Conjectured arithmetic congruences for the 9-color overpartition function

From papers

Let p9(n)\overline{p}_{9}(n) denote the 99-color overpartition function, and let n0n\ge0 and α1\alpha\ge1. Arithmetic congruence conjecture. For all n0n\ge0 and α1\alpha\ge1,

p9(2α(8n+7))0(mod16),p9(36n+33)0(mod128).\begin{aligned} \overline{p}_{9}\bigl(2^{\alpha}(8n+7)\bigr)&\equiv0\pmod{16},\\ \overline{p}_{9}(36n+33)&\equiv0\pmod{128}. \end{aligned}

These are presented as conjectured arithmetic properties based on the paper's computations; no proof or resolution is supplied.

Progress summary

Open

The two predicted divisibility patterns were recorded in a July 2026 preprint, but no proof, disproof, or independent verification has been found.

The conjecture predicts that p9(2α(8n+7))\overline{p}_{9}(2^{\alpha}(8n+7)) is divisible by 1616 and p9(36n+33)\overline{p}_{9}(36n+33) is divisible by 128128, for the stated ranges of nn and α\alpha.

July 2026 preprint

H. S. Sumanth Bharadwaj, N. Sujatha, and S. Chandankumar present both statements explicitly as conjectures based on computation. The preprint gives no proof, counterexample, or verification beyond the reported computations.

Current status (as of August 2026): Both congruences remain open conjectures; no publicly retrieved source establishes or refutes either one.

Sources
Sources & referencesView supporting material

Primary source

H. S. Sumanth Bharadwaj, N. Sujatha and S. Chandankumar, “Arithmetic properties of the 2-color overpartition function p_(n)”, arXiv:2607.16608 (2026).

Additional references

7 papers in this index state this conjecture (2013–2026). The statement above is taken from the most recent of them; the others are arXiv:2412.19998, arXiv:2307.04687, arXiv:2307.02579, arXiv:2101.04058, arXiv:1801.06990, arXiv:1304.0684.

Solutions 1

Proof

Both congruences hold. In fact, the first extends to every parameter 1(mod8)\ell\equiv1\pmod8 and every exponent α0\alpha\ge0, strengthening both restrictions in the conjecture:

p(2α(8n+7))0(mod16)(1 ⁣ ⁣(mod8), α,n0),\boxed{\overline p_\ell(2^\alpha(8n+7)) \equiv0\pmod{16} \quad (\ell\equiv1\!\!\pmod8,\ \alpha,n\ge0),}

and

p9(36n+33)0(mod128)(n0).\boxed{\overline p_9(36n+33)\equiv0\pmod{128} \quad(n\ge0).}

For the first identity, put

U(q)=j1(1)jqj2,V(q)=U(q).U(q)=\sum_{j\ge1}(-1)^jq^{j^2}, \qquad V(q)=U(q^\ell).

The theta identity gives

N0p(N)qN=1(1+2U)(1+2V)\sum_{N\ge0}\overline p_\ell(N)q^N = \frac1{(1+2U)(1+2V)}

and consequently, modulo 1616,

12(U+V)+4(U2+UV+V2)8(U3+U2V+UV2+V3).(1)1-2(U+V) +4(U^2+UV+V^2) -8(U^3+U^2V+UV^2+V^3). \tag{1}

Its terms count representations by at most three positive weighted squares, with weights 1,1(mod8)1,\ell\equiv1\pmod8.

Write M=8n+7M=8n+7. If α=2a\alpha=2a, any representation of 4aM4^aM by at most three weighted squares has every root divisible by 2a2^a. Dividing yields a representation of M7(mod8)M\equiv7\pmod8 by at most three residues from {0,1,4}\{0,1,4\}, which is impossible.

If α=2a+1\alpha=2a+1, dividing roots by 2a2^a leaves 2M6(mod8)2M\equiv6\pmod8, which is neither a weighted square nor a sum of two weighted squares. For each cubic term in (1), swap two roots of the same weight. A fixed point would satisfy

2a0x2+b0z2=2M,a0,b0{1,}.2a_0x^2+b_0z^2=2M, \qquad a_0,b_0\in\{1,\ell\}.

Necessarily z=2wz=2w and xx is odd, so

M=a0x2+2b0w21 or 3(mod8),M=a_0x^2+2b_0w^2\equiv1\text{ or }3\pmod8,

contradicting M7(mod8)M\equiv7\pmod8. Therefore every cubic coefficient is even, and its factor 88 makes it vanish modulo 1616. This proves the stronger all-\ell, all-α\alpha theorem.

For the second congruence, write fj=(qj;qj)f_j=(q^j;q^j)_\infty. The source's exact three-dissection gives

B(q)=r0p9(3r)qr=f24f34f18f62.B(q) = \sum_{r\ge0}\overline p_9(3r)q^r = \frac{f_2^4f_3^4}{f_1^8f_6^2}.

The target corresponds to exponents 12n+1112n+11 of BB. Form the eta quotient

E(τ)=η(2τ)4η(3τ)4η(12τ)72η(τ)8η(6τ)2=q36f1272B(q).E(\tau) = \frac{\eta(2\tau)^4\eta(3\tau)^4\eta(12\tau)^{72}} {\eta(\tau)^8\eta(6\tau)^2} = q^{36}f_{12}^{72}B(q).

At level 2424, its eta exponent vector is

(r1,r2,r3,r6,r12)=(8,4,4,2,72).(r_1,r_2,r_3,r_6,r_{12})=(-8,4,4,-2,72).

The weighted eta sums are

rδ=70,δrδ=864,(24/δ)rδ=24.\sum r_\delta=70,\qquad \sum\delta r_\delta=864,\qquad \sum(24/\delta)r_\delta=24.

At cusp denominators 1,2,3,4,6,8,12,241,2,3,4,6,8,12,24, the orders are respectively

1, 6, 19, 12, 18, 12, 36, 36.1,\ 6,\ 19,\ 12,\ 18,\ 12,\ 36,\ 36.

Hence EE is holomorphic of weight 3535, level 2424, and character χ4\chi_{-4}.

Project onto residue 11(mod12)11\pmod{12} using all four Dirichlet characters modulo 1212:

S11=ψmod12ψ(11)(Eψ)=4r11 (12)[qr]Eqr.S_{11} = \sum_{\psi\bmod12}\psi(11)(E\otimes\psi) = 4\sum_{r\equiv11\ (12)}[q^r]E\,q^r.

Primitive twisting and the standard oldform correction for imprimitive characters show

S11M35(Γ0(3456),χ4).S_{11}\in M_{35}(\Gamma_0(3456),\chi_{-4}).

The subgroup index is 69126912, giving the complete Sturm bound

B=35691212=20160.B=\frac{35\cdot6912}{12}=20160.

The required exact finite certificate is reproducible entirely with integer arithmetic:

a(0)=1,a(r)=2j2r(1)j+1a(rj2)(mod128),a(0)=1,\qquad a(r)= 2\sum_{j^2\le r}(-1)^{j+1}a(r-j^2) \pmod{128},

and

p9(r)=j=0r/9a(r9j)a(j)(mod128).\overline p_9(r) = \sum_{j=0}^{\lfloor r/9\rfloor} a(r-9j)a(j) \pmod{128}.

Computing through

3(B36)=603723(B-36)=60372

gives

p9(36j+33)0(mod128)(0j1676).\overline p_9(36j+33)\equiv0\pmod{128} \qquad(0\le j\le1676).

These 16771677 exact coefficients cover the full Sturm bound after the q36q^{36} shift and multiplication by the unit f1272f_{12}^{72}. Therefore

S110(mod512)S_{11}\equiv0\pmod{512}

identically by Sturm's theorem. Dividing its coefficients by 44 and removing the residue-preserving unit proves the desired modulo-128128 congruence for every nn.

Source: H. S. Sumanth Bharadwaj, N. Sujatha and S. Chandankumar, arXiv:2607.16608, equation (3.3) and Conjecture 3, equations (5.7)–(5.8).

0 endorsements
Shivam Patel ·