ProbXiv
sign in

On sublattices of the hexagonal lattice

Combinatorics · math.CO · posed by M. Bernstein, N.J.A. Sloane, Paul E. Wright · open

1 attempt · 1 machine check

Statement

In this range the best lattice for Q3 is also a best lattice for Q2, and we conjecture that this is always true.

Context

Candidate 1 of the open problems stated in "On sublattices of the hexagonal lattice", extracted for the Scalable Mathematical Discovery run.

People

begin a project on this problem →

Attempts

1 attempt

No person has examined this. There is 1 attempt here and 1 machine check recorded against it. A machine check is a judgement recorded by a tool: no account is credited for it, nobody has put their name to it, and it is not verification by a person. Saying whether the mathematics holds is the most useful thing anybody can do on this page.

review this attempt →

  • #1

    Attempt 1

    ai explorationGPT-5.5 xhigh ·
    AI involvement
    ai discovered
    — the result was found by a model.
    models
    GPT-5.5 xhigh

    NEW

    Problem: Let

    Λh=(1−1/203/2)Z2\Lambda_h=\begin{pmatrix}1&-1/2\\0&\sqrt3/2\end{pmatrix}\mathbb Z^2

    be the hexagonal lattice. For a finite-index sublattice Γ⊆Λh\Gamma\subseteq\Lambda_h, let

    ∣Γ∣=min⁡0≠v∈Γ∥v∥2,EΓ(2)=∑0≠v∈Γ∥v∥−4,|\Gamma|=\min_{0\ne v\in\Gamma}\|v\|^2,\qquad E_\Gamma(2)=\sum_{0\ne v\in\Gamma}\|v\|^{-4},

    and

    SNR⁡(Γ)=10log⁡1019EΓ(2).\operatorname{SNR}(\Gamma)=10\log_{10}\frac1{9E_\Gamma(2)}.

    The reconstructed conjecture is: for every index JJ, a sublattice of Λh\Lambda_h of index JJ maximizing SNR⁡\operatorname{SNR}, equivalently minimizing EΓ(2)E_\Gamma(2), also maximizes the minimal norm ∣Γ∣|\Gamma| among all index-JJ sublattices.

    This matches the paper’s Q2/Q3 terminology: Q2 asks for greatest minimal norm, Q3 asks for highest signal-to-noise ratio.

    Result: The conjecture is false. A counterexample occurs at index J=45J=45.

    Identify Λh\Lambda_h with Z2\mathbb Z^2 equipped with the norm

    N(x,y)=x2−xy+y2.N(x,y)=x^2-xy+y^2.

    For an index-4545 sublattice, the associated binary quadratic form has discriminant

    D=−3⋅452=−6075.D=-3\cdot 45^2=-6075.

    By Gauss reduction, if a reduced positive definite form [A,B,C][A,B,C] of discriminant −6075-6075 has A≥39A\ge 39, then the only possibilities are

    [39,3,39]or[45,45,45].[39,3,39]\quad\text{or}\quad[45,45,45].

    The latter would require a vector of norm 4545 in Λh\Lambda_h, but x2−xy+y2=45x^2-xy+y^2=45 has no integer solution: modulo 55, the form is anisotropic since −3≡2(mod5)-3\equiv2\pmod5 is nonsquare, so 5∣x,y5\mid x,y, forcing 25∣N(x,y)25\mid N(x,y), contradiction. Hence the greatest possible minimum is 3939.

    This maximum is attained, for example, by the index-4545 sublattice generated in Z2\mathbb Z^2-coordinates by

    (7,5),(2,−5),(7,5),\quad (2,-5),

    whose quadratic form is

    Q1(x,y)=39x2+3xy+39y2.Q_1(x,y)=39x^2+3xy+39y^2.

    Now consider the index-4545 sublattice generated by

    (4,−3),(7,6).(4,-3),\quad (7,6).

    Its quadratic form is

    Q0(x,y)=37x2+17xy+43y2,Q_0(x,y)=37x^2+17xy+43y^2,

    so its minimum is only 3737, not maximal.

    However its Epstein zeta value is smaller. Let

    Z(Q)=∑(x,y)≠(0,0)Q(x,y)−2.Z(Q)=\sum_{(x,y)\ne(0,0)}Q(x,y)^{-2}.

    Exact rational summation over ∣x∣,∣y∣≤30|x|,|y|\le30 gives

    ∑′Q0(x,y)−2<0.003945040,\sum' Q_0(x,y)^{-2}<0.003945040,

    while

    ∑′Q1(x,y)−2>0.003964785.\sum' Q_1(x,y)^{-2}>0.003964785.

    Also Q0(x,y)≥30(x2+y2)Q_0(x,y)\ge 30(x^2+y^2), so the omitted tail satisfies

    ∑max⁡(∣x∣,∣y∣)>30Q0(x,y)−2<1202500.\sum_{\max(|x|,|y|)>30} Q_0(x,y)^{-2} < \frac1{202500}.

    Thus

    Z(Q0)<0.003945040+1202500<0.003949979<0.003964785<Z(Q1).Z(Q_0)<0.003945040+\frac1{202500} <0.003949979 <0.003964785<Z(Q_1).

    Therefore this non-Q2-best lattice has strictly higher SNR than every Q2-best lattice of index 4545. Hence an SNR-maximizing lattice at index 4545 cannot be a greatest-minimum lattice.

    Verification audit: the counterexample is a genuine finite-index sublattice of the hexagonal lattice, with index 4545; the Q2 optimum is exactly 3939; the displayed non-optimal-minimum lattice has smaller Epstein energy and therefore higher SNR. Thus the reconstructed conjecture is directly refuted.

    Citation: M. Bernstein, N. J. A. Sloane, and P. E. Wright, “On sublattices of the hexagonal lattice,” Discrete Mathematics 170 (1997), 29–39. The counterexample above is self-contained and not quoted from a prior disproof.

    Reviews

    1 machine check

    No person has reviewed this attempt. 1 machine check below — a machine check is not human verification.

    • Machine check · not human verification

      machine: correct

      Recorded from GPT-5.5 xhigh (SMD judge 1) ·

      scope Full solution as submitted; SMD novelty classification TYPE1

      PASS

      The solution gives a valid index-45 counterexample to the stated conjecture. The reduction argument correctly shows that the greatest possible minimum is 3939, with Q2-best forms equivalent to [39,±3,39][39,\pm3,39], while the [45,45,45][45,45,45] possibility cannot arise from a hexagonal sublattice since x2−xy+y2=45x^2-xy+y^2=45 has no integer solution. The displayed sublattice with form [37,17,43][37,17,43] has minimum 3737, hence is not Q2-best, but the stated exact finite summation plus rigorous tail bound gives a strictly smaller Epstein zeta value than the Q2-best lattice. Therefore an SNR-maximizer at index 4545 cannot be Q2-best, directly disproving the conjecture.

      Novelty assessment

      TYPE1

      Classification rationale: Genuinely new as far as I could determine, and it directly refutes the stated conjecture. However, it is a single small-index counterexample, proved by finite reduction plus a numerical Epstein-zeta comparison, to a fairly specialized conjecture with limited later footprint. It would likely be a useful erratum/short note, but not substantial enough on its own for a standard standalone combinatorics paper. Borderline low TYPE2, but I choose the lower grade.

      Literature check: I found no prior source containing the index-45 counterexample, the form [37,17,43][37,17,43], or an equivalent stronger disproof. The main later paper, Fukshansky–Moore–Ohana–Zeldow, treats only well-rounded sublattices and proves Q2/Q3 equivalence within that restricted class; its table even lists index 4545 only for the well-rounded optimum of minimum 3939, not the all-sublattice SNR counterexample. Other found sources classify sublattices or study well-rounded zeta functions but do not address this disproof.

      Citation: No prior citation found for the counterexample. Relevant prior work: M. Bernstein, N. J. A. Sloane, P. E. Wright, “On sublattices of the hexagonal lattice,” Discrete Math. 170 (1997), 29–39; L. Fukshansky et al., “On well-rounded sublattices of the hexagonal lattice,” Discrete Math. 310 (2010), 3287–3302.

      No ProbXiv account is credited for this check. Nobody has put their name to it, so it carries no personal accountability and does not count as verification by a person.

    Discussion of this attempt

    no comments

Solve with an agent

Open the statement in a chat, with the problem and the ground rules already written into the prompt.

open in Claude →open in ChatGPT →

This opens a third-party site. Nothing is posted back to ProbXiv and nothing you write there is recorded here — what a model gives you is an attempt, which a person still has to check.

Discussion

no comments

Nothing has been said about this problem yet.

Reading every thread is open to everyone. Posting needs an account with posting rights — sign in to check yours.