Every Nonnegative Integer Is a Sum of a Triangular, a Pentagonal, and a Heptagonal Number
Yichuan Cao \institution State Key Laboratory of Mathematical Sciences, Academy of Mathematics and Systems Science, Chinese Academy of Sciences, and University of Chinese Academy of Sciences. This work is supported by the Strategic Priority Research Program of Chinese Academy of Sciences under Grant XDA0480502 and XDA0480503, Dakai Guo \instref 1, Ruichen Qiu \instref 1, Ruyong Feng \instref 1, Xiao-Shan Gao \instref 1 · 2026
Read Every Nonnegative Integer Is a Sum of a Triangular, a Pentagonal, and a Heptagonal Number by Yichuan Cao \institution State Key Laboratory of Mathematical Sciences, Academy of Mathematics and Systems Science, Chinese Academy of Sciences, and University of Chinese Academy of Sciences. This work is supported by the Strategic Priority Research Program of Chinese Academy of Sciences under Grant XDA0480502 and XDA0480503, Dakai Guo \instref 1, Ruichen Qiu \instref 1, Ruyong Feng \instref 1, Xiao-Shan Gao \instref 1 (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2606-26035.txt.
Introduction
Yichuan Cao, Dakai Guo, Ruichen Qiu, Ruyong Feng, and Xiao-Shan Gao settle a number theory conjecture about polygonal numbers. Triangular, pentagonal, and heptagonal numbers count dots in growing polygon patterns. Zhi-Wei Sun had conjectured the three families cover everything together. One value from each family should sum to any nonnegative integer. The claim sat recorded as sequence A287616 in the OEIS database.
Proof now exists, with a modern twist. The natural-language proof and its Lean 4 formalization were both generated by the MechMath Agent Team, a system built by the authors. Nearly every step is machine-checked in Lean. Two pieces stand outside that check. One is an externally cited theorem. The other is a statement verified by symbolic computation. The argument leans on ternary quadratic forms, genus theory, and finite descent. The result closes the universality question for this triple.
Catalog record
| CREATOR | Yichuan Cao \institution State Key Laboratory of Mathematical Sciences, Academy of Mathematics and Systems Science, Chinese Academy of Sciences, and University of Chinese Academy of Sciences. This work is supported by the Strategic Priority Research Program of Chinese Academy of Sciences under Grant XDA0480502 and XDA0480503, Dakai Guo \instref 1, Ruichen Qiu \instref 1, Ruyong Feng \instref 1, Xiao-Shan Gao \instref 1 |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 66 KB |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2606.26035
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Every Nonnegative Integer Is a Sum of a Triangular, a Pentagonal, and a Heptagonal Number Yichuan Cao \institution State Key Laboratory of Mathematical Sciences, Academy of Mathematics and Systems Science, Chinese Academy of Sciences, and University of Chinese Academy of Sciences. This work is supported by the Strategic Priority Research Program of Chinese Academy of Sciences under Grant XDA0480502 and XDA0480503, Dakai Guo \instref 1, Ruichen Qiu \instref 1, Ruyong Feng \instref 1, Xiao-Shan Gao \instref 1 Abstract In this paper, it is proved that any nonnegative integer can be written in the following form x ( x + 1 ) 2 + y ( 3 y + 1 ) 2 + z ( 5 z + 1 ) 2 , x , y , z ∈ ℕ . \frac{x(x+1)}{2}+\frac{y(3y+1)}{2}+\frac{z(5z+1)}{2},\qquad x,y,z\in\mathbb{N}. This settles the conjecture recorded as OEIS A287616. All parts of the proof have been formalized in Lean 4, with the exception of two results: one externally cited theorem and one statement verified by symbolic computation. Both the natural-language proof and the Lean formalization were generated by the MechMath Agent Team developed by the authors. 2020 Mathematics Subject Classification. Primary 11E25, 11D85; Secondary 11E20, 11Y50, 68V20. Keywords. Sums of polygonal numbers, ternary quadratic forms, genus theory, finite descent, computer-assisted proof, Lean formalization. 1 Introduction For a nonnegative integer x let T(x)=\frac{x(x+1)}{2},\qquad P(y)=\frac{y(3y+1)}{2},\qquad H(z)=\frac{z(5z+1)}{2} be, respectively, the x -th triangular number, the y -th second pentagonal number, and the z -th second heptagonal number. In his study of universal sums of three quadratic polynomials, Zhi-Wei Sun conjectured that the triple (T,P,H) is universal : every nonnegative integer is a sum of one value from each family [ 7 ] . The sequence of integers admitting such a representation is recorded as OEIS A287616 [ 3 ] . The present paper proves the conjecture. Theorem 1 (OEIS A287616) . For every n\in\mathbb{N} , there exist x,y,z\in\mathbb{N} such that n=T(x)+P(y)+H(z). Strategy. The three elementary identities 8T(x)+1=(2x+1)^{2},\quad 24P(y)+1=(6y+1)^{2},\quad 40H(z)+1=(10z+1)^{2} turn Theorem 1 into a constrained representation problem for the diagonal ternary form Q(u,v,w)=15u^{2}+5v^{2}+3w^{2}: an integer n has the desired representation if and only if m=120n+23 is represented by Q with u odd, v\equiv 1\pmod{6} , and w\equiv 1\pmod{10} . The proof then has two parts. First, an unconditional primitive seed : every m\equiv 23\pmod{120} is primitively represented by Q . This combines p -adic Hensel lifting, the two-class genus structure of Q , and an explicit transfer from the genus mate g_{2}=2x^{2}-2xy+8y^{2}+15z^{2} . Second, a finite descent : three rational similarities of Q lower a lexicographic potential \Phi until the congruence side conditions are met. Away from a thin residual cone the descent is elementary. On the cone, a finite-key determinacy theorem shows that the success of every short move-word depends only on a finite sign-and-residue key of the starting vector, and an exact-rational linear-programming certificate proves that every one of the finitely many remaining cells escapes. The descent is well founded for each fixed m , so it terminates at a good representation. Computer assistance and formalization. The reduction, the primitive seed, the move calculus, and the residual structure are ordinary mathematics. The single computational input is the finite cover of the residual cone, discharged by exact integer and rational certificates with no floating-point feasibility decision. The mathematical core — the statement, the reduction, the primitive seed, the transport law and elementary descents, and the residual gap analysis — has in addition been formalized in Lean 4 over Mathlib, and the conjecture itself is machine-checked. In fact every conclusion of the proof is formalized except two: one classical theorem that we cite — the genus-theoretic primitive local–global statement for Q (the Hasse–Minkowski theorem together with the two-class genus fact) — and one statement established by the computer-verified finite cover. These two enter the development as explicitly named inputs, and everything built on them is machine-checked. Section 8 describes this development. Both the natural-language proof and the Lean formalization were generated by the MechMath Agent Team developed by the authors [ 2 ] . Organization. Section 2 states the square reduction. Sections 3 – 7 give the mathematical proof: the primitive seed, the move calculus and descent potential, the residual cone, the finite cover that discharges the residual descent, and the final descent. Section 8 describes the Lean formalization, and Section 9 gives conclusions. The appendices collect the move ledger, the determinacy and cover certificates, and the soundness details. 2 Statement and square reduction Define T(x)=\frac{x(x+1)}{2},\qquad P(y)=\frac{y(3y+1)}{2},\qquad H(z)=\frac{z(5z+1)}{2}. The target is the following statement. Theorem 2 (A287616) . For every n\in\mathbb{N} there exist x,y,z\in\mathbb{N} such that n=T(x)+P(y)+H(z). The nonnegativity of x,y,z is part of the statement recorded in OEIS A287616. It is not the “generalized polygonal” variant in which the variables are allowed to range over all of \mathbb{Z} ; that all-integer variant is a different result in Sun’s work [ 7 ] . The elementary identities 8T(x)+1=(2x+1)^{2},\quad 24P(y)+1=(6y+1)^{2},\quad 40H(z)+1=(10z+1)^{2} give 120\bigl(T(x)+P(y)+H(z)\bigr)+23=15(2x+1)^{2}+5(6y+1)^{2}+3(10z+1)^{2}. Thus put Q(u,v,w)=15u^{2}+5v^{2}+3w^{2},\qquad A=\operatorname{diag}(15,5,3), so that Q(x)=x^{t}Ax . We shall prove the following equivalent representation theorem. Theorem 3 (Constrained ternary representation) . Every integer m\equiv 23\pmod{120} has a representation m=Q(u,v,w) with u positive and odd, v\equiv 1\pmod{6} , and w\equiv 1\pmod{10} . Equivalence of Theorems 2 and 3 . If n=T(x)+P(y)+H(z) , then m=120n+23 is represented by (u,v,w)=(2x+1,6y+1,10z+1) , and these coordinates satisfy the displayed positivity and congruence conditions. Conversely, if m=120n+23=Q(u,v,w) with the displayed conditions, then x=\frac{u-1}{2},\qquad y=\frac{v-1}{6},\qquad z=\frac{w-1}{10} are nonnegative integers, and substituting into the three square identities recovers n=T(x)+P(y)+H(z) . Because Q is diagonal, changing signs of u,v,w does not change Q ; in the descent below it is therefore enough to reach |v|\equiv 1\pmod{6} and |w|\equiv 1\pmod{10} after normalization. ∎ We call a representation primitive if \gcd(u,v,w)=1 . We call its normalized state (|u|,|v|,|w|) good if all three coordinates are odd, |v|\equiv 1\pmod{6} , and |w|\equiv 1\pmod{10} . The proof first obtains a primitive representation of m and then lowers a lexicographic potential until a good state is reached. Equivalently, among the finitely many primitive representations of a fixed m the descent locates one whose normalized state is good; by the equivalence above this yields the desired representation of n . 3 The primitive seed 3.1 Local primitive representability We begin with the local input. Let B(x,y) be the polar form of Q : B(x,y)=15x_{1}y_{1}+5x_{2}y_{2}+3x_{3}y_{3}. Lemma 4 (Local primitive representation) . Let m\equiv 23\pmod{120} . For every prime p , there is x_{p}\in\mathbb{Z}_{p}^{3} such that Q(x_{p})=m and at least one coordinate of x_{p} is a p -adic unit. The form also represents m over \mathbb{R} . Proof. We use two standard elementary facts, recalled for completeness. First, we use the following strong Hensel form, in the valuation-theoretic version of 6 . If f\in\mathbb{Z}_{p}[t] and t_{0}\in\mathbb{Z}_{p} satisfy v_{p}(f(t_{0}))\geq 2k+1,\qquad v_{p}(f^{\prime}(t_{0}))=k, then there is t\in\mathbb{Z}_{p} with f(t)=0 and t\equiv t_{0}\pmod{p^{k+1}} . Indeed, Newton iteration t_{j+1}=t_{j}-f(t_{j})/f^{\prime}(t_{j}) is valid because the derivative valuation remains k ; Taylor’s formula f(X+H)=f(X)+f^{\prime}(X)H+H^{2}g(X,H) shows the error valuation increases, while the increments have valuations tending to infinity. Completeness of \mathbb{Z}_{p} gives the limit. Second, for odd p\nmid ab , the binary form ax^{2}+by^{2} takes every residue class modulo p . The two sets \{ax^{2}:x\in\mathbb{F}_{p}\} and \{c-by^{2}:y\in\mathbb{F}_{p}\} both have size (p+1)/2 in