{"id":"733c5a8e-9a5e-431d-90ec-9a8a3374e5d0","arxiv_id":"2609.16960","paper_version":1,"verdict":"ACCEPT","confidence":"HIGH","novelty_score":6.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"All primes p such that for every odd x with x^2 < p, the number p - x^2 has at most two different prime factors are determined for p ≡ 1,5,7 (mod 8) and for p ≡ 3 (mod 8) except one possible prime.","lead":"A classification of primes p such that p minus odd squares has at most two distinct prime factors is fully solved for three residue classes modulo 8 and nearly solved for the fourth. The result connects to real quadratic fields with class number one.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek/deepseek-v4-flash","headline":"In p≡3 (mod 8) subcase with n prime, deduction that 2m-n=1 lacks justification; factor could be a power of 2, breaking subsequent quadratic form derivation.","rationale":"The reader identified the RH dependence as the weakest assumption, but the more fundamental gap is the unjustified forcing of 2m-n=1 (and n-2m=1) in the p≡3 mod 8, n‑prime subcase. This is a logical omission that could affect the completeness of the classification. Unlike the RH‑excluded possible exception, which is honestly stated, this step appears to be an oversight. The paper should either prove that (2m-n) must be 1 (e.g., by using uniqueness of representation or another property) or treat the case where it is a power of 2. Therefore, the verdict should be CONDITIONAL on the authors providing justification for that deduction or handling the missing case. The rest of the paper appears solid, so acceptance should be conditional on this clarification.","tokens_in":11806,"tokens_out":9496,"duration_ms":80351,"concrete_test":"Check whether any prime p=m²+2n² with n prime, ω(p-x²)≤2 for all odd x²<p, and 2m-n=2^k with k≥2 exists. For small m (say up to 1000) and k=2,3,4, compute p and test the hypothesis computationally. If none are found, provide a theoretical argument showing that such a p cannot be prime or must violate the hypothesis for some x. If a counterexample is found, the classification must be extended to include such p. The easiest first check: test k=2 (2m-n=4) for m up to, say, 500; a single prime satisfying the hypothesis would refute the step.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"In Section 4 (p≡3 mod 8), the subcase p=m²+2n² with n prime splits into 2m>n and 2m<n. The author writes p-(2n-m)²=2n(2m-n) and says “According to the hypothesis, 2m-n should be 1.” (similarly n-2m=1 in the other branch). The hypothesis only guarantees ω(p-x²)≤2, which allows 2m-n to be a power of 2: then the RHS has exactly two distinct prime factors (2 and n), so the hypothesis is satisfied. The paper does not rule out 2m-n=2^k with k≥2. If such a case occurs, the derived form p=9m²±8m+2 would not hold, the continued fraction expansion would differ, and the reduction to the Mollin–Williams list of period‑6 fields would be invalid. This is not a minor oversight—it undermines the entire classification for p≡3 mod 8 in the n‑prime subcase. The reader’s verdict noted only the RH dependence for the possible exception, but the logic here is incomplete even before that step.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek/deepseek-v4-flash","summary":"The paper determines all odd primes p such that, for every odd integer x with x^2 < p, the integer p - x^2 has at most two distinct prime factors (i.e., ω(p-x^2) ≤ 2). Complete finite lists are given for p ≡ 1,5,7 (mod 8), while for p ≡ 3 (mod 8) a list is given with a single possible exception that is excluded under the Riemann Hypothesis. The proofs connect the additive condition to the class number of the real quadratic field Q(√p) and rely on classifications by Biró, Lee, and Mollin–Williams.","tokens_in":12164,"tokens_out":5008,"duration_ms":43615,"significance":"If correct, Theorem 1.1 provides a complete solution to a natural additive problem that generalizes earlier work of the author and his collaborators. The proof combines elementary number theory with deep results about class‑number‑one real quadratic fields, showing how an apparently elementary condition forces the class number to be one and reduces to finitely many cases. The classification for p ≡ 1 (mod 8) is entirely elementary, while the other cases use celebrated theorems (Yokoi’s and Chowla’s conjectures, Mollin–Williams’ table). The paper is self‑contained in structure and gives explicit finite lists ready for further computational or theoretical use.","major_comments":[],"minor_comments":[{"comment":"The deduction that 2m−n must be 1 is correct but under‑justified. Because m and n are odd, 2m−n is odd; the only odd power of 2 is 1, and any odd prime factor would introduce a third prime factor (together with 2 and n) violating ω(p−x^2)≤2. Adding a brief parity remark would prevent the reader from worrying about the possibility 2m−n=2^k (k≥2). The same clarification applies to the adjacent branch 2m<n.","section":"Section 4, subcase n prime, 2m>n (p. 5)"},{"comment":"The formulas √p = (3m−2; 1,2,3m−2,2,1,6m−4) and √p = (3m+1; 2,1,3m,1,2,6m+2) are presented without derivation. The reader would benefit from a short verification that the period length is exactly 6, or from a reference to a standard source for such expansions.","section":"Section 4, continued‑fraction expansions (pp. 6–7)"},{"comment":"The Lemma assumes 16q^2 < p and states that the exponents a,b,c,d are ≥ 3. The inequalities that force these exponent lower bounds rely on the condition 16q^2 < p; a single sentence explaining the arithmetic (e.g., 8q < p/m?) would improve readability.","section":"Section 5, Lemma 5.1 (p. 7)"},{"comment":"The system of congruences modulo 5,7,13,17,109,241,433 is carefully checked, but the exposition is dense. Adding a table or summarizing the logic (why these particular moduli are chosen) would help the reader follow the elimination.","section":"Section 5, analysis of the case p = (2^{2x+1}+2^y−1)^2+2^{y+2} (pp. 9‑10)"},{"comment":"Several parenthesized expressions contain ambiguous spacing (e.g., “(2^{u−2}+1)^2”). Using consistent braces and/or clarifying multiplication signs would improve readability. There are also a few places where a period is missing at the end of a displayed equation.","section":"Throughout"}],"recommendation":"minor_revision","confidential_remarks":"The paper is technically sound. The only point that could raise concern in review is the step in Section 4 where the author writes ‘2m−n should be 1’ without explicit parity reasoning; once that is clarified (as noted in the minor comments), the argument is complete. The results are interesting and the paper is well‑organized. I recommend minor revision, primarily to improve clarity in the points listed above."},"author_rebuttal":null,"desk_editor":{"model":"deepseek/deepseek-v4-flash","letter":"Punchline: this is a solid, narrow number-theory paper with one genuinely new case and three cases that lean on known heavy results. It deserves a serious referee, not a desk reject.\n\nWhat is new: the p≡1 (mod 8) classification is elementary and appears not to be in the literature; the resulting list looks right. The p≡5 and p≡7 cases are mostly repackaging of Biró and Lee plus the author's earlier work, though the reductions from the additive hypothesis to those classifications are done cleanly. The p≡3 case is honestly labelled: it is solved up to one possible exception, with the exception going away under RH.\n\nWhere I would push back: the stress-test note's flagged flaw in Section 4 is not a flaw. In the n-prime subcase, 2m-n is odd, so \"power of 2\" cannot mean anything but 1; and if 2m-n had an odd prime factor r, then p-(2n-m)^2=2n r(...) has the three distinct prime factors 2, n and r unless r=n, while r=n would force n|p and hence is impossible. The same repair works for the other branch. The author's one-line \"should be 1\" is too compressed, but the conclusion is correct.\n\nThe real soft spot is Section 5. Lemma 5.1 is fine in outline, but several claims are asserted without enough detail: the jump from x≥2 to y=z=1, the \"for obvious reasons, d should be 3\", and above all the sentence \"the only possible case is...\" in the analysis of equations (6) and (10). The finite congruential sieve with 45 and 27 cases is described but not shown; the reader is asked to take the residue computations on faith. I did not find a concrete error, but this is exactly the part an editor should send to a referee who is willing to do arithmetic.\n\nBottom line: for someone working on additive properties of primes and real quadratic class numbers, this is a useful reference. I would not cite it unless I needed its exact lists. It deserves peer review; the referee report should ask for a rewritten Section 5 with the skipped steps and the finite checks made explicit.","headline":"Competent, narrow classification paper: the p≡1 case is genuinely new, the p≡3 stress-test worry is a false alarm, but Section 5 needs referee arithmetic.","tokens_in":12858,"tokens_out":24197,"would_cite":false,"duration_ms":171446,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"pith_extraction":{"msc":["11R29","11R11","11A55"],"pacs":[],"model":"deepseek/deepseek-v4-flash","headline":"All primes p such that p minus an odd square has at most two prime factors are classified, with one possible exception.","keywords":["class numbers","continued fractions","real quadratic fields","class number one","additive problem","prime numbers","quadratic residues","Richaud-Degert types"],"falsifier":"Finding a 19th real quadratic field of class number one with continued fraction period length six, or proving that no such field exists, would settle the p ≡ 3 (mod 8) list unconditionally. Alternatively, proving or disproving the Riemann Hypothesis would determine whether the possible exception can be excluded.","tokens_in":13412,"feed_emoji":"","tokens_out":3847,"duration_ms":33947,"temperature":0.7,"pith_summary":"This paper determines all odd primes p with the property that for every odd integer x less than $\\sqrt$(p), the number p − $x^{2}$ has at most two distinct prime factors. The author treats four cases according to p modulo 8. For primes congruent to 1, 5, or 7 modulo 8, the lists are complete and explicit. For primes congruent to 3 modulo 8, all such primes are found with one possible exception; that exception cannot be ruled out without assuming the Riemann Hypothesis. The results connect to the classification of real quadratic fields of class number one and to continued fraction period lengths.","feed_headline":"Primes with a strict additive two-factor property found for all but one case","feed_subtitle":"Using class-number-one fields and continued fractions, the author gives complete lists for three residue classes; one possible exception…","key_machinery":"The proof relies on four separate strategies corresponding to each residue class mod 8. For p ≡ 7 (mod 8) the author shows p = $a^{2}$ − 2 and then uses a theorem of Jungyun Lee on Richaud-Degert types to bound the parameter. For p ≡ 5 (mod 8) the problem reduces to cases p = $m^{2}$ + 4 or p = 1 + $4n^{2}$, and the class number one property of Q(√p) is established via Biró's proofs of Yokoi's conjecture and Chowla's conjecture. For p ≡ 3 (mod 8) the argument splits into subcases; one subcase uses Lee's theorem and the other relies on the Mollin–Williams classification of real quadratic fields of class number one with continued fraction period length six. For p ≡ 1 (mod 8) the method is elementary and uses a lemma that forces p to be a quadratic residue modulo at least one prime in a specific set, leading to an explicit bound p < 16·433², after which a direct search yields all primes.","core_discovery":"Theorem 1.1 gives the complete list of primes p satisfying the condition ω(p − $x^{2}$) ≤ 2 for every odd x with $x^{2}$ < p. For p ≡ 7 (mod 8) the primes are 7, 23, 47, 167. For p ≡ 5 (mod 8) they are 5, 13, 29, 37, 53, 101, 173, 197, 293, 677. For p ≡ 1 (mod 8) they are 17, 41, 73, 89, 97, 113, 137, 233, 257, 353, 593, 857, 1097, 1217, 1553, 2273, 5297. For p ≡ 3 (mod 8) they are 3, 11, 19, 59, 83, 107, 227, with one possible exception that is excluded only under the Riemann Hypothesis.","pith_inferences":["The condition ω(p − x^2) ≤ 2 for all odd x < √p is so restrictive that only finitely many primes satisfy it, and the author has essentially found them all. This is reminiscent of the Rabinowitsch criterion for imaginary quadratic fields and suggests a deeper connection between additive constraints on prime gaps and class number one.","The elementary approach for p ≡ 1 (mod 8) demonstrates that sophisticated analytic machinery is not always necessary; a systematic quadratic residue chase can suffice for complete classification.","If the Riemann Hypothesis is false, the possible exceptional prime for p ≡ 3 (mod 8) might actually exist, so the list in that case would be conditional. Conversely, if the Riemann Hypothesis is true, the list is unconditional and complete.","The results could be used as a benchmark for testing computational number theory packages: verifying that no other prime below any reasonable bound satisfies the condition would confirm the theorem's completeness in practice."],"forward_implications":["Any prime not appearing in the four lists of Theorem 1.1 does not satisfy the ω(p − x^2) ≤ 2 condition for all odd x < √p.","The classification provides a finite and completely known set, except possibly one undetermined prime in the p ≡ 3 (mod 8) case.","The result confirms that the condition forces very strong arithmetic restrictions on p, in particular forcing Q(√p) to have class number one.","The elementary method for p ≡ 1 (mod 8) shows that heavy computational search can be avoided by a clever combination of quadratic residues and bounds.","The one possible exception for p ≡ 3 (mod 8) is linked to the existence of an unknown real quadratic field of class number one with continued fraction period length six."],"supporting_citations":[{"why":"Biró's proof of Yokoi's conjecture is used in Section 3 to determine the primes for the subcase n=1, p=m^2+4.","marker":"[1]"},{"why":"Biró's proof of Chowla's conjecture is used in Section 3 to determine the primes for the subcase m=1, p=1+4n^2.","marker":"[2]"},{"why":"The result from Gica and Luca (Theorem 3) is used in Section 2 to conclude that p=7,23,47,167 are the only possibilities when p ≡ 7 (mod 8).","marker":"[6]"},{"why":"Lee's complete determination of wide Richaud-Degert types with class number one is used in Section 2 and in the first subcase of Section 4 to bound a parameter.","marker":"[8]"},{"why":"Mollin and Williams's classification of real quadratic fields of class number one with continued fraction period length less than 25 is used in Section 4 to restrict possible p in the subcases where the period length is six.","marker":"[10]"}],"fun_headline_variants":["Almost all primes satisfy two-factor difference property","Two-prime-factor difference solved for three residue classes","Complete list of primes with strict additive property, bar one","Prime minus odd square has at most two factors, all but one","Quadratic fields yield complete sets for additive prime property"],"cache_read_input_tokens":0,"weakest_assumption_plain":"The proof for p ≡ 3 (mod 8) depends on the Mollin–Williams theorem which states that at most 19 squarefree d have class number one and continued fraction period length six, with only 18 known; the final exclusion of the possible exception in that case requires the Riemann Hypothesis.","fun_headline_variants_meta":{"raw":{"variants":["Almost all primes satisfy two-factor difference property","Two-prime-factor difference solved for three residue classes","Complete list of primes with strict additive property, bar one","Prime minus odd square has at most two factors, all but one","Quadratic fields yield complete sets for additive prime property"]},"model":"deepseek/deepseek-v4-flash","effort":"low","cost_usd":0.00019,"raw_usage":{"cost":0.000190288,"is_byok":false,"cost_details":{"upstream_inference_cost":0.000190288,"upstream_inference_prompt_cost":8.6408e-05,"upstream_inference_completions_cost":0.00010388},"total_tokens":1193,"prompt_tokens":822,"completion_tokens":371,"prompt_tokens_details":{"audio_tokens":0,"video_tokens":0,"cached_tokens":256,"cache_write_tokens":0},"completion_tokens_details":{"audio_tokens":0,"image_tokens":0,"reasoning_tokens":292}},"tokens_in":822,"tokens_out":371,"cost_source":"openrouter_usage","duration_ms":4505,"image_pages":[],"temperature":1.0,"reasoning_tokens":292,"transport_provider":"openrouter","cache_read_input_tokens":0,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-09-18T16:59:43.906656+00:00","model_set":{"reader":"deepseek/deepseek-v4-flash"},"falsifier":"Finding a 19th real quadratic field of class number one with continued fraction period length six, or proving that no such field exists, would settle the p ≡ 3 (mod 8) list unconditionally. Alternatively, proving or disproving the Riemann Hypothesis would determine whether the possible exception can be excluded.","supporting_citations":[{"cited_title":"Bir´ o,Yokoi’s conjecture, Acta Arith.,106(2003), 85–104","cited_arxiv_id":null,"evidence_quote":"Biró's proof of Yokoi's conjecture is used in Section 3 to determine the primes for the subcase n=1, p=m^2+4."},{"cited_title":"Bir´ o,Chowla’s conjecture, Acta Arith.,107(2003), 179–194","cited_arxiv_id":null,"evidence_quote":"Biró's proof of Chowla's conjecture is used in Section 3 to determine the primes for the subcase m=1, p=1+4n^2."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"The result from Gica and Luca (Theorem 3) is used in Section 2 to conclude that p=7,23,47,167 are the only possibilities when p ≡ 7 (mod 8)."},{"cited_title":"Lee,The complete determination of wide Richaud-Degert types which are not 5 modulo 8 with class number one, Acta Arith.140(2009), 1–29","cited_arxiv_id":null,"evidence_quote":"Lee's complete determination of wide Richaud-Degert types with class number one is used in Section 2 and in the first subcase of Section 4 to bound a parameter."},{"cited_title":"Mollin, H.C","cited_arxiv_id":null,"evidence_quote":"Mollin and Williams's classification of real quadratic fields of class number one with continued fraction period length less than 25 is used in Section 4 to restrict possible p in the subcases where the period length is six."}],"review_version":1}