Conditional positivity conjecture for Q01Q_{01}

From papers

Let pp and qq be the parameters defining the polynomial Q01Q_{01}. A polynomial has nonnegative coefficients when every coefficient in its expansion is at least zero.

Conditional positivity conjecture. Q01Q_{01} has nonnegative coefficients if and only if p=2p=2 and q1mod4q\equiv 1 \mod 4.

This gives the precise parameter condition for positivity in the P21P_{21} and Q01Q_{01} case. The supplied text does not state whether the claim has been proved or remains open.

Progress summary

Open

No public discussion or published progress was found on this conjecture.

No public discussion or published progress was found.

Current status (as of August 2026): it appears open with no recorded activity.

Sources & referencesView supporting material

Primary source

Evelyn Fiore, George D. Nasr and Cooper Stone, “Dice Relabeling Using Square-Sided Dice”, arXiv:2606.20311 (2026).

Solutions 1

Proof

Proof of the full conditional-positivity conjecture

The source already proves the p=2p=2 classification in Theorem 5.8. It remains to prove that Q01Q_{01} has a negative coefficient for every odd prime pp and every distinct prime qq.

Write

p=2h+1,P=p2,T(x)=Φq(x)2,Δ(x)=(1xp)T(x).p=2h+1,\qquad P=p^2,\qquad T(x)=\Phi_q(x)^2,\qquad \Delta(x)=(1-x^p)T(x).

The coefficients of TT are

tj={j+1,0j<q,2q1j,qj2q2,0,otherwise.t_j= \begin{cases} j+1,&0\le j<q,\\ 2q-1-j,&q\le j\le2q-2,\\ 0,&\text{otherwise}. \end{cases}

The exact cyclotomic identity

Φp2q(x)=(1xp2q)(1xp)(1xp2)(1xpq)\Phi_{p^2q}(x) =\frac{(1-x^{p^2q})(1-x^p)} {(1-x^{p^2})(1-x^{pq})}

gives

Q01(x)x=Δ(x)(1xPq)(1xP)(1xpq).\frac{Q_{01}(x)}x =\frac{\Delta(x)(1-x^{Pq})} {(1-x^P)(1-x^{pq})}.

Therefore, whenever 0k<pq0\le k<pq,

[xk]Q01(x)x=j0ΔkjP,Δi=titip.(1)[x^k]\frac{Q_{01}(x)}x =\sum_{j\ge0}\Delta_{k-jP}, \qquad \Delta_i=t_i-t_{i-p}. \tag{1}

If qhq\le h, take k=pk=p. Since 2q2<p2q-2<p,

[xp](Q01/x)=tpt0=1.[x^p](Q_{01}/x)=t_p-t_0=-1.

If h<q<Phh<q<P-h, take k=q+h<Pk=q+h<P. Then

[xq+h](Q01/x)=tq+htq+hp=(qh1)(qh)=1.[x^{q+h}](Q_{01}/x) =t_{q+h}-t_{q+h-p} =(q-h-1)-(q-h)=-1.

This remains valid at q=h+1q=h+1.

It remains to consider qPhq\ge P-h. The support of Δ\Delta lies in

[0,M],M=2q+p2.[0,M],\qquad M=2q+p-2.

The complete block

MkM+P1(2)M\le k\le M+P-1 \tag{2}

lies below pqpq: for p5p\ge5,

(p2)q(p2)(p2p+1)=p2+p2+p(p24p+2)p2+p2.(p-2)q\ge(p-2)(p^2-p+1) =p^2+p-2+p(p^2-4p+2) \ge p^2+p-2.

For p=3p=3, the assumption q8q\ge8 forces the distinct prime q11q\ge11, and M+P1=2q+9<3qM+P-1=2q+9<3q.

For r{0,,P1}r\in\{0,\ldots,P-1\}, define

Lr=0iM\ir(modP)Δi.L_r=\sum_{\substack{0\le i\le M\i\equiv r\pmod P}}\Delta_i.

By (1), the coefficients in block (2) are exactly the PP numbers LrL_r, each occurring once. Their sum is

r=0P1Lr=Δ(1)=0.\sum_{r=0}^{P-1}L_r=\Delta(1)=0.

If all these coefficients were nonnegative, every LrL_r would vanish. Reducing Δ(x)\Delta(x) modulo xP1x^P-1 would then imply

xp21(1xp)Φq(x)2.x^{p^2}-1\mid(1-x^p)\Phi_q(x)^2.

Since

xp21=(xp1)Φp2(x),x^{p^2}-1=(x^p-1)\Phi_{p^2}(x),

cancellation would give

Φp2(x)Φq(x)2,\Phi_{p^2}(x)\mid\Phi_q(x)^2,

contradicting the coprimality of distinct cyclotomic polynomials. Therefore at least one coefficient in (2) is strictly negative.

This covers every odd prime pp. Combining it with the already proved p=2p=2 case of Theorem 5.8 establishes the complete classification

  Q01(x)Z0[x]p=2 and q1(mod4).  \boxed{\; Q_{01}(x)\in\mathbb Z_{\ge0}[x] \quad\Longleftrightarrow\quad p=2\ \text{and}\ q\equiv1\pmod4. \;}
0 endorsements
Shivam Patel ·