{"id":"47d9db27-32d7-4d44-a4d4-c80df2b00964","arxiv_id":"1908.08349","paper_version":1,"verdict":"ACCEPT","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"low","formal_verification":"none","parameter_count":0,"one_line_summary":"A pairwise mapping is combinatorially similar to an ultrametric exactly when its fibers are coherent, its derived value order is antisymmetric, all triangles are isosceles, and its value set can be embedded like a subset of the real line.","lead":"This paper finds exact rules for deciding when a table of pairwise distance labels is secretly an ultrametric, a tree-like distance structure, after relabeling the points and the distance values. It extends the rules to distance tables whose values live in abstract ordered sets instead of ordinary numbers.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"No significant objection identified: the order-embeddability condition in Theorem 4.21 is load-bearing but correctly established in both directions.","rationale":"Reader's ACCEPT is justified. I focused on Theorem 4.21(iii) because the order-embeddability condition is the only hypothesis not purely fiber-theoretic; it is also the one Example 4.9 shows to be essential. I checked that the proof of (i)->(ii) does not secretly require rho(Y^2)=R+ or that the canonical order be total: it only needs the inclusion <=_rho subseteq <=, which follows from Lemma 3.22 because rho is a <=_rho-pseudoultrametric and its range is the full poset Q=rho(Y^2). The step via Theorem 4.4 is legitimate because canonical orders are generated by u and f preserves u in both directions. The sufficiency direction is a direct application of Proposition 3.24 after shifting the embedding to send a0 to 0. The a0-coherence notion is formally terse (strong consistency is defined for equivalence relations), but Proposition 2.5 fixes the intended meaning, and the subsequent proofs use that meaning. The Continuum Hypothesis is confined to Theorem 4.15 and flagged. Omitted proofs and the (4.22) typo are routine and do not affect the central equivalence. Therefore I recommend keeping ACCEPT.","tokens_in":29698,"tokens_out":35027,"duration_ms":330939,"concrete_test":"As a check, independently prove the containment used in the necessity direction: for a real pseudoultrametric rho, every pair <y1,y2> in u_rho satisfies y1 <= y2, hence <=_rho = u_t^rho union Delta_{range} subseteq <=; then verify that the bijection f obtained from Theorem 4.4 pulls <= back to a total order on Phi(X^2) with <=_Phi subseteq <= and order-isomorphic to a subset of R+. If either containment fails for some rho, the characterization would overclaim.","verdict_should_be":"UNCHANGED","load_bearing_attack":"After tracing the theorem's logical spine, I find no load-bearing flaw. The decisive premise is indeed the order-embeddability of the canonical value order into (R+, <=). Necessity: if Phi is similar to a real pseudoultrametric rho, Lemma 3.22 and Proposition 3.21 give <=_rho = u_t^rho union Delta_{rho(Y^2)} subseteq <=, and Theorem 4.4 upgrades the combinatorial similarity to a weak similarity, so the pullback of <= under the order-isomorphism f is a linear order on Phi(X^2) that contains <=_Phi and embeds into R+. Sufficiency: if such a linear order exists, its embedding, shifted to send a0 to 0, is an injective isotone map from (Phi(X^2), <=_Phi) into R+; Proposition 3.24 makes the composition a real pseudoultrametric, and injectivity gives the required combinatorial similarity. Example 4.9 correctly shows the embeddability hypothesis is not redundant. The only issues I found are expository: Proposition 2.4's proof is omitted, Theorem 4.20 is proved by analogy, and formula (4.22) has an undefined qx_2 typo. None of these threatens Theorem 4.21 or Corollary 4.22.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies the notion of combinatorial similarity between arbitrary mappings on X^2 and (pseudo)ultrametrics, asking when a mapping Φ with domain X^2 is just a value-insensitive relabeling of a real-valued ultrametric or pseudoultrametric. The main results are Theorem 3.10, which handles the countable-range case; Theorem 3.18, which characterizes combinatorial similarity to a Q-pseudoultrametric for a poset Q; and Theorems 4.21 and Corollary 4.22, which give the general real-valued characterization in terms of symmetry, a0-coherence, the isosceles-triangle property, antisymmetry of the transitive closure of the canonical relation u_Phi, and the existence of a linear order on the value set that extends u_Phi and embeds into R+. The paper also studies weak similarities between such mappings, with applications to Priess-Crampe–Ribenboim ultrametric distances, and provides a sharpness example (Example 4.9) showing that the order-embeddability condition is not redundant.","tokens_in":29857,"tokens_out":38214,"duration_ms":313624,"significance":"If correct, this is a complete and checkable combinatorial characterization of when a mapping is secretly an ultrametric or pseudoultrametric. The conditions are explicit and the proofs are constructive, including the rigid ultrametric construction of Proposition 4.7 and the countable embedding via Cantor's lemma. The countable-range theorem is clean, and Example 4.9 sharply delineates the boundary of the general case. The paper also connects to the established literature on weak similarities and on Priess-Crampe–Ribenboim ultrametric distances, and it makes a believable conjecture (Conjecture 4.24) that frames the remaining order-theoretic question. Overall, the central results appear sound and would be a useful contribution to the theory of ultrametric spaces.","major_comments":[],"minor_comments":[{"comment":"The definition of a0-coherence should explicitly require that Φ^{-1}(a0) is an equivalence relation, because strong consistency was defined only for equivalence relations; otherwise Remark 2.3 is false (a two-point mapping with a0 on the off-diagonal and a different value on the diagonal satisfies implication (2.1) without the fiber being reflexive) and the inference Φ(x,x)=a0 in the proof of Theorem 3.10 is unjustified.","section":"Definition 2.2 and Remark 2.3"},{"comment":"The displayed chain after (3.4) uses ≥ where the strong triangle inequality gives ≤ (since ⟨y_i,y_{i+1}⟩∈uΦ means the first coordinate is the base of an isosceles triangle); the contradiction still works after reversing the signs, but please correct the inequalities.","section":"Theorem 3.10, proof of (ii)⇒(iii)"},{"comment":"The symbol q^x_2 in formula (4.22) is undefined; it should refer to a point such as q^y_0 with y>x (for instance y=x+1) to make the injectivity argument work. Also clarify the direction of f: if g:R0→X is a weak similarity from d to ρ, then f should map d(R0^2) to ρ(X^2), not the reverse.","section":"Example 4.9, formula (4.22)"},{"comment":"The theorem statement does not mention the Continuum Hypothesis, but the proof uses 2^{ℵ0}=ℵ1; please state the assumption explicitly in the theorem or indicate that the result is conditional.","section":"Theorem 4.15"},{"comment":"The proof is omitted with the comment that it is straightforward, but the proposition is used later (for instance in Proposition 4.3); please include a proof or at least a detailed sketch.","section":"Proposition 2.4"},{"comment":"The proof is given only by analogy with Theorem 4.18; please provide the details or clearly state the modifications needed when Lemma 4.19 is used instead of Lemma 4.17.","section":"Theorem 4.20"},{"comment":"The bijection g is written as g:Z→Y, but Y was previously defined as Φ(X^2); it should be g:Z→X.","section":"Theorem 3.10, proof of (ii)⇒(iii)"},{"comment":"There are several small typos and formatting issues (for example, 's imilar' and 'Combina torial' in the abstract, 'Φ be a mappings' in Proposition 2.5, and the repeated use of /greaterorequalslant where ≤ is meant); a careful proofreading pass would resolve these.","section":"Throughout"}],"recommendation":"minor_revision","confidential_remarks":"The paper is essentially correct and the main characterization is valuable. The issues are local and easily fixed; I would be happy to see the revised version. No concerns about novelty or scope."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Two things you should know. First, the paper actually delivers: a mapping Phi on X^2 is a relabeling of a real pseudoultrametric iff it is symmetric, satisfies the a0-coherence condition, every triple is isosceles in the right sense, and the canonical relation u_Phi can be extended to a linear order on the value set that embeds into (R+, <=). The countable-range version (Theorem 3.10) is clean, and Example 4.9 shows the order-embeddability condition is not redundant. Second, I traced the main arguments and the stress-test note is right: there is no load-bearing flaw. The equivalence in Theorem 4.21 goes through in both directions, and the necessity side using weak similarity is particularly neat.\n\nWhat is new: the u_Phi relation and a0-coherence are genuinely new devices, and the characterization theorems (3.10, 3.18, 4.21, 4.22) plus the poset-valued extensions in Section 3 go beyond the author's earlier pseudometric paper [16]. The weak-similarity results in Section 4, especially the sharp boundary between countable and continuum ranges (Example 4.9 and Proposition 4.11), are the real content and give the paper its significance. The author also explicitly flags the use of the Continuum Hypothesis in Theorem 4.15, which is the right call.\n\nSoft spots, in proportion: Proposition 2.4 is asserted without proof; it is routine, but should be supplied or given a reference. Theorem 4.20 is proved \"similarly\" to Theorem 4.18, which is acceptable but a bit lazy for a main result. And formula (4.22) has a typo: qx_2 is undefined, presumably meant to be qx_1 or another point. None of these threatens the main theorems. The citation pattern is fine: self-citations to [16] are to the pseudometric analogue, and the new results are not in the cited literature. No data or code, but none is needed; the results stand on their proofs.\n\nWho this is for: specialists in ultrametric spaces, hierarchical clustering, and Priess-Crampe–Ribenboim distances. It is a serious paper with real content, and a competent referee will find the main claims correct and the presentation mostly careful. I would send it to review, and I expect it to be accepted after minor revisions.","headline":"Dovgoshey's combinatorial characterizations of (pseudo)ultrametrics are real and the main proofs hold up; only minor expository gaps stand between this and a solid accept.","tokens_in":30643,"tokens_out":1917,"would_cite":true,"duration_ms":18688,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["54E35","06A05","06A06"],"pacs":[],"model":"deepseek-v4-flash","headline":"The paper proves that a mapping with domain $X^2$ is combinatorially similar to a (pseudo)ultrametric exactly when its triples are isosceles, its zero-fiber is coherent (or the diagonal), and its canonical value order extends to a linear…","keywords":["ultrametric","pseudoultrametric","combinatorial similarity","generalized ultrametric","poset-valued ultrametric distance","a0-coherence","isosceles triangles","order embedding into the reals"],"falsifier":"To test the characterization, take any symmetric, $a_0$-coherent mapping whose triples are isosceles and whose canonical relation extends to an $\\mathbb{R}_+$-embeddable linear order, and apply the paper's construction $f^*\\circ\\Phi$; if that function ever violates the strong triangle inequality, Theorem 4.21 is false. The sharp test case is the lexicographic ultrametric of Example 4.9: it satisfies every condition except order-embeddability into $\\mathbb{R}_+$, and the proof of its non-realizability reduces to the claim that an embedding would create an injection from $\\mathbb{R}_+$ into $\\mathbb{Q}_+$; any explicit embedding of its value order into $\\mathbb{R}_+$ would refute the paper.","tokens_in":29302,"feed_emoji":"📏","tokens_out":11957,"duration_ms":109687,"temperature":0.7,"pith_summary":"The paper asks when an arbitrary table of values on $X \\times X$ is, up to renaming points and relabeling the values, a genuine (pseudo)ultrametric. It proves that this happens exactly when the table is symmetric; its $a_0$-level set is an equivalence relation and every other level is a union of such classes (the $a_0$-coherence condition); every triple of points is isosceles in the value pattern; and the canonical comparison order on the values extends to a linear order isomorphic to a subposet of $(\\mathbb{R}_+,\\leq)$. For countable value sets the last condition is automatic, because every countable linear order embeds into the nonnegative rationals. The paper also characterizes poset-valued ultrametric distances and gives a lexicographic example showing the real-line order condition is genuinely needed. If the paper is right, the strong triangle inequality of an ultrametric is invisible to combinatorial similarity: only the order type of the value set survives.","feed_headline":"Disguised ultrametrics are pure order, not numbers","feed_subtitle":"A value table is a relabeled ultrametric exactly when its triples are isosceles and its value order fits the real line.","key_machinery":"The engine of the argument is the canonical relation $u_\\Phi$ on the value set $V=\\Phi(X^2)$: put $\\langle y_1,y_2\\rangle\\in u_\\Phi$ when there exist $x_1,x_2,x_3$ with $y_1=\\Phi(x_1,x_3)$ and $y_2=\\Phi(x_1,x_2)=\\Phi(x_2,x_3)$. This relation records, in purely combinatorial form, which value can sit at the base of an isosceles triangle while the larger value sits on its two equal sides. Its transitive closure together with the diagonal forms a partial order $\\preceq_\\Phi$ on the values; when $\\Phi$ is $a_0$-coherent, $a_0$ is its smallest element. The proofs build an actual real-valued pseudoultrametric by extending $\\preceq_\\Phi$ to a linear order and then embedding that order into $(\\mathbb{R}_+,\\leq)$ through standard theorems: every countable linear order embeds into the nonnegative rationals, every partial order extends to a linear order, and a linear order is a subposet of $(\\mathbb{R}_+,\\leq)$ exactly when its order topology is second countable. The sharpness boundary is the lexicographic value set $\\mathbb{R}_+\\times\\{0,1\\}$, whose canonical order cannot be embedded into $\\mathbb{R}_+$.","core_discovery":"On the paper's own terms, the central discovery is Theorem 4.21 and its ultrametric counterpart Corollary 4.22. A mapping $\\Phi$ with domain $X^2$ is combinatorially similar to a pseudoultrametric if and only if: $\\Phi$ is symmetric; there is a value $a_0$ such that $\\Phi^{-1}(a_0)$ is an equivalence relation and $\\Phi$ is $a_0$-coherent; every triple of points admits a permutation with two equal $\\Phi$-values; the canonical relation $u_\\Phi$ is contained in a linear order on $\\Phi(X^2)$ with $a_0$ as its smallest element; and that linear order is order-isomorphic to a subposet of $(\\mathbb{R}_+, \\leq)$. For ultrametrics the coherence condition is replaced by the stricter equality $\\Phi^{-1}(a_0)=\\Delta_X$. For countable value sets these conditions collapse to symmetry, $a_0$-coherence, antisymmetry of the transitive closure of $u_\\Phi$, and the isosceles-triangle condition, by Theorem 3.10. The paper further characterizes when a mapping is combinatorially similar to a poset-valued ultrametric distance, showing that the same local conditions are sufficient and that the real-line condition is precisely what distinguishes real-valued ultrametrics from merely poset-valued ones.","pith_inferences":["If the characterization is right, ultrametric similarity is a purely order-theoretic invariant: two pseudoultrametrics are combinatorially similar exactly when their canonical value posets are order-isomorphic, so numerical distances play no role beyond their ordering.","The continuum example implies a non-localizability result: no finite collection of triple conditions can certify real-ultrametric similarity in general, since every countable subtable is realizable while the full table is not.","This suggests a practical test for hierarchical clusterability of finite dissimilarity data: check symmetry, zero-fiber coherence, the isosceles-triangle condition, and acyclicity of the canonical relation; for finite tables the real-line embeddability condition is automatic, so four-point checks decide whether the data is a monotone relabeling of an ultrametric.","Conjecture 4.24, if true, would turn the real-line condition into an internal criterion on the value poset: cardinality at most continuum and every totally ordered subposet embeddable in $\\mathbb{R}_+$. Testing that conjecture on lexicographic products like $\\mathbb{R}_+\\times\\{0,1\\}$ is the natural next step."],"forward_implications":["Every symmetric, $a_0$-coherent mapping with countable range whose triples are isosceles and whose canonical relation has an antisymmetric transitive closure is a relabeled rational-valued pseudoultrametric; for ultrametrics, the zero fiber must be the diagonal.","A mapping with uncountably many values can satisfy all local ultrametric-pattern conditions and still fail to be a real ultrametric: the lexicographic example $\\mathbb{R}_+\\times\\{0,1\\}$ is combinatorially similar to no real ultrametric, despite every countable restriction being realizable.","For poset-valued ultrametric distances, the same conditions characterize combinatorial similarity to a generalized ultrametric, and the passage to a real ultrametric is governed solely by whether the canonical value order embeds into $\\mathbb{R}_+$.","Whenever combinatorial similarity holds, it can be realized by a weak similarity, meaning that the relabeling of values is an order isomorphism of the canonical value posets; combinatorial and order-theoretic sameness coincide on the class the paper characterizes."],"supporting_citations":[{"why":"Defines combinatorial similarity and supplies the pseudometric characterization that the paper extends to ultrametrics.","marker":"[16]"},{"why":"Introduces the class of poset-valued ultrametric distances that later theorems characterize.","marker":"[57]"},{"why":"Develops the generalized ultrametric spaces used for poset-valued pseudoultrametrics.","marker":"[58]"},{"why":"Continues the development of generalized ultrametric spaces and their order-theoretic framework.","marker":"[59]"},{"why":"Provides the extension of partial orders to linear orders used in the construction of real-valued distances.","marker":"[67]"},{"why":"Supplies the result that countable linear orders embed into the nonnegative rationals, used for the countable range theorem.","marker":"[65]"},{"why":"Characterizes subposets of the real line by second-countability of the order topology, used in the uncountable case.","marker":"[9]"},{"why":"Supplies the earlier notion of generalized ultrametrics that the paper contrasts with the poset-valued distances.","marker":"[56]"}],"fun_headline_variants":["Hidden ultrametrics are just order, not numbers","Isosceles triples and linear orders define disguised ultrametrics","Combinatorial lookalikes of ultrametrics revealed","A value table is an ultrametric if triples are isosceles","From symmetric tables to ultrametrics: an order condition"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing assumption is that the value set, equipped with the canonical comparison order coming from isosceles triangles, can be extended to a linear order that is order-isomorphic to a subset of the nonnegative reals; if that order-embedding fails, the whole characterization collapses, and Example 4.9 shows the failure is possible.","fun_headline_variants_meta":{"raw":{"variants":["Hidden ultrametrics are just order, not numbers","Isosceles triples and linear orders define disguised ultrametrics","Combinatorial lookalikes of ultrametrics revealed","A value table is an ultrametric if triples are isosceles","From symmetric tables to ultrametrics: an order condition"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000257,"raw_usage":{"total_tokens":1592,"prompt_tokens":975,"completion_tokens":617,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":591,"completion_tokens_details":{"reasoning_tokens":533}},"tokens_in":591,"tokens_out":617,"duration_ms":6154,"temperature":1.0,"reasoning_tokens":533,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T11:44:07.301363+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"To test the characterization, take any symmetric, $a_0$-coherent mapping whose triples are isosceles and whose canonical relation extends to an $\\mathbb{R}_+$-embeddable linear order, and apply the paper's construction $f^*\\circ\\Phi$; if that function ever violates the strong triangle inequality, Theorem 4.21 is false. The sharp test case is the lexicographic ultrametric of Example 4.9: it satisfies every condition except order-embeddability into $\\mathbb{R}_+$, and the proof of its non-realizability reduces to the claim that an embedding would create an injection from $\\mathbb{R}_+$ into $\\mathbb{Q}_+$; any explicit embedding of its value order into $\\mathbb{R}_+$ would refute the paper.","supporting_citations":[{"cited_title":"Combinatorial characterization of pseudometrics","cited_arxiv_id":"1906.07411","evidence_quote":"Defines combinatorial similarity and supplies the pseudometric characterization that the paper extends to ultrametrics."},{"cited_title":"Priess-Crampe and P","cited_arxiv_id":null,"evidence_quote":"Introduces the class of poset-valued ultrametric distances that later theorems characterize."},{"cited_title":"Priess-Crampe and P","cited_arxiv_id":null,"evidence_quote":"Develops the generalized ultrametric spaces used for poset-valued pseudoultrametrics."},{"cited_title":"Priess-Crampe and P","cited_arxiv_id":null,"evidence_quote":"Continues the development of generalized ultrametric spaces and their order-theoretic framework."},{"cited_title":"Szpilrajn","cited_arxiv_id":null,"evidence_quote":"Provides the extension of partial orders to linear orders used in the construction of real-valued distances."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Supplies the result that countable linear orders embed into the nonnegative rationals, used for the countable range theorem."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Characterizes subposets of the real line by second-countability of the order topology, used in the uncountable case."},{"cited_title":"Priess-Crampe","cited_arxiv_id":null,"evidence_quote":"Supplies the earlier notion of generalized ultrametrics that the paper contrasts with the poset-valued distances."}],"review_version":1}