Congruence conjectures for DSOME modulo 8 and 16

From papers

Let DSOME(n)DSOME(n) denote the sum of all odd parts in the partitions of nn into distinct parts minus the sum of all even parts. DSOME congruence conjecture. For all integers n0n\geq0,

DSOME(250n+21)DSOME(250n+71)DSOME(250n+121)DSOME(250n+171)DSOME(250n+221)0(mod8),DSOME(250n+21)\equiv DSOME(250n+71)\equiv DSOME(250n+121)\equiv DSOME(250n+171)\equiv DSOME(250n+221)\equiv0\pmod{8},

and

DSOME(500n+71)DSOME(500n+171)DSOME(500n+271)DSOME(500n+371)DSOME(500n+471)0(mod16).DSOME(500n+71)\equiv DSOME(500n+171)\equiv DSOME(500n+271)\equiv DSOME(500n+371)\equiv DSOME(500n+471)\equiv0\pmod{16}.

The conjecture is based on numerical calculations and proposes further congruences for the distinct-part function DSOMEDSOME; the supplied source gives no resolution.

Progress summary

Open

A 2026 paper claims to prove the first divisibility pattern, but the stronger second pattern remains unproved.

Baruah and Gogoi proposed the two DSOMEDSOME congruence families in 2026, based on numerical evidence. The first is equivalent to DSOME(50n+21)0(mod8)DSOME(50n+21)\equiv0\pmod{8}, and the second to DSOME(100n+71)0(mod16)DSOME(100n+71)\equiv0\pmod{16}; the supplied source did not resolve either one.

August 2026 claimed modulo-88 proof

Bardhan and Saikia state a proof of the modulo-88 conjecture, in fact proving DSOME(50n+10t+1)0(mod8)DSOME(50n+10t+1)\equiv0\pmod{8} for 1t41\leq t\leq4. The case t=2t=2 implies all five modulo-88 progressions in the problem. Their paper does not prove the modulo-1616 family.

Current status (as of August 2026): The modulo-88 assertion has a claimed but independently unverified arXiv proof, while the modulo-1616 assertion remains open.

Sources
Sources & referencesView supporting material

Primary source

Nayandeep Deka Baruah and Pankaj Gogoi, “Arithmetic properties of DSOME function”, arXiv:2602.20025 (2026).

Solutions 1

Proof

The two congruences for DSOME(n)DSOME(n)

For a nonnegative integer nn, let DSOME(n)DSOME(n) be the sum of the odd parts minus the sum of the even parts, taken over all partitions of nn into distinct parts. We prove Baruah and Gogoi's Conjecture 4.1: for every n0n\geq0,

DSOME(50n+21)0(mod8),DSOME(100n+71)0(mod16).DSOME(50n+21)\equiv0\pmod8, \qquad DSOME(100n+71)\equiv0\pmod{16}.

The first congruence was proved by Bardhan and Saikia in Theorem 4.11 and Remark 4.12. We prove the second. We will also use their quadratic-representation involution, explicitly given below. The additional step is a coefficientwise theta-series calculation modulo 1616; two parity decompositions then finish the argument.

1. A weighted pentagonal identity

All series in the proof are formal power series with integer coefficients. Put

f=r1(1qr),X=m1(1)mqm2,Xd=X(qd),D=qddq.f=\prod_{r\geq1}(1-q^r),\qquad X=\sum_{m\geq1}(-1)^m q^{m^2},\qquad X_d=X(q^d),\qquad D=q\frac{d}{dq}.

Thus X1=XX_1=X. Write ϕ=1+2X\phi=1+2X. The generating-function identity in Baruah and Gogoi, Theorem 1.1, together with ϕ=r1(1qr)2/r1(1q2r)\phi=\prod_{r\geq1}(1-q^r)^2/\prod_{r\geq1}(1-q^{2r}), gives

F(q):=n0DSOME(n)qn=f8(ϕ1ϕ3).(1)\mathcal F(q):=\sum_{n\geq0}DSOME(n)q^n =\frac{f}{8}\left(\phi^{-1}-\phi^3\right). \tag{1}

Set Pj=j(3j1)/2P_j=j(3j-1)/2 for jZj\in\mathbb Z. We use the classical identities recorded as (2.2) and (4.54) in Bardhan and Saikia:

f=jZ(1)jqPj,fϕ2=jZ(16j)qPj.(2)f=\sum_{j\in\mathbb Z}(-1)^j q^{P_j}, \qquad f\phi^2=\sum_{j\in\mathbb Z}(1-6j)q^{P_j}. \tag{2}

Define U=X(1+X)U=X(1+X) and Y=fUY=fU. Since ϕ2=1+4U\phi^2=1+4U, (2) implies

Y=jZ(1)jvjqPj,v2k=3k,v2k+1=3k+1.(3)Y=\sum_{j\in\mathbb Z}(-1)^j v_j q^{P_j}, \qquad v_{2k}=-3k,\quad v_{2k+1}=3k+1. \tag{3}

A direct calculation, valid for every integer jj, gives

3Pj=2vj2+vj.3P_j=2v_j^2+v_j.

Consequently vjPj(mod4)v_j\equiv P_j\pmod4. Substituting this into 3(Pjvj)=2vj(vj1)3(P_j-v_j)=2v_j(v_j-1) modulo 88 gives vj2Pj2Pj(mod8)v_j\equiv2P_j^2-P_j\pmod8. Therefore

Y(2D2D)f(mod8),YDf(mod4).(4)Y\equiv(2D^2-D)f\pmod8, \qquad Y\equiv Df\pmod4. \tag{4}

In particular, both ff and YY have zero coefficient at every index not of the form PjP_j.

2. The coefficient calculation modulo 1616

Fix an integer N0N\geq0 such that N3(mod4)N\equiv3\pmod4 and NPjN\neq P_j for every jZj\in\mathbb Z. For a series HH, write H=[qN]H\langle H\rangle=[q^N]H, and put

ad=fXd(d=1,2,4,8,16).a_d=\langle fX_d\rangle \qquad(d=1,2,4,8,16).

We claim that

DSOME(N)2Na112a48(a2+a8+a16)(mod16).(5)DSOME(N)\equiv -2Na_1-12a_4-8(a_2+a_8+a_{16}) \pmod{16}. \tag{5}

Here are the details. Factoring the numerator of (1) gives the exact identity

F=Y(1+2X21+2X).\mathcal F =-Y\left(1+\frac{2X^2}{1+2X}\right).

Since XX has zero constant term, expansion of the inverse is valid formally. With

Z=X22X3+4X4,Z=X^2-2X^3+4X^4,

we obtain FY2YZ(mod16)\mathcal F\equiv-Y-2YZ\pmod{16}. Our assumption on NN gives Y=0\langle Y\rangle=0, so

DSOME(N)2YZ(mod16).(6)DSOME(N)\equiv-2\langle YZ\rangle\pmod{16}. \tag{6}

For any series HH,

(Df)H=NfHfDH,(D2f)H=N2fH2NfDH+fD2H.\begin{aligned} \langle(Df)H\rangle &=N\langle fH\rangle-\langle fDH\rangle,\\ \langle(D^2f)H\rangle &=N^2\langle fH\rangle -2N\langle fDH\rangle+\langle fD^2H\rangle. \end{aligned}

Using (4), then N3(mod4)N\equiv3\pmod4, yields

YZ(2N)fZ+5fDZ+2fD2Z(mod8).(7)\langle YZ\rangle\equiv (2-N)\langle fZ\rangle +5\langle fDZ\rangle +2\langle fD^2Z\rangle \pmod8. \tag{7}

The following congruences follow coefficientwise by separating odd and even mm in the definition of XX:

DXX+X42X16(mod4),D2XX+X4(mod2),(DX)2X2+X8(mod2).(8)\begin{aligned} DX&\equiv X+X_4-2X_{16}\pmod4,\\ D^2X&\equiv X+X_4\pmod2,\\ (DX)^2&\equiv X_2+X_8\pmod2. \end{aligned} \tag{8}

For the last line, square the preceding parity expression and use Xd2X2d(mod2)X_d^2\equiv X_{2d}\pmod2.

Differentiating ZZ gives

DZ2UDX(mod8),2D2Z4(DX)2+4UD2X(mod8).\begin{aligned} DZ&\equiv2U\,DX\pmod8,\\ 2D^2Z&\equiv4(DX)^2+4U\,D^2X\pmod8. \end{aligned}

Substitution into (7), followed by (8), gives

YZf(NX2+4X4+6UX44UX16+4X2+4X8)(mod8).(9)\begin{aligned} \langle YZ\rangle\equiv \big\langle f\big(&-NX^2+4X^4+6UX_4\\ &-4UX_{16}+4X_2+4X_8\big)\big\rangle \pmod8. \end{aligned} \tag{9}

Indeed, before using N3(mod4)N\equiv3\pmod4, the coefficient of X3X^3 in this simplification is 2N+22N+2, which vanishes modulo 88.

To simplify (9), use fX2=a1\langle fX^2\rangle=-a_1, which follows from Y=0\langle Y\rangle=0, and X4X4(mod2)X^4\equiv X_4\pmod2. Also (4) and DXd0(modd)DX_d\equiv0\pmod d give

YX4Na4(mod4),YX16Na16(mod2).\langle YX_4\rangle\equiv Na_4\pmod4, \qquad \langle YX_{16}\rangle\equiv Na_{16}\pmod2.

It follows that

YZNa1+(4+6N)a44Na16+4a2+4a8Na1+6a4+4(a2+a8+a16)(mod8).\begin{aligned} \langle YZ\rangle &\equiv Na_1+(4+6N)a_4 -4Na_{16}+4a_2+4a_8\\ &\equiv Na_1+6a_4+4(a_2+a_8+a_{16}) \pmod8. \end{aligned}

Together with (6), this proves (5).

3. The progression 100n+71100n+71

Now let N=100n+71N=100n+71 with n0n\geq0, and put

M=24N+1=5(480n+341).M=24N+1=5(480n+341).

Thus the exponent of 55 in MM is exactly one. In particular MM is not a square, so NN is not pentagonal, since 24Pj+1=(6j1)224P_j+1=(6j-1)^2. Also N3(mod4)N\equiv3\pmod4, and (5) applies.

By (2), the sums ada_d have the explicit form

ad=jZ, m1Pj+dm2=N(1)j+m.(10)a_d= \sum_{\substack{j\in\mathbb Z,\ m\geq1\\ P_j+dm^2=N}}(-1)^{j+m}. \tag{10}

For d=2d=2, a summand would give M=(6j1)2+48m2M=(6j-1)^2+48m^2. Reduction modulo 55 forces both 6j16j-1 and mm to be divisible by 55, because 22 is a quadratic nonresidue modulo 55. This contradicts the exponent of 55 in MM. The same argument for d=8d=8 uses the nonresidue 33. Hence

a2=a8=0.(11)a_2=a_8=0. \tag{11}

Consider the finite set

R={(j,m)Z×Z>0:Pj+m2=N}.\mathcal R= \{(j,m)\in\mathbb Z\times\mathbb Z_{>0}: P_j+m^2=N\}.

We use the involution from the proof of Bardhan and Saikia's Theorem 4.11. For (j,m)R(j,m)\in\mathcal R, put u=6j1u=6j-1. Then u2+24m2=Mu^2+24m^2=M. Neither uu nor mm is divisible by 55, and there is a unique s{1,1}s\in\{1,-1\} such that 5u+sm5\mid u+sm. Define

u=24smu5,m=su+m5,j=u+16.(12)u' =\frac{24sm-u}{5},\qquad m'=\frac{|su+m|}{5},\qquad j'=\frac{u'+1}{6}. \tag{12}

Both uu' and mm' are integers, u1(mod6)u'\equiv-1\pmod6, and m>0m'>0: equality su+m=0su+m=0 would make M=25m2M=25m^2. Moreover,

(u)2+24(m)2=M,(u')^2+24(m')^2=M,

so (j,m)R(j',m')\in\mathcal R. To check the inverse explicitly, let ε\varepsilon be the sign of su+msu+m and put s=sεs'=s\varepsilon. Then

sm=u+sm5,u+sm=5sm.s'm'=\frac{u+sm}{5},\qquad u'+s'm'=5sm.

Thus ss' is the unique sign selected for (u,m)(u',m'), and applying (12) again returns uu and mm.

Furthermore,

mm+1(mod2),j=4smj+15j+1(mod2).m'\equiv m+1\pmod2,\qquad j'=\frac{4sm-j+1}{5}\equiv j+1\pmod2.

The involution therefore exchanges even and odd mm and preserves the sign (1)j+m(-1)^{j+m}.

For d=2,4,8d=2,4,8, set

Ed=(j,m)Rdm(1)j.E_d=\sum_{\substack{(j,m)\in\mathcal R\\ d\mid m}}(-1)^j.

Pairing by (12) gives a1=2E2a_1=2E_2. In addition, simply separating even and odd values of m/2m/2 and m/4m/4 gives

E2+a4=2E4,E4+a16=2E8.(13)E_2+a_4=2E_4, \qquad E_4+a_{16}=2E_8. \tag{13}

Using N3(mod4)N\equiv3\pmod4 and (13), we obtain

NE2+3a4+2a163(E2+a4)+2a16=6E4+2a162(E4+a16)=4E80(mod4).\begin{aligned} NE_2+3a_4+2a_{16} &\equiv3(E_2+a_4)+2a_{16}\\ &=6E_4+2a_{16}\\ &\equiv2(E_4+a_{16}) =4E_8\equiv0\pmod4. \end{aligned}

Finally, (5), (11) and a1=2E2a_1=2E_2 imply

DSOME(N)4(NE2+3a4+2a16)0(mod16).DSOME(N)\equiv -4(NE_2+3a_4+2a_{16}) \equiv0\pmod{16}.

This proves the second congruence for every n0n\geq0. Together with the previously proved modulo-88 congruence, it establishes all of Conjecture 4.1.

0 endorsements
Shivam Patel ·