{"id":"18014e5c-ea3c-4174-8bb9-0a98e1b481a3","arxiv_id":"2411.14629","paper_version":2,"verdict":"ACCEPT","confidence":"MODERATE","novelty_score":7.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"For shifts of finite type, shift equivalence over Z+ (eventual conjugacy) implies flow equivalence.","lead":"This mathematics paper proves that shifts of finite type that are eventually conjugate must also be flow equivalent. The result resolves a long-standing open question in symbolic dynamics and has consequences for the classification of C*-algebras.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Proof rests on unproved external classification Theorem 5.7, whose supporting [6] contains an erratum the paper patches by assertion; if the patch is incomplete, the final step fails.","rationale":"The central argument is a reduction: shift equivalence over Z+ gives a PSE equation (Theorem 4.8), setting t=1 yields an SLP(Z) equivalence of stabilizations (6.3), and positivity on cycle components is verified in Section 6 via the direct argument for the first factor and Proposition 5.8 for the second. I read the algebra in Sections 3–4 and the positivity argument in Section 6 as internally sound; the only displayed slip is the bracketed product M in Section 6, whose factors are written in reverse order, while the surrounding prose uses the correct composition phi2 composed with phi1. That typo is not load-bearing. What is load-bearing is Theorem 5.7, the classification theorem that converts the algebraic equivalence into flow equivalence. The paper neither proves it nor reproduces a proof; it extracts it from [6] (Remark 5.13) and simultaneously reports an error in [6, Lemma 2.6] that must be repaired. The repair is asserted to leave all uses of the lemma valid, but this is a nontrivial external claim. If the corrected classification is invalid—whether because the erratum is incomplete or because the extraction of the finite-index positivity condition is inaccurate—then the proof of Theorem 1.1 fails at its final step, even though every computation leading to that step is correct. This is the same concern the reader identified as the weakest assumption; I agree it is the single most load-bearing risk. It does not, in my judgment, lower the reader's verdict, because the external theorem is by the same author, the erratum is disclosed and explained at the two cited use sites, and no counterexample or known gap in the repair is identified.","tokens_in":15887,"tokens_out":15747,"duration_ms":147893,"concrete_test":"Obtain [6] and independently audit the repair in Remark 5.12: (1) enumerate every occurrence of Lemma 2.6 in [6] and verify that Lemma 5.9 of the present paper supplies the needed statement at each occurrence, especially in Lemma A.3 where a=1 is deduced; (2) verify that the finite-index direct-limit formulation and the positivity-on-cycle-components condition of Theorem 5.7 are exactly equivalent to the 'positive stabilized SL_P(Z) equivalence' condition of [6, Theorem 3.1], with no weakened positivity requirement. If both checks pass, the external assumption is supported; if either fails, Theorem 1.1 is unproved.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The proof's final step is Theorem 5.7, a flow-equivalence classification imported from [6] and not proved here. This is load-bearing: Sections 3–6 only produce an SL_P(Z) equivalence of stabilizations of I-A and I-B that is positive on cycle components, and without Theorem 5.7 that algebraic object is not known to imply flow equivalence. The paper's own Remark 5.12 discloses that [6, Lemma 2.6] contained an error and asserts that replacing it by Lemma 5.9 makes the proofs go through. If that repair is incomplete—for example, if Lemma 2.6(3) is used beyond the two cited places, or if the 'positive on cycle components' condition in Definition 5.6 does not exactly match the positivity notion in [6, Theorem 3.1]—then the central implication has no completed proof. This is not an internal algebraic gap in the PSE/partitioned argument; it is an external correctness risk at the final classification step.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper proves that if A and B are nonnilpotent square matrices with entries in Z+ and are shift equivalent over Z+, then the corresponding edge shifts of finite type are flow equivalent (Theorem 1.1). The proof constructs an explicit polynomial shift equivalence equation (3.2) from a shift equivalence, then a P-partitioned version (4.9) via a unique partitioning result (Proposition 4.3). Setting t = 1 yields an SL_P(Z) equivalence of stabilizations of I - A and I - B (Eq. (6.3)), and the paper verifies the positivity on cycle components required by the flow equivalence classification Theorem 5.7. Theorem 1.2 states the equivalent formulation that eventual conjugacy implies flow equivalence for SFTs, and Section 7 supplies examples showing the implication fails outside the subshift class.","tokens_in":15943,"tokens_out":18574,"duration_ms":156227,"significance":"The result is a significant advance for the classification of shifts of finite type: it proves a long-standing implication that was previously known only in the irreducible case and directly yields that SE over Z+ preserves the algebraic invariants of flow equivalence. The explicit PSE equation is a useful tool that may have further applications, and the paper is careful with the partition structure, especially in Proposition 4.3. The main limitation is that the final classification step uses the author's 2002 theorem [6, Theorem 3.1], whose statement is not proved here and which relies on an erratum in [6] that the paper repairs only by assertion. With a completed verification of this dependence, the paper would be a strong contribution.","major_comments":[{"comment":"The proof of Theorem 1.1 in Section 6 depends on Theorem 5.7 as the final classification step, but Theorem 5.7 is not proved in this paper; it is extracted from [6, Theorem 3.1] by Remark 5.13. Since Remark 5.12 states that [6, Lemma 2.6] contains an error and asserts that replacing it by Lemma 5.9 makes the proofs in [6] go through, the validity of this classification result is not self-contained. Please include a proof of Theorem 5.7, or at least a complete verification that Lemma 5.9 repairs every use of Lemma 2.6 that is needed for [6, Theorem 3.1].","section":"Theorem 5.7 and Remark 5.12"},{"comment":"The matrix M defined to give the induced cokernel isomorphism is printed as (I - R; 0 I)_p (I 0; S 0)_p. This does not match the left-hand side of (6.3), where the multipliers appear as (I 0; S I) followed by (I - R; 0 I), and it conflicts with the subsequent phi_1, phi_2 composition. The correct expression is M = (I 0; S I)_p (I - R; 0 I)_p. The intended argument is clear, but the displayed definition should be corrected.","section":"Section 6, after Eq. (6.3)"},{"comment":"After obtaining the SL_P equivalence (6.3), the paper says that Theorem 5.7 applies provided positivity on cycle components is verified. However, condition (a) of Theorem 5.7, namely that A and B have the same cycle components, is not explicitly established. This follows from the shift equivalence equations: if A{p,p} is essentially cyclic then A^ell{p,p} = R{p,tilde p} S{tilde p,p} is a permutation matrix, forcing B^ell{tilde p,tilde p} = S{tilde p,p} R{p,tilde p} to be a permutation matrix as well; but the proof should state this to make the application of Theorem 5.7 complete.","section":"Section 6, application of Theorem 5.7"}],"minor_comments":[{"comment":"In the proof of claim (3), the sentence 'Similarly, S{tilde p, r} != 0 implies 0 != B^ell{tilde p, tilde r} != 0, hence p <= r' is garbled; the displayed inequality should be a single nonzero condition and the conclusion about p <= r in P_A needs a clearer derivation.","section":"Proposition 4.3(3)"},{"comment":"There are several typos, e.g., 'consideral umprovement' in the acknowledgments and the odd '(T k,k)k' in Section 7; these should be fixed.","section":"Throughout"},{"comment":"Proposition 2.1 is invoked to reduce to M^{sq,delta}(Z+), but its proof is only sketched; this is acceptable as a standard exercise, but a reference to a full proof would improve the paper.","section":"Proposition 2.1"}],"recommendation":"major_revision","confidential_remarks":"The paper is essentially a proof of a folklore theorem using the author's own classification from 2002. The erratum to [6] is a point that the editor may want checked by a second referee familiar with the flow equivalence classification. The reliance on self-citation is not problematic given the central role of [6] and [5] in the argument."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"You should know that this paper closes a long-open gap in the basic equivalences for shifts of finite type: shift equivalence over Z+ implies flow equivalence. We already knew the irreducible case from Franks; the general reducible case is new and the proof is direct.\n\nWhat the paper does well: it constructs an explicit polynomial shift equivalence (PSE) equation from a given shift equivalence (Section 3), develops a partitioned version that respects the poset of irreducible components (Section 4), and then verifies the positivity-on-cycle-components condition needed for the final classification step (Section 6). The algebra in the PSE equation and the partitioned lemma (Prop 4.3) is explicit and checkable. Proposition 5.8 and Lemma 5.9 are proved in detail and are the right tools for the positivity argument. This is a serious, careful piece of work.\n\nThe soft spot is the one the stress-test flags: the final step invokes Theorem 5.7, a flow-equivalence classification taken from the author's 2002 paper [6], which is not re-proved here. And [6] had a known error in Lemma 2.6; the paper patches it (Remark 5.12) by replacing it with Lemma 5.9 and asserts the proofs go through. I do not think this is a load-bearing flaw. The author identifies exactly where Lemma 2.6 was used, the replacement is proved with the needed generality, and Remark 5.13 explains how Theorem 5.7 follows from [6, Theorem 3.1]. The remaining worry is the usual one about depending on a published external theorem; it is not a concrete gap. A referee should still check the two cited uses and the consistency of the positivity definitions, but that is a bounded verification task, not a reason to reject.\n\nOverall: the result is new, the proof is transparent, and I believe the claim is correct. This paper deserves a serious refereeing and will be cited. I would accept it for peer review and expect it to be published, with the erratum discussion kept as is.","headline":"A long-open implication in SFT classification, proved with explicit PSE equations and a partitioned argument; the main caveat is a justified reliance on the author's earlier classification theorem, but the dependence is transparent.","tokens_in":16590,"tokens_out":3276,"would_cite":true,"duration_ms":28670,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["37B10","46L35"],"pacs":[],"model":"deepseek-v4-flash","headline":"Shift equivalence over Z+ implies flow equivalence for shifts of finite type.","keywords":["shift equivalence","flow equivalence","shifts of finite type","eventual conjugacy","PSE equation","partitioned matrices","SL(Z[t]) equivalence","cycle components"],"falsifier":"Exhibit one pair of square nonnilpotent matrices over $\\mathbb{Z}_+$ that are shift equivalent over $\\mathbb{Z}_+$ but whose edge shifts of finite type are not flow equivalent; no such pair is known, and finding one would refute Theorem 1.1. A direct place to look is the family of reducible shifts of finite type known to be eventually conjugate but not conjugate: computing their flow equivalence invariants (the K-web, equivalently the stabilized $\\mathrm{SL}_P(\\mathbb{Z})$ class with positivity on cycle components) would decide the theorem for those examples.","tokens_in":15534,"feed_emoji":"🔄","tokens_out":12295,"duration_ms":105116,"temperature":0.7,"pith_summary":"This paper proves that two fundamental equivalence relations on shifts of finite type are nested: if square nonnilpotent matrices $A$ and $B$ with entries in $\\mathbb{Z}_+$ are shift equivalent over $\\mathbb{Z}_+$, then the edge shifts of finite type $\\sigma_A$ and $\\sigma_B$ are flow equivalent. In dynamical language, eventual conjugacy implies flow equivalence for shifts of finite type. The proof is constructive: it turns a shift equivalence into an explicit polynomial matrix equation (the PSE equation), evaluates the equation at $t=1$, and shows the resulting stabilized $\\mathrm{SL}(\\mathbb{Z})$ equivalence is positive on cycle components, exactly what the flow equivalence classification demands. A closing section shows the implication fails for general topological systems, which can be eventually conjugate without being flow equivalent.","feed_headline":"Shift equivalence implies flow equivalence for shifts of finite type","feed_subtitle":"Eventual conjugacy now implies flow equivalence for all SFTs, including reducible systems.","key_machinery":"The engine is the PSE (polynomial shift equivalence) equation. From a lag-$\\ell$ shift equivalence $A^\\ell=RS$, $B^\\ell=SR$, $AR=RB$, $BS=SA$ over a ring, the paper writes an explicit $2\\times 2$ block matrix identity over $\\mathbb{Z}[t]$ showing that stabilizations of $I-tA$ and $I-tB$ are $\\mathrm{SL}(\\mathbb{Z}[t])$-equivalent. The proof first reduces $A$ and $B$ to block upper triangular form with irreducible diagonal blocks, repartitions the shift equivalence according to the component poset (Proposition 4.3), and applies the $P$-partitioned version of the PSE equation. Setting $t=1$ produces a stabilized $\\mathrm{SL}_P(\\mathbb{Z})$-equivalence of $I-A$ and $I-B$; a nonnegativity argument (Proposition 5.8) shows this equivalence is positive on cycle components, which is precisely the hypothesis of the flow equivalence classification theorem (Theorem 5.7) that yields the conclusion.","core_discovery":"The central claim is Theorem 1.1: for square nonnilpotent matrices $A$ and $B$ over $\\mathbb{Z}_+$, shift equivalence over $\\mathbb{Z}_+$ implies that the edge shifts of finite type $\\sigma_A$ and $\\sigma_B$ are flow equivalent. Since every shift of finite type is conjugate to an edge shift, and shift equivalence over $\\mathbb{Z}_+$ characterizes eventual conjugacy, the equivalent Theorem 1.2 states that eventual conjugacy implies flow equivalence for shifts of finite type. The result was already known for irreducible shifts of finite type through the earlier algebraic classification, and it was remarked without proof in a 2000 survey; this paper supplies the full proof for reducible systems. It also records a consequence for C*-algebras: any equivariant stable isomorphism of not necessarily simple Cuntz–Krieger algebras (a class of C*-algebras built from finite 0-1 matrices) can be replaced by a diagonal-preserving one, as follows from an existing diagram of implications.","pith_inferences":["The explicit PSE equation gives future invariants a canonical test object: to check whether a proposed invariant separates two flow-equivalent SFTs, one can plug in the stabilized $\\mathrm{SL}_P(\\mathbb{Z})$ equivalence this construction produces, rather than hunting for one.","Because the proof's last step leans on a classification theorem quoted from an earlier paper and patched with a corrected lemma, a careful reader will want to re-verify the two places that earlier lemma was used; if the patch is incomplete, the bridge has a hole.","The same construction might adapt to other semirings (for example, replacing $\\mathbb{Z}_+$ by $\\mathbb{R}_+$), since the positivity argument only uses nonnegativity of the factors; the paper does not explore this.","Question 6.6—whether eventually conjugate subshifts can fail to be flow equivalent—now looks sharper: if such examples exist, they can only be non-SFT subshifts, since the SFT case is settled by this theorem."],"forward_implications":["Theorem 1.2: eventual conjugacy implies flow equivalence for all shifts of finite type, not just irreducible ones.","The proof is an explicit algorithm: from the matrices $R$ and $S$ witnessing a shift equivalence, one writes the PSE equation, specializes at $t=1$, and reads off the stabilized $\\mathrm{SL}_P(\\mathbb{Z})$ equivalence.","Via the diagram of implications in the C*-algebra literature, Theorem 1.1 implies that any equivariant stable isomorphism of not necessarily simple Cuntz–Krieger algebras can be replaced by a diagonal-preserving one.","The theorem leaves open whether eventual conjugacy implies flow equivalence for arbitrary subshifts; Section 7 shows the implication is false for general (non-subshift) systems."],"supporting_citations":[{"why":"Supplies the reduction of a general nonnilpotent $\\mathbb{Z}_+$ matrix to block upper triangular form with irreducible diagonal blocks, the starting point of the proof.","marker":"[3]"},{"why":"Provides the flow equivalence classification theorem (Theorem 5.7 here) into which the proof plugs; the paper also corrects an error in that source's Lemma 2.6.","marker":"[6]"},{"why":"Gives the irreducible case classification of flow equivalence, the model for the general classification used in the proof.","marker":"[15]"},{"why":"Establishes that shift equivalence over $\\mathbb{Z}_+$ characterizes eventual conjugacy and that every SFT is conjugate to an edge SFT, used to state Theorem 1.2.","marker":"[20]"}],"fun_headline_variants":["Eventual conjugacy implies flow equivalence for all SFTs","Shift equivalence now implies flow equivalence for reducible SFTs","Proof complete: shift equivalence yields flow equivalence for every SFT","All SFTs: shift equivalence ensures flow equivalence (reducible included)"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The proof depends on a previously published classification theorem (Theorem 5.7) that is quoted, not proved here; if that classification is wrong, the conclusion need not follow.","fun_headline_variants_meta":{"raw":{"variants":["Eventual conjugacy implies flow equivalence for all SFTs","Shift equivalence now implies flow equivalence for reducible SFTs","Proof complete: shift equivalence yields flow equivalence for every SFT","All SFTs: shift equivalence ensures flow equivalence (reducible included)"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000914,"raw_usage":{"total_tokens":3820,"prompt_tokens":735,"completion_tokens":3085,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":351,"completion_tokens_details":{"reasoning_tokens":3011}},"tokens_in":351,"tokens_out":3085,"duration_ms":22557,"temperature":1.0,"reasoning_tokens":3011,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T15:05:56.025759+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Exhibit one pair of square nonnilpotent matrices over $\\mathbb{Z}_+$ that are shift equivalent over $\\mathbb{Z}_+$ but whose edge shifts of finite type are not flow equivalent; no such pair is known, and finding one would refute Theorem 1.1. A direct place to look is the family of reducible shifts of finite type known to be eventually conjugate but not conjugate: computing their flow equivalence invariants (the K-web, equivalently the stabilized $\\mathrm{SL}_P(\\mathbb{Z})$ class with positivity on cycle components) would decide the theorem for those examples.","supporting_citations":[{"cited_title":"Shift equivalence and the Jordan form away from zero.Ergodic Theory Dynam","cited_arxiv_id":null,"evidence_quote":"Supplies the reduction of a general nonnilpotent $\\mathbb{Z}_+$ matrix to block upper triangular form with irreducible diagonal blocks, the starting point of the proof."},{"cited_title":"Flow equivalence of shifts of finite type via positive factorizations.Pacific J","cited_arxiv_id":null,"evidence_quote":"Provides the flow equivalence classification theorem (Theorem 5.7 here) into which the proof plugs; the paper also corrects an error in that source's Lemma 2.6."},{"cited_title":"Flow equivalence of subshifts of finite type.Ergodic Theory Dynam","cited_arxiv_id":null,"evidence_quote":"Gives the irreducible case classification of flow equivalence, the model for the general classification used in the proof."},{"cited_title":"Cambridge University Press, Cambridge, second edition, 2021","cited_arxiv_id":null,"evidence_quote":"Establishes that shift equivalence over $\\mathbb{Z}_+$ characterizes eventual conjugacy and that every SFT is conjugate to an edge SFT, used to state Theorem 1.2."}],"review_version":1}