functions and application s, Compos","cited_arxiv_id":null,"evidence_quote":"Supplies the correlation formulas whose local factors $M_p$ are expanded to obtain the prime series from convergence of the correlation product."},{"cited_title":"Bergelson, J","cited_arxiv_id":null,"evidence_quote":"Proves that rationally almost periodic sequences are generic, yielding (iii) implies (i)."},{"cited_title":"Matomäki, M","cited_arxiv_id":null,"evidence_quote":"Introduces strongly non-pretentious functions and the corrected correlation conjecture used in the proof of Corollary 0.6."},{"cited_title":"Konieczny, M","cited_arxiv_id":null,"evidence_quote":"Classifies automatic multiplicative sequences, used in Proposition 0.4 to locate automatic functions within the Toeplitz picture."}],"review_version":1}