{"id":"1f90a3ab-3b51-4cf2-96cd-e7f089edea1f","arxiv_id":"1908.09387","paper_version":2,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":1,"one_line_summary":"The paper proves that for every n, [0,n]∪{ω} is the spectrum of recursive models of a flat, non-disintegrated, model complete strongly minimal theory in a finite relational language.","lead":"This paper constructs a strongly minimal theory whose recursive models have spectrum exactly [0,n]∪{ω}, a shape not previously known for any strongly minimal theory. The construction introduces a new 'unblockable extension' technique in Hrushovski-style amalgamation, which may simplify future interactions between recursion theory and Hrushovski constructions.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Lemma 6.1's 'never removed' obstruction claim fails when the algebraic witness Y uses only R and the dynamic R_i; the diagonalization may leave intermediate spectra alive.","rationale":"The reader's weakest assumption pinpoints the exact place where I find a concrete gap. The diagonalization in Lemma 6.1 depends on the assertion that the algebraic witness Y, once supplemented by the dynamically holding relation R_i(\\bar b c), forms an obstruction that is never removed. But the paper's own removal criterion in Step 1 says an obstruction is removed when δ_t(Y) ≥ |\\bar b_i|, with δ_t counting only non-suspicious relation symbols. In the infinite outcome R_i is suspicious forever, so excluding R_i from δ_t makes δ_t(Y) equal to δ_{\\hat L}(Y) unless Y contains some other relation symbol that is limited away but not suspected. The proof does not establish that such a symbol must occur in Y, and the dynamic R_i(\\bar b c) is witnessed by Ω-extensions using only R, so Y may well be R-only. In that case the obstruction is removed under the stated definition, the strategy does not get stuck, and the contradiction in the infinite outcome collapses. This would allow c to remain algebraic, potentially admitting a recursive model of dimension n+1 and contradicting the claimed spectrum. Because this is a gap in a central proof step rather than a refuted theorem, the reader's CONDITIONAL verdict is the appropriate one: the diagonalization needs an independent check or a repair (e.g., an argument that every such Y contains a limited-away non-suspicious relation, or a different obstruction definition that is genuinely permanent). I therefore leave the verdict unchanged.","tokens_in":24020,"tokens_out":12293,"duration_ms":125906,"concrete_test":"Test Lemma 6.1 with an explicit diagonalization run where S1[j]=∅ for all j≠i and B_i is a recursive model of T with basis \\bar b_i of size n+2, containing an R-only finite Y with \\bar b_i∪{c}⊆Y and δ_{\\hat L}(Y)=|\\bar b_i|. Simulate Step 1: after R_i(\\bar b c) is observed, Y is an obstruction; when the first new number enters S1[i], R_i becomes suspicious and δ_t(Y)=|Y|−#_R(Y)=|\\bar b_i|, so by the Step 1 removal rule the obstruction is gone and the strategy can enter Step 2. If this run is consistent and B_i is a recursive model of T, the conclusion of Lemma 6.1 fails; if it is inconsistent, say exactly which condition rules it out.","verdict_should_be":"UNCHANGED","load_bearing_attack":"In Lemma 6.1 (Section 6), the infinite outcome argues: if B_i |= T, \\bar b_i is independent, and c ∈ acl(\\bar b_i) as witnessed by Y with δ_{\\hat L}(Y)=|\\bar b_i|, then adding the dynamic relation R_i(\\bar b c) makes Y an obstruction that is never removed, contradicting the infinite outcome. However, under the Step 1 definitions, an obstruction Y is removed as soon as δ_t(Y) ≥ |\\bar b_i|, where δ_t counts only non-suspicious relation symbols. Once a new number is enumerated into S1[i], R_i becomes suspicious and is excluded from L_t. If Y contains no other relation symbol that is limited away but never suspected, then δ_t(Y) = |Y| − #_R(Y) − Σ_{j≠i, R_j ∈ \\hat L} #_{R_j}(Y) = δ_{\\hat L}(Y) = |\\bar b_i|, so Y is removed. The proof never shows that such a relation must occur in Y; R_i(\\bar b c) holding is witnessed by n+6 Ω-extensions that use only R, so Y could be an R-only algebraic set. In that case the strategy proceeds through Step 2 infinitely while c remains algebraic, and a recursive model of dimension n+1 would survive, making SRM(T)∩[n+1,ω) nonempty. This is exactly the fragility identified by the reader.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper aims to construct, for each finite n, a strongly minimal, flat, non-disintegrated, model complete theory in a finite signature whose spectrum of recursive models is exactly [0,n] ∪ {ω}. The construction uses a Hrushovski amalgamation class with an infinite auxiliary language, a dimension-reducing amalgamation lemma, and a family of 'unblockable' extensions that are claimed to be guaranteed to realize the maximal number of extensions over any base. A recursive saturated model and recursive models of dimensions at most n are built in Section 5, and Section 6 presents a diagonalization intended to rule out recursive models of intermediate finite dimension by enumerating a set S1 so that any candidate model with a purported basis of size k > n is either not a model or has an element algebraic over the basis that the strategy makes non-algebraic. The main theorem (Theorem 6.2) asserts that SRM(T) = [0,n] ∪ {ω}.","tokens_in":24260,"tokens_out":21294,"duration_ms":214908,"significance":"If the proof were correct, the paper would add [0,n] ∪ {ω} to the list of known spectra of recursive models of strongly minimal theories, and it would do so in the technically demanding setting of flat, model complete theories in finite signatures. The paper also introduces a useful technical tool: the notion of k-unblockable extensions, which are shown to have maximal realizations over arbitrary bases in the amalgamation class. The construction in Sections 2–5 is detailed and self-contained, and the algebraic-amalgamation lemmas appear carefully argued. However, the central claim of the paper depends on the correctness of the diagonalization in Section 6. That diagonalization has a load-bearing gap: it does not correctly handle algebraic witnesses that appear only after the strategy has already begun enumerating the column S1[i]. As a result, the paper does not establish the claimed exclusion of intermediate dimensions, and the main theorem is not proven.","major_comments":[{"comment":"The infinite-outcome argument assumes that any witness Y to c ∈ acl(\\bar b_i) forms an obstruction that is never removed. This is not established for witnesses that appear after the first entry into Step 1. Since B_i is only a partially enumerated recursive structure, an R-only witness Y with δ_{\\hat L}(Y) = |\\bar b_i| may first become visible only after the strategy has already passed through Step 2 and R_i has been declared suspicious. In that case R_i is excluded from L_t, so δ_t(Y) = |Y| − #_R(Y) = δ_{\\hat L}(Y) = |\\bar b_i|, and the strategy's own removal rule discards the obstruction (or, if the obstruction set is not redefined on later visits, Y is never added to it). Nothing in the construction forces all algebraic witnesses to be present at the first visit to Step 1. Moreover, the fact that R_i(\\bar b c) is witnessed by Ω-extensions using only R shows that a delayed R-only algebraic witness is compatible with the strategy's dynamic relation. Thus the strategy can run through Step 2 infinitely often while B_i is a recursive model of T with basis \\bar b_i and c ∈ acl(\\bar b_i), so the claimed exclusion of SRM(T) ∩ [n+1, ω) does not follow. The proof's 'never removed' claim is correct only for witnesses already present when R_i is not yet suspicious; that special case does not cover the general situation.","section":"Section 6, Step 1 and Lemma 6.1"},{"comment":"The step 'from some stage onward this Y forms an obstruction that is never removed' conflates the time at which Y exists with the time at which the set of obstructions is defined. If the obstruction set is defined only on the first entry into Step 1, a later-appearing Y is never considered. If the obstruction set is redefined on each entry, then by that later time R_i is already suspicious, and for an R-only Y we have δ_t(Y) = |\\bar b_i|, so Y is immediately removed. Either reading breaks the contradiction with the infinite outcome. This is not a presentation issue but a load-bearing gap in the diagonalization that is essential to Theorem 6.2.","section":"Lemma 6.1"}],"minor_comments":[{"comment":"The symbol δ is used for the initial obstruction condition without an explicit definition of which relations are counted at that moment, while δ_t is defined separately with a subscript; the relation between the two should be clarified.","section":"Section 6, Step 1"}],"recommendation":"reject","confidential_remarks":"The unblockable-extensions technique in Section 3 appears to be a sound and potentially reusable contribution, and it may be worth publishing separately. However, the main spectrum theorem is not established because the Section 6 diagonalization does not rule out recursive models of dimension in [n+1, ω). The flaw is not a minor gap: the strategy's core invariant fails for algebraic witnesses that appear after the first Step 2. If the authors can repair the diagonalization, the result would be significant, but the current manuscript does not support the main claim."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"The new spectrum [0,n]∪{ω} is real news, and the unblockable-extension technique is the best part of the paper. Those extensions are a clean solution to a known obstruction in Hrushovski-plus-recursion constructions, and the lower-bound construction (Section 5) is detailed and credible. If the paper did only that, it would still be worth reading.\n\nThe problem is the other half. Lemma 6.1 is supposed to kill all recursive models of intermediate dimension, and its infinite-outcome argument rests on the claim that an algebraic witness Y for c ∈ acl(b_i) becomes an obstruction that is never removed. Under the paper's own definitions, that claim fails. Once a new number enters S[i]_1, R_i is called suspicious, so L_t excludes R_i. If Y witnesses algebraicity in the true theory, then δ_{\\hat L}(Y) = |b_i|, and because all relations in Y are either R, R_i, or genuine \\hat L-relations, δ_t(Y) = δ_{\\hat L}(Y) = |b_i| as soon as R_i is excluded. The removal condition is δ_t(Y) ≥ |b_i|, so Y is removed immediately. The proof never shows that some non-suspicious relation must occur in Y, and the Ω-extensions are R-only, so the stress-test's scenario — an R-only algebraic set plus the dynamic R_i — is exactly what the construction allows. The infinite outcome then carries no contradiction: c stays algebraic, and a recursive model of dimension n+1 can survive.\n\nThe reader's conditional acceptance was generous. I think this is not a minor gap in presentation; the diagonalization's key mechanism is doing the opposite of what the proof claims. A fix might exist — for example, redefining removal so that a newly suspicious relation still counts against the obstruction for some window, or using a different diagonalization — but it is not in the text. The lower bound and the unblockable extensions stand on their own; the claimed equality SRM(T) = [0,n]∪{ω} is not established.\n\nWho should read it: people working on recursive spectra of strongly minimal theories will want the unblockable extension tool, and the recursive construction of the positive spectrum is a solid model for future work. But citing the main theorem now would be premature.\n\nRecommendation: send to peer review, but with a clear instruction that Lemma 6.1 is the make-or-break point. The referee should demand either a corrected diagonalization or a formalized argument for the never-removed claim. As written, it is not ready for acceptance.","headline":"Genuinely new spectrum and a genuinely useful tool, but the proof of the upper bound in Lemma 6.1 has a load-bearing gap that looks real and needs fixing.","tokens_in":24876,"tokens_out":7824,"would_cite":false,"duration_ms":74500,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["03C57","03D47","03C30"],"pacs":[],"model":"deepseek-v4-flash","headline":"A strongly minimal, flat, non-disintegrated, model complete theory in a finite signature has spectrum of recursive models exactly $[0,n]\\cup\\{\\omega\\}$.","keywords":["spectrum of recursive models","spectrum of computable models","flatness","Hrushovski construction","strongly minimal theory","model completeness","finite signature","non-disintegrated geometry"],"falsifier":"Run the diagonalization of Section 6 for a fixed $n$ and produce a computable atomic diagram of a model of $T$ with an independent tuple of size $n+1$ whose algebraic closure is not the whole model; such a recursive model of dimension in $[n+1,\\omega)$ would immediately falsify the equality $\\mathrm{SRM}(T)=[0,n]\\cup\\{\\omega\\}$. More narrowly, exhibit a stage in the infinite outcome of Lemma 6.1 at which an obstruction $Y$ with $\\delta(Y)=|\\bar b_i|$ is removed after the relation $R_i$ has been added, since that would break the key claim that $c \\notin \\mathrm{acl}(\\bar b_i)$.","tokens_in":23749,"feed_emoji":"♾️","tokens_out":12276,"duration_ms":98918,"temperature":0.7,"pith_summary":"This paper proves that for every natural number $n$ there is a strongly minimal theory—flat, non-disintegrated, model complete, and expressed in a finite signature—whose recursively presentable models are exactly the models of dimensions $0$ through $n$ together with the saturated model of dimension $\\omega$. This adds the set $[0,n]\\cup\\{\\omega\\}$ to the known spectra of recursive models of strongly minimal theories. The result matters because the three prototypical geometric cases (disintegrated, modular, and field-like theories) admit only a few spectra, while this flat theory exhibits a finer mixture: some positive-dimensional models are recursive and others are not. The construction combines an amalgamation construction with a coding of an r.e. set into the theory, and introduces 'unblockable' extensions to guarantee that certain algebraic extensions occur as often as the amalgamation class permits.","feed_headline":"One theory's computable models are exactly dimensions 0..n and ω","feed_subtitle":"The construction yields a flat, non-disintegrated strongly minimal theory with this new spectrum of recursive models.","key_machinery":"The central object is an amalgamation class $C^\\zeta_\\mu$ built from a predimension $\\delta(A)=|A|-\\#R(A)$, where $\\#R(A)$ counts occurrences of relations up to permutation, and a function $\\mu(A,B,m)$ giving the maximum number of disjoint minimally simply algebraic extensions of the form $B$ over $A$ allowed over a base $A$. The argument is carried by a new supply of '3-unblockable' extensions, namely generalized ternary paths in which every added vertex lies on at most three edges; such an extension is guaranteed to realize its $\\mu$-maximal number of copies over any base whatsoever, exactly the property needed to make the coding robust. A second ingredient is a dimension-dropping extension that lets the construction remove an unwanted relation from a tuple while keeping the predimension of every subset fixed, which is what allows the recursive models of dimension at most $n$ to delete the relations $R_i$ that do not belong to the true language.","core_discovery":"The central claim, proved as Theorem 6.2, is that for any $n \\in \\omega$ the theory $T$ built in Section 4 is strongly minimal, flat, non-disintegrated, and model complete in the finite signature $\\{R\\}$, and that its Spectrum of Recursive Models is exactly $[0,n]\\cup\\{\\omega\\}$. To get the positive inclusion, the paper constructs a recursive presentation of the saturated model and, uniformly, recursive copies of the models of dimension at most $n$ as algebraic closures of independent tuples. To rule out intermediate dimensions, it enumerates a set $S_1$ so that any purported recursive model with an independent tuple of size larger than $n$ either fails to satisfy $T$ or contains an element outside the algebraic closure of that tuple, hence the tuple is not a basis. The language is first built with infinitely many relations $R_i$, and the final theory is the reduct to the single ternary relation $R$; the function $\\mu$ is chosen so that each $R_i$ is both existentially and universally definable in the reduct, which is what preserves model completeness.","pith_inferences":["One might try to generalize the coding so that arbitrary finite sets replace the initial interval $[0,n]$, or so that the excluded block is not contiguous, by varying how the r.e. set $S_1$ is represented in the $\\mu$ function.","Because $T$ is model complete, the relative computability of its models may be sharper than the general $\\Sigma^0_4$ bound; one could study the Turing degrees that compute each model and ask whether the gap $[n+1,\\omega)$ persists at other degrees.","The unblockable-extension construction suggests a testable local criterion for when a minimally simply algebraic type is guaranteed maximal multiplicity, which could be applied to higher-arity relations or to other amalgamation classes."],"forward_implications":["The set $[0,n]\\cup\\{\\omega\\}$ becomes a known spectrum of a strongly minimal theory, and the first such example that is at once flat, non-disintegrated, model complete, and in a finite signature.","For each $n$, the same construction yields a distinct theory with this spectrum, so the method produces infinitely many new spectra of recursive models.","The unblockable-extension technique provides a reusable tool for combining recursion-theoretic requirements with amalgamation constructions, since it removes the need for fragile special-case arguments about when a tuple has maximal realizations.","The results show that flatness escapes the trichotomy dichotomy: unlike disintegrated, modular, or field-like finite-signature theories, a flat finite-signature theory can have some but not all positive-dimensional models recursive."],"supporting_citations":[{"why":"Establishes that the countable models of an uncountably categorical theory are determined by a single dimension, which defines the object whose spectrum is studied.","marker":"[BL71]"},{"why":"Supplies the amalgamation construction and the flatness of the resulting strongly minimal geometry on which the whole construction relies.","marker":"[Hru93]"},{"why":"Shows that $[0,n]$ is a spectrum of a flat strongly minimal theory in a finite language, providing the coding-by-extension-count technique that this paper adapts.","marker":"[And11a]"},{"why":"Shows that $\\{\\omega\\}$ and $[0,\\omega)$ are spectra of flat strongly minimal theories in finite languages and introduces the removal-of-relations technique that the current construction refines.","marker":"[And11b]"}],"fun_headline_variants":["Spectrum of recursive models: [0,n] ∪ {ω}","Computable models exist only for dims 0..n and ω","Exactly these recursive models: [0,n] and ω","Flat theory's recursive models are precisely dims 0..n and ω"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The proof that no model of intermediate dimension can have a recursive presentation rests on the claim that in the infinite outcome of the diagonalization, a set $Y$ that witnesses $c \\in \\mathrm{acl}(\\bar b_i)$ together with the relation $R_i(\\bar b, c)$ forms an obstruction that is never removed; if that obstruction could in fact be removed, intermediate spectra might survive.","fun_headline_variants_meta":{"raw":{"variants":["Spectrum of recursive models: [0,n] ∪ {ω}","Computable models exist only for dims 0..n and ω","Exactly these recursive models: [0,n] and ω","Flat theory's recursive models are precisely dims 0..n and ω"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000407,"raw_usage":{"total_tokens":2046,"prompt_tokens":810,"completion_tokens":1236,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":426,"completion_tokens_details":{"reasoning_tokens":1159}},"tokens_in":426,"tokens_out":1236,"duration_ms":10479,"temperature":1.0,"reasoning_tokens":1159,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T11:14:07.201859+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Run the diagonalization of Section 6 for a fixed $n$ and produce a computable atomic diagram of a model of $T$ with an independent tuple of size $n+1$ whose algebraic closure is not the whole model; such a recursive model of dimension in $[n+1,\\omega)$ would immediately falsify the equality $\\mathrm{SRM}(T)=[0,n]\\cup\\{\\omega\\}$. More narrowly, exhibit a stage in the infinite outcome of Lemma 6.1 at which an obstruction $Y$ with $\\delta(Y)=|\\bar b_i|$ is removed after the relation $R_i$ has been added, since that would break the key claim that $c \\notin \\mathrm{acl}(\\bar b_i)$.","supporting_citations":[],"review_version":1}