{"id":"dc690d69-d698-4bfe-a7e8-42d1b9b74b93","arxiv_id":"1908.02102","paper_version":4,"verdict":"CONDITIONAL","confidence":"MODERATE","novelty_score":8.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":0,"one_line_summary":"For every m except 7 and 9, the minimal number of generalized m-gonal numbers representing all integers is m-4 (or 3 or 4 for small m), with optimal repeated-sum versions also determined.","lead":"This paper pins down, for almost every polygon shape, the fewest generalized polygonal numbers needed to write every positive integer as a sum. It also gives exact optimal counts when some summands are repeated, closing a gap from Richard Guy's 1994 question.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Lemma 2.2's universality of Q = sum_{i<=j} x_i x_j is asserted, not shown; the 290-theorem check is omitted, so the upper-bound engine is unverified.","rationale":"The reader's weakest assumption identifies exactly the load-bearing premise: Lemma 2.2's assertion about the auxiliary quaternary form Q. All upper bounds in the paper hinge on representing multiples of m-2 in five variables, and the lemma is the only source of that capability. I agree that this is the most fragile point in the argument. The concern is not that Q is obviously defective; on the contrary, small cases such as n=1 through 15 are readily represented, so the lemma may well be true. The issue is that the paper neither proves the class-number/local-representability claim nor supplies the finite 290-theorem verification it mentions. That is a gap in the written proof, not a demonstrated mathematical error. Since the reader's CONDITIONAL verdict already reflects the need for such verification, my stress-test does not change the recommendation. The proposed concrete test is deliberately simple: a finite check over a bounded box for all n up to 290, which would fully settle the 290-theorem route and thereby remove the main obstacle to accepting the proof.","tokens_in":15994,"tokens_out":28650,"duration_ms":278849,"concrete_test":"Run an exhaustive search over x in Z^4 with |x_i| <= ceiling(sqrt(2*290)) and compute Q(x)=sum_{1<=i<=j<=4} x_i x_j. Check that every integer 1 <= n <= 290 occurs; if any is missing, Lemma 2.2's 290-theorem route fails. For a stricter check, print a witness representation for each of the 29 exceptional integers in the Bhargava-Hanke 290-theorem list; if all are present, Q is universal and the upper-bound chain stands.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Lemma 2.2 is the engine for every upper bound in the paper: it is used to represent multiples of m-2 in five variables, and Theorem 1.1 parts (1), (2), and (3) all reduce to it. The lemma asserts that the quaternary form Q(x)=sum_{1<=i<=j<=4} x_i x_j has class number one and represents every integer locally, with an alternative that one may verify the 290-theorem criterion. Neither claim is demonstrated in the manuscript: the class-number and local-representability facts are stated without proof, and the 290-check is not exhibited. Because this is a fixed four-variable form, a failure here would invalidate the five-variable decomposition used throughout, collapsing the claimed optimal values. My own spot checks of small values suggest Q is in fact universal, so the concern is one of missing verification rather than an identified falsehood; nevertheless, the paper leaves a load-bearing, easily checkable computational assertion unverified, and a referee cannot certify the proof as written without either a proof of class number one plus local representability or an explicit 290-theorem verification.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper studies sums of generalized polygonal numbers P_m(x)=((m-2)x^2-(m-4)x)/2 with x in Z, allowing repetitions of summands with prescribed multiplicities. The main result, Theorem 1.1, gives the minimal number of generalized m-gonal numbers needed to represent every positive integer: for m not in {7,9}, this number is m-4 for m>=10, 3 for m in {3,5,6}, and 4 for m in {4,8}; it also gives sharp results for repeated sums for 7<=r<m-3, and for r=2,...,6 under suitable lower bounds on m. These results answer a question of Guy for all m except 7 and 9, which are explicitly left open. The upper bounds are obtained by reducing sums of five generalized m-gonal numbers to a fixed quaternary quadratic form Q via the identity in Lemma 2.2, then combining this with classical results (Siegel-Weil, Lagrange, Gauss, Legendre, Sun) and extensive finite case analysis. A corollary gives a lower bound for the universal-exceptional constant gamma_m.","tokens_in":16234,"tokens_out":13092,"duration_ms":121573,"significance":"If correct, the paper resolves Guy's question in the generalized setting for all m except 7 and 9 and provides the first optimal results for repeated generalized polygonal numbers. The proof strategy is natural: the hyperplane identity in Lemma 2.2 cleanly reduces a five-variable sum of generalized polygonal numbers to a universal quaternary form, and the remaining work is finite case analysis. The paper is clearly organized and relies on standard external results rather than on the authors' own prior work, so there is no circularity in the main argument. However, the manuscript currently leaves a load-bearing computational verification unperformed: Lemma 2.2 asserts without proof that Q has class number one and represents every integer locally, and the alternative appeal to the Bhargava-Hanke 290-theorem omits the required check. Consequently, the correctness of all upper bounds in the paper depends on an unverified auxiliary claim. No machine-checked proofs or reproducible code are provided.","major_comments":[{"comment":"Lemma 2.2 is the engine for every upper bound in the paper: the hyperplane identity reduces sum_{j=1}^5 P_m(x_j) to (m-2)Q(x') for the fixed quaternary form Q(x')=sum_{1<=i<=j<=4} x'_i x'_j, and the lemma asserts that Q has class number one and represents every integer locally. Neither assertion is proved or supported by a reference; the alternative use of the Bhargava-Hanke 290-theorem is not accompanied by the required verification that Q represents every integer up to 290. Since the proof of Theorem 1.1(1) (Section 3.1), Proposition 3.3, and the proofs in Section 4 all use Lemma 2.2 to represent multiples of m-2 in five variables, the proof as written is incomplete. I ask the authors to provide either a proof of class number one and local representability for Q or an explicit 290-theorem verification (for example, a table of representations of 1,...,290), and to specify which version of the 290-theorem is being applied, because Q is integer-valued but its Gram matrix is not classically integral.","section":"Section 2, Lemma 2.2"},{"comment":"In the r=4,5,6 part of the proof, the sets S_r are defined and then asserted to be precisely the integers less than r(m-2) represented by sum_{j=1}^{r-1} P_m(x_j); this characterization underpins the decomposition n = s + r(m-2)k_1 + rk_2 used with Tables 4.3-4.5. No proof of this characterization is given. Similarly, Table 4.2 is said to follow from Guy's argument without exhibiting the argument for each entry, and the r=3 case ends with 'there remain finitely many choices of j for each k_3 in K_3 and we check these as in the r=2 case' without specifying the choices or presenting the check. Because these are finite verifications from the explicit list (3.1), they should be made explicit (or provided in an appendix or supplementary material) so that the proof can be certified.","section":"Section 4, proof of Theorem 1.1(3)"}],"minor_comments":[{"comment":"The phrase 'this may be done by hand' for the finite cases 10<=m<k+9<=15 and 0<=k1<20 is not a proof; please supply the cases or a systematic table, since the accompanying remark only outlines the method.","section":"Section 3.1, proof of Theorem 1.1(1)"},{"comment":"The proof of Corollary 1.2 is a sketch: 'one can see that every smaller integer is represented' is an assertion without demonstration. This corollary is not needed for Theorem 1.1, but if kept it needs a complete finite verification.","section":"Section 4, proof of Corollary 1.2"},{"comment":"The graph encoding of representations in Lemma 3.1 is very difficult to parse; a table of representations or a recursive algorithm would be much clearer and would make the proof easier to verify.","section":"Lemma 3.1"},{"comment":"In equation (3.5), the expression 'ell >= m-3/r + r-2' should be written with parentheses as 'ell >= (m-3)/r + r-2' and, since ell is an integer, the stronger statement 'ell >= ceil((m-3)/r) + r-2' should be used.","section":"Proposition 3.4"}],"recommendation":"major_revision","confidential_remarks":"The paper is potentially important and the overall strategy is sound, but the proof as written has a clear gap in Lemma 2.2, which is load-bearing for all upper bounds. The missing verification is a fixed, finite computation and should be straightforward to supply; I therefore recommend major revision rather than rejection. The manuscript also cites an unpublished paper [1] and a paper 'in preparation' [11], but these are not used in the proof of the main theorem, so they do not affect correctness."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Short version: this paper settles Guy's optimality question for generalized m-gonal numbers for every m except 7 and 9, and also gives optimal counts for repeated summands in several regimes. That is a real advance, and the proof structure is sound. The one thing a referee should insist on is a verification of Lemma 2.2, the engine behind every upper bound.\n\nWhat's actually new: Theorem 1.1(1) pins down the minimal number l_m for all m not in {7,9}: m-4 for m>=10, and 3 or 4 for the small cases. Previously the literature had Guy's gap m-4 <= l_m <= m plus a few isolated values. Parts (2) and (3) are also new, giving exact values for repeats with r>=7 and for r=2,...,6 with explicit lower bounds on m. The corollary on gamma_m is a nice side benefit. The paper is open about the remaining m=7,9 cases and even about the fact that its method has a hard limit around m>=27 for r=4 etc. That is good practice.\n\nSoft spots, in proportion. The biggest is the one flagged in the stress test: Lemma 2.2 asserts that the quaternary form Q = sum_{i<=j} x_i x_j has class number one and represents every integer locally, and says one could instead verify the 290-theorem. Neither is shown. This is a fixed, small form, and both claims are very likely true—I checked small values and they behave—but as written, the upper bounds for every part of Theorem 1.1 rest on an unproved assertion. This is fixable: produce the class-number computation or a table of the 29 critical values. It is not a fatal flaw, but it is a genuine gap.\n\nA second, smaller issue: the abstract says 'the minimal number of generalized m-gonal numbers' without mentioning the m=7,9 exception. The introduction is correct, but the abstract oversells slightly.\n\nThe casework is extensive and some of it is 'by hand' with no code. The ranges are small enough that I don't see this as a problem, though a referee may want to see more systematic tables for the r=4,5,6 checks.\n\nWho this is for: anyone working on universal quadratic polynomials or the arithmetic of polygonal numbers. The paper deserves a serious referee. I would send it to review and expect a revision that fills the Lemma 2.2 verification.","headline":"Solid resolution of Guy's polygonal-number optimality question for all m except 7 and 9, with a fixable gap in the key lemma.","tokens_in":16821,"tokens_out":4051,"would_cite":true,"duration_ms":38988,"reading_group":"yes","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":["11E12","11E25","11E08"],"pacs":[],"model":"deepseek-v4-flash","headline":"The polygonal-number theorem is exact: m-4 summands suffice for m ≥ 10.","keywords":["generalized polygonal numbers","m-gonal numbers","universal quadratic polynomials","Diophantine equations","sums with repeats","quadratic forms","class number one","polygonal number theorem"],"falsifier":"Compute the class number of $Q(x)=\\sum_{1\\le i\\le j\\le4}x_i x_j$ and check whether $Q$ represents every integer from $1$ through $290$; if the class number is not one or any of those integers is missed, the asserted universality is false and the proof's engine fails. Separately, an exhaustive search for $m=10$ for any positive integer not representable as a sum of six generalized decagonal numbers would refute the $m\\ge10$ clause of the theorem.","tokens_in":15832,"feed_emoji":"🔢","tokens_out":16380,"duration_ms":148053,"temperature":0.7,"pith_summary":"Every positive integer is a sum of at most $m$ ordinary $m$-gonal numbers by the classical polygonal-number theorem, but allowing the summands to be generalized (index $x\\in\\mathbb Z$ instead of $x\\ge 0$) can reduce the count. This paper determines the exact minimum number of generalized $m$-gonal numbers needed to represent every positive integer: for $m\\ge 10$ the minimum is $m-4$, and for $m=3,5,6$ it is $3$ while for $m=4,8$ it is $4$, leaving only $m=7$ and $m=9$ unresolved. It also finds optimal counts when the summands include repetitions: for $7\\le r<m-3$, the minimal length is $\\lceil (m-3)/r\\rceil+(r-2)$, and for $r=2,\\dots,6$ it gives exact formulas under explicit lower bounds on $m$. The result matters because it sharpens a 400-year-old theorem from an upper bound to the true threshold for almost every polygon type.","feed_headline":"Exact count: m-4 polygonal summands suffice for m ≥ 10","feed_subtitle":"Proof also fixes optimal sums with repeated summands; only m=7 and m=9 stay open.","key_machinery":"The engine is a five-variable identity. On the hyperplane $x_1+\\cdots+x_5=0$, the five-term sum collapses to $\\sum_{j=1}^5 P_m(x_j)=(m-2)\\sum_{1\\le i\\le j\\le4}x_i x_j$; therefore every multiple of $m-2$ is a five-term generalized $m$-gonal sum whenever the quaternary form $\\sum_{1\\le i\\le j\\le4}x_i x_j$ is universal. The paper asserts that this quaternary form has class number one and represents every integer locally, so the local-global mass formula makes it universal. The proof then writes $n=(m-2)k_1+k_2$ with $k_2$ in a bounded interval and supplies explicit representations of the residue $k_2$ using a short list of small values of $P_m(x)$; finitely many remaining cases for small $m$ are checked directly. For the repeated-sum results the same five-variable lemma is combined with estimates on how many repeated copies are needed to absorb the residue terms.","core_discovery":"The central discovery is Theorem 1.1, an exact sharpening of the polygonal-number theorem for generalized inputs. With $P_m(x)=\\frac{(m-2)x^2-(m-4)x}{2}$ for $x\\in\\mathbb Z$, the paper proves that the least $\\ell$ for which the equation $\\sum_{j=1}^\\ell P_m(x_j)=n$ is solvable for every $n\\in\\mathbb N$ is $\\ell_m=m-4$ for $m\\ge10$, $\\ell_m=3$ for $m\\in\\{3,5,6\\}$, and $\\ell_m=4$ for $m\\in\\{4,8\\}$; the cases $m=7$ and $m=9$ are the only ones left open. When $r-1$ single copies are followed by $r$ repeated copies of each summand, the paper proves the optimal total length for $7\\le r<m-3$ is $\\lceil(m-3)/r\\rceil+(r-2)$; for $r=2$ and $m\\ge14$ it is $\\lfloor m/2\\rfloor$; for $r=3$ it is $m-2$ for $m\\not\\equiv2\\pmod3$ and $\\frac{2m-4}{3}$ for $m\\equiv2\\pmod3$; and for $r=4,5,6$ it is $\\lceil(m-2)/4\\rceil+2$, $\\lceil(m-3)/5\\rceil+3$, and $\\lceil(m-3)/6\\rceil+4$ under the stated bounds $m\\ge62,78,93$. The proof also yields a corollary on the largest integer that must be checked to decide universality of weighted $m$-gonal sums.","pith_inferences":["The stated lower bounds on $m$ in the small-$r$ cases (e.g. $m\\ge14$ for $r=2$) are likely artifacts of the proof rather than true boundaries; finite computer searches in the omitted ranges could reveal whether the formulas extend.","The method's apparent need for at least six unrepeated variables suggests that for a fixed repeat pattern the ceiling-plus-constant shape should persist for all sufficiently large $m$, with the constant determined by the finitely many residues that must be handled by hand.","The $r=3$ obstruction comes from the congruence $P_m(2)\\equiv P_m(-1)\\pmod3$; analogous congruence coincidences for other $r$ should predict exactly where the optimal formulas gain additive constants."],"forward_implications":["For $m\\ge10$, every positive integer is a sum of at most $m-4$ generalized $m$-gonal numbers, and no smaller number of summands works, so the classical upper bound is improved to the exact threshold.","For $7\\le r<m-3$, the minimal length with $r-1$ single copies followed by $r$ repeated copies is $\\lceil(m-3)/r\\rceil+(r-2)$.","For $r=2,\\dots,6$, the exact values are $\\lfloor m/2\\rfloor$, $m-2$ or $\\frac{2m-4}{3}$, and $\\lceil(m-2)/4\\rceil+2$, $\\lceil(m-3)/5\\rceil+3$, $\\lceil(m-3)/6\\rceil+4$, each valid under a stated lower bound on $m$.","For $m\\ge14$, the largest integer that must be tested to certify universality is at least $3m-12$ when $m\\not\\equiv2\\pmod3$ and at least $2m-9$ when $m\\equiv2\\pmod3$."],"supporting_citations":[{"why":"proves the classical upper bound of at most m m-gonal numbers, which the paper improves to m-4.","marker":"[5]"},{"why":"poses the optimality question and supplies the lower bound from the smallest nonzero generalized m-gonal value; every upper bound must match it.","marker":"[7]"},{"why":"gives a finite verification criterion used as an alternative route to prove universality of the auxiliary quaternary form.","marker":"[3]"},{"why":"part of the local-global mass formula that turns local representability into global universality for class-number-one forms.","marker":"[12]"},{"why":"companion part of the local-global mass formula used to obtain the same universality conclusion.","marker":"[13]"},{"why":"establishes the mass formula on the representation side, completing the local-to-global step behind Lemma 2.2.","marker":"[15]"},{"why":"supplies the known universal representation for m=8 used as one of the exceptional base cases.","marker":"[14]"}],"fun_headline_variants":["Exact m-gonal counts: m-4 for m>=10, 3 for 3,5,6, 4 else","Sharpened Fermat: minimal m-gonal summands nailed except m=7,9","m-4 generalized m-gonal numbers represent all n for m>=10","Repeated generalized polygonal sums: optimal lengths for r=2..6","Fermat's theorem generalized: exact counts for all m except 7,9"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is the unproved assertion that the auxiliary quaternary form $\\sum_{1\\le i\\le j\\le4}x_i x_j$ is universal (class number one plus local representability of every integer); if that assertion fails, the five-variable lemma and all upper bounds built on it collapse.","fun_headline_variants_meta":{"raw":{"variants":["Exact m-gonal counts: m-4 for m>=10, 3 for 3,5,6, 4 else","Sharpened Fermat: minimal m-gonal summands nailed except m=7,9","m-4 generalized m-gonal numbers represent all n for m>=10","Repeated generalized polygonal sums: optimal lengths for r=2..6","Fermat's theorem generalized: exact counts for all m except 7,9"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.00089,"raw_usage":{"total_tokens":3864,"prompt_tokens":992,"completion_tokens":2872,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":608,"completion_tokens_details":{"reasoning_tokens":2753}},"tokens_in":608,"tokens_out":2872,"duration_ms":21577,"temperature":1.0,"reasoning_tokens":2753,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T14:54:46.455578+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Compute the class number of $Q(x)=\\sum_{1\\le i\\le j\\le4}x_i x_j$ and check whether $Q$ represents every integer from $1$ through $290$; if the class number is not one or any of those integers is missed, the asserted universality is false and the proof's engine fails. Separately, an exhaustive search for $m=10$ for any positive integer not representable as a sum of six generalized decagonal numbers would refute the $m\\ge10$ clause of the theorem.","supporting_citations":[{"cited_title":"Cauchy, D´ emonstration du th´ eor` em g´ en´ eral de Fermat sur les nombres polygones, M´ em","cited_arxiv_id":null,"evidence_quote":"proves the classical upper bound of at most m m-gonal numbers, which the paper improves to m-4."},{"cited_title":null,"cited_arxiv_id":null,"evidence_quote":"poses the optimality question and supplies the lower bound from the smallest nonzero generalized m-gonal value; every upper bound must match it."},{"cited_title":"Bhargava and J","cited_arxiv_id":null,"evidence_quote":"gives a finite verification criterion used as an alternative route to prove universality of the auxiliary quaternary form."},{"cited_title":"Siegel, Indefinite quadratische Formen und Funktionentheorie, I , Math","cited_arxiv_id":null,"evidence_quote":"part of the local-global mass formula that turns local representability into global universality for class-number-one forms."},{"cited_title":"Siegel, Indefinite quadratische Formen und Funktionentheorie, II , Math","cited_arxiv_id":null,"evidence_quote":"companion part of the local-global mass formula used to obtain the same universality conclusion."},{"cited_title":"Weil, Sur la formule de Siegel dans la th´ eorie des groupes classiq ues, Acta","cited_arxiv_id":null,"evidence_quote":"establishes the mass formula on the representation side, completing the local-to-global step behind Lemma 2.2."},{"cited_title":"Sun, A result similar to Lagrange’s Theorem , J","cited_arxiv_id":null,"evidence_quote":"supplies the known universal representation for m=8 used as one of the exceptional base cases."}],"review_version":1}