{"id":"c3158608-d4ff-48c8-9baf-bde84e5ade22","arxiv_id":"2412.00042","paper_version":1,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"The paper refutes the Karvatskyi-Pratsiovytyi conjecture by exhibiting series a≤c≤b such that the achievement sets of a and b are Cantor-type while c's is a Cantorval.","lead":"This paper gives a concrete counterexample to a 2023 conjecture about when the set of all subsums of a series keeps its shape when the series is sandwiched between two others. It shows the conjecture's 'squeeze theorem' analogy fails and proposes a strengthened version that avoids the counterexample.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The counterexample is arithmetically sound, but the printed Section 2 has two load-bearing typos: the proof of Proposition 1 verifies r^b_n < b_n instead of the required r^b_n < a_n, and the final paragraph calls E(b_n) a Cantorval after Proposition 3 proved it Cantor-type; both must be corrected…","rationale":"The reader identified a self-contradictory sentence and recommended conditional acceptance, which is correct. However, the reader's stated weakest assumption -- reliance on Theorem 1.3(e) and the count |Sigma_2(b_n)| = 14 -- is not the main vulnerability: I independently checked the count (the 14 distinct values are correct) and the strict inequality q^2 = 1/16 < 1/14, so the external theorem is applied properly. The real issue is that the manuscript's own text, as printed, fails to prove condition (1) in Proposition 1 and mislabels the Cantorval as E(b_n) rather than E(c_n). Both are evidently typos, since the required inequality holds numerically and the sequence c_n matches the Guthrie-Nymann example, but a reader should not have to infer corrections to verify the central counterexample. The mathematical construction itself is sound and the counterexample is valid once these typos are fixed.","tokens_in":3386,"tokens_out":30313,"duration_ms":282007,"concrete_test":"Check the two flagged passages in Section 2 against the definitions: (1) replace the displayed r^b_n < b_n by r^b_n < a_n and verify that 7*beta/3 < alpha; (2) replace E(b_n) by E(c_n) in the final sentence and verify that c_{2k-1} = 3/4^k, c_{2k} = 2/4^k is the Guthrie-Nymann sequence of [4]. If both substitutions make the text coherent and the inequalities hold, the counterexample stands; otherwise the central claim fails.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The central claim is that (a_n), (b_n), (c_n) satisfy conditions (1) with E(a) and E(b) Cantor-type while E(c) is a Cantorval. I verified the non-obvious numerical ingredients: |Sigma_2(b_n)| = 14 and q^2 = 1/16 < 1/14, so Theorem 1.3(e) from [1] does apply; and E(c_n) is exactly the Guthrie-Nymann sequence from [4]. The load-bearing problem is in the printed text, not in the mathematics. In Proposition 1, the even case is proved by 'r^b_n < b_n', but condition (1) requires 'r^b_n < a_n'. The intended inequality is true because 7*beta/3 = 1.8667 < 1.95 = alpha, so the proof needs a one-letter correction. More seriously, the concluding paragraph of Section 2 says 'the achievement set E(b_n) is Cantorval', contradicting Proposition 3. It must read E(c_n). Taken literally, the counterexample is self-contradictory; with the correction, it is valid. These are not deep flaws, but they are load-bearing because the counterexample cannot be checked from the manuscript as written.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper claims to refute the Karvatskyi–Pratsiovytyi conjecture by presenting three multigeometric sequences a, b, c that satisfy the conjecture's interleaving and remainder conditions (1), while E(a) and E(b) are Cantor-type sets and E(c) is a Cantorval. The intended counterexample shows that two outer sequences sharing the same achievement-set type do not force the middle sequence to have that type. The paper also proposes an improved conjecture adding the condition lim_{n→infty} b_n/a_n = 1, which it says the counterexample does not refute.","tokens_in":3711,"tokens_out":3848,"duration_ms":32448,"significance":"If the intended corrections are made, the paper provides a simple, explicit counterexample to a published conjecture, with hand-checkable constants and no parameter fitting. The proof relies legitimately on a known classification theorem for self-similar sets and on the standard identification of multigeometric achievement sets with K(Σ;q). The counterexample is narrow but sufficient to falsify the conjecture as stated, and the proposed refined conjecture is a reasonable direction for future work.","major_comments":[{"comment":"The even case is proved by showing r^b_n < b_n, but condition (1) requires r^b_n < a_n. Since a_n ≤ b_n, the displayed inequality does not imply the required one. The intended inequality is true: for n = 2k, r^b_{2k} = 7β/(3·4^k) < α/4^k = a_{2k} because 7β/3 = 1.866… < α = 1.95; the proof needs this one-letter correction.","section":"Section 2, Proposition 1"},{"comment":"The sentence 'the achievement set E(b_n) is Cantorval' directly contradicts Proposition 3, which proved E(b_n) is a Cantor-type set. The intended statement is that E(c_n) is a Cantorval; indeed c is exactly the Guthrie–Nymann sequence. Without this correction the counterexample is internally inconsistent and cannot be checked from the manuscript as written.","section":"Section 2, final paragraph"},{"comment":"The prose says 'we suggest adding a condition lim_{n→∞} b_n/a_n ≠ 1', but the stated improved conjecture in (2) and the preceding explanation require lim_{n→∞} b_n/a_n = 1. The displayed '≠' is a typo that should be corrected to '='.","section":"Section 3, introductory paragraph"}],"minor_comments":[{"comment":"The set Σ_2(b_n) is listed explicitly, but a short explanation of why |Σ_2(b_n)| = 14 would help the reader verify the crucial count on which the application of Theorem 1.3(e) depends.","section":"Section 2, Proposition 3"},{"comment":"The phrase 'In [1, Theorem 1.3(e)] proved that...' is ungrammatical; it should read 'Theorem 1.3(e) in [1] proves that...' or similar.","section":"Throughout"},{"comment":"The definition of a Cantor-type set as 'homeomorphic to the Cantor set, which is the achievement set of the sequence (u_n) with u_n = 2/3^n' could be phrased more clearly by saying that the standard Cantor set is the achievement set of that representative sequence.","section":"Introduction"}],"recommendation":"major_revision","confidential_remarks":"The mathematical content appears sound after correcting the internal typo; however, the erroneous 'E(b_n) is Cantorval' sentence means the paper as submitted does not state a consistent counterexample. I recommend major revision rather than rejection because the fix is local and the intended construction is verifiable."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Quick take: the counterexample is real, but the manuscript as submitted is not. Two typos in Section 2 make the claim unverifiable as written; both are one-letter fixes and the mathematics works after that. I agree with the stress-test assessment.\n\nWhat is actually new: explicit multigeometric sequences a, b, c with hand-picked constants α=1.95, β=0.8 showing the Karvatskyi–Pratsiovytyi conditions (1) hold, while E(a) and E(b) are Cantor-type and E(c) is the Guthrie–Nymann Cantorval. The use of Banakh et al.'s Theorem 1.3(e) is legitimate: |Σ_2(b_n)|=14 and (1/4)^2 < 1/14, so the application is correct. The improved conjecture with the limit condition lim b_n/a_n = 1 is a sensible proposal, though it's only motivated by this one example.\n\nSoft spots: Proposition 1's even case proves r^b_n < b_n instead of the required r^b_n < a_n. The intended inequality happens to be true, since 7β/3 ≈ 1.8667 < 1.95 = α, so it's a one-letter correction. The concluding paragraph of Section 2 states 'E(b_n) is Cantorval' right after Proposition 3 proved it Cantor-type; it must read E(c_n). These typos are load-bearing because a referee cannot check the counterexample from the text as it stands. They are not deep mathematical flaws, but they need to be fixed before the claim is supportable. I'd also like the authors to double-check that the improved conjecture isn't refuted by some other construction; as stated it's a reasonable guess, not a theorem.\n\nWho this is for: the small group working on achievement sets and multigeometric sequences. For that audience the counterexample matters and the improved conjecture gives a concrete direction. The paper deserves a serious referee, but only after the obvious corrections. My recommendation: send it to peer review with a note that the author must correct the typos; if the referee confirms the fixes, accept.","headline":"A real counterexample, currently obscured by two fixable typos that make Section 2 self-contradictory; the math holds up once they are corrected.","tokens_in":4124,"tokens_out":2973,"would_cite":true,"duration_ms":24529,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["40A05","28A80","11K31","11B05"],"pacs":[],"model":"deepseek-v4-flash","headline":"Three explicit multigeometric sequences satisfy the Karvatskyi–Pratsiovytyi conjecture's inequalities yet produce achievement sets of different topological types, refuting the conjecture.","keywords":["achievement set","set of subsums","Cantor-type set","Cantorval","multigeometric sequence","self-similar set","counterexample"],"falsifier":"A reader could directly compute the fourteen values in Σ_2(b_n) to confirm the count and verify that (1/4)^2 < 1/14, then approximate E(c) by finite sums to check that it contains intervals. If E(c) turns out to be Cantor-type rather than a Cantorval, or if either E(a) or E(b) contains an interval, the counterexample would fail.","tokens_in":3226,"feed_emoji":"","tokens_out":5775,"duration_ms":52057,"temperature":0.7,"pith_summary":"The paper sets out to disprove a 2023 conjecture by Karvatskyi and Pratsiovytyi, which claimed that if one positive series is sandwiched termwise between two others—with additional tail inequalities—and the outer two have the same kind of achievement set, then the middle series must have that kind too. The achievement set of a positive absolutely summable sequence is the set of all possible subsums, and it is known to be one of three types: a finite union of intervals, a Cantor-type set, or a Cantorval. The paper constructs three explicit multigeometric sequences a, b, and c that satisfy the conjecture's inequalities for every n, with E(a) and E(b) being Cantor-type sets while E(c) is a Cantorval. That counterexample shows the squeeze-theorem analogy behind the conjecture is invalid, and the paper proposes a strengthened version of the conjecture that this counterexample does not refute.","feed_headline":"Sandwiched series can have different achievement-set types","feed_subtitle":"Explicit multigeometric sequences satisfy the conjecture's inequalities but give a Cantorval where Cantor-type sets were forced.","key_machinery":"The central object is the achievement set $E(u_n)=\\{\\sum_{n=1}^\\infty \\varepsilon_n u_n : \\varepsilon_n\\in\\{0,1\\}\\}$, the set of all subsums of a positive absolutely summable sequence. For multigeometric sequences, the achievement set is a self-similar set $K(\\Sigma;q)=\\{\\sum_{n=0}^\\infty d_n q^n : d_n\\in\\Sigma\\}$, where $\\Sigma$ is the finite digit set of one-period partial sums. The decisive tool is the criterion from Theorem 1.3(e): $K(\\Sigma;q)$ is a Cantor-type set if $q^n<1/|\\Sigma_n|$ for some n. The paper uses this criterion at different depths for $a$ and $b$, while the digit set for $c$ produces overlapping intervals that yield the Cantorval structure.","core_discovery":"The central claim is that the Karvatskyi–Pratsiovytyi conjecture, stated in the paper's introduction, is false. For every k ≥ 1, set $a_{2k-1}=a_{2k}=\\alpha/4^k$ with $\\alpha=1.95$, $b_{2k-1}=4\\beta/4^k$ and $b_{2k}=3\\beta/4^k$ with $\\beta=0.8$, and $c_{2k-1}=3/4^k$, $c_{2k}=2/4^k$. The paper proves that $a_n \\le c_n \\le b_n$ and that the tail inequalities $b_n \\le r^a_n$ and $r^b_n < a_n$ hold for all n. Using the known representation of multigeometric achievement sets as self-similar sets $K(\\Sigma;q)$ with $q=1/4$, it applies Theorem 1.3(e) from Banakh et al. to conclude that $E(a)$ and $E(b)$ are Cantor-type sets, because $|\\Sigma(a)|=3$ and $|\\Sigma_2(b)|=14$ with $(1/4)^2<1/14$. However, $E(c)$ is the Guthrie–Nymann achievement set, a Cantorval. Thus the conclusion of the conjecture—that all three achievement sets have the same topological type—fails, even though the hypotheses hold completely.","pith_inferences":["The counterexample suggests that the ratio condition $\\lim b_n/a_n = 1$ may be essential; one testable extension is to search for additional counterexamples that violate only that ratio condition but satisfy all other inequalities.","The self-similar digit-set viewpoint used here could be turned into a systematic search method: vary the digit sets $\\Sigma(a), \\Sigma(b), \\Sigma(c)$ with $q=1/4$ to find other triples where the theorem criterion applies to two sequences but not the third.","If the improved conjecture is true, it would imply a genuine squeezing result for achievement sets under asymptotic closeness of the sandwiching sequences, which would be a useful classification tool for intermediate series."],"forward_implications":["The squeeze-theorem analogy does not carry over to achievement sets: termwise sandwiching plus tail inequalities does not force equal topological type.","The explicit counterexample provides a concrete boundary case that any corrected version of the conjecture must exclude or handle.","The proposed improved conjecture adds the condition $\\lim_{n\\to\\infty} b_n/a_n = 1$, and the paper shows the counterexample does not refute that weaker statement.","The construction demonstrates that multigeometric sequences with the same ratio $q=1/4$ can realize different achievement-set types despite satisfying identical interleaving and tail inequalities."],"supporting_citations":[{"why":"Supplies Theorem 1.3(e), the criterion used to prove E(a) and E(b) are Cantor-type, and the representation of multigeometric achievement sets as K(Σ;q).","marker":"[1]"},{"why":"Defines the Guthrie–Nymann achievement set, the Cantorval that E(c) is identified with.","marker":"[4]"},{"why":"Establishes the trichotomy of achievement sets into finite unions of intervals, Cantor-type sets, and Cantorvals, which frames the conjecture.","marker":"[6]"},{"why":"States the original Karvatskyi–Pratsiovytyi conjecture whose conditions (1) the counterexample satisfies and whose conclusion it refutes.","marker":"[7]"}],"fun_headline_variants":["Counterexample topples achievement-set conjecture","Sandwiched series produce different set types","Multigeometric sequences defy topological conjecture","Explicit series counterexample breaks conjecture","Cantorval appears where Cantor sets were expected"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The counterexample depends on the external theorem that K(Σ;q) is Cantor-type when q^n < 1/|Σ_n| for some n, and on the exact count |Σ_2(b_n)| = 14; if that theorem is misstated, misapplied, or the count is wrong, the counterexample loses its ground.","fun_headline_variants_meta":{"raw":{"variants":["Counterexample topples achievement-set conjecture","Sandwiched series produce different set types","Multigeometric sequences defy topological conjecture","Explicit series counterexample breaks conjecture","Cantorval appears where Cantor sets were expected"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000697,"raw_usage":{"total_tokens":3148,"prompt_tokens":938,"completion_tokens":2210,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":554,"completion_tokens_details":{"reasoning_tokens":2145}},"tokens_in":554,"tokens_out":2210,"duration_ms":16701,"temperature":1.0,"reasoning_tokens":2145,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-12T14:30:34.506183+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A reader could directly compute the fourteen values in Σ_2(b_n) to confirm the count and verify that (1/4)^2 < 1/14, then approximate E(c) by finite sums to check that it contains intervals. If E(c) turns out to be Cantor-type rather than a Cantorval, or if either E(a) or E(b) contains an interval, the counterexample would fail.","supporting_citations":[{"cited_title":"Banakh, A","cited_arxiv_id":null,"evidence_quote":"Supplies Theorem 1.3(e), the criterion used to prove E(a) and E(b) are Cantor-type, and the representation of multigeometric achievement sets as K(Σ;q)."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Defines the Guthrie–Nymann achievement set, the Cantorval that E(c) is identified with."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"Establishes the trichotomy of achievement sets into finite unions of intervals, Cantor-type sets, and Cantorvals, which frames the conjecture."},{"cited_title":"Pratsiovytyi and D","cited_arxiv_id":null,"evidence_quote":"States the original Karvatskyi–Pratsiovytyi conjecture whose conditions (1) the counterexample satisfies and whose conclusion it refutes."}],"review_version":1}