Asymptotic size conjecture for the constructed system D

Let n=4k+7n=4k+7 and let D\mathcal{D} be the constructed maximal commutative algebraic system in Pn()\mathcal{P}_n(*). D\mathcal{D} asymptotic-size conjecture. Is

limnD2n2=1?\lim_{n\to\infty}\frac{|\mathcal{D}|}{2^{n-2}}=1?

Numerical checks through k=249k=249 found ratios above 11 and decreasing toward it. The conjecture asks whether the ratio converges to 11 as nn tends to infinity.

References

Primary source

Victor A. Bovdi and Ho-Hon Leung, “Maximal commutative subalgebras of a Grassmann algebra”, arXiv:1803.03457 (2018).

Progress summary

Refreshed
Claimed progress

A reader-submitted calculation claims the conjecture is true, but no independent verification has been found.

Bovdi and Leung posed the conjecture in 2018 for their maximal commutative algebraic system D\mathcal{D}, asking whether its normalized size approaches 11. Their numerical tests supported this, but did not prove convergence.

Known results

  • Bovdi and Leung (2018): D\mathcal{D} is maximal and the computed ratios satisfy 1<dk1<d_k through k=249k=249.
  • Bovdi and Leung (2018): d2=1.0188d_2=1.0188 and d249=1.0031d_{249}=1.0031, with the observed sequence decreasing.

August 26, 2026 community submission

A submitted proof argues the exact identity

D2n2=2(4k+3)(2k1)(2k+1)(2k+2)(4k2k),|\mathcal{D}|-2^{n-2}=\frac{2(4k+3)(2k-1)}{(2k+1)(2k+2)}\binom{4k}{2k},

for k2k\ge 2, and concludes that the excess divided by 2n22^{n-2} tends to zero. If correct, this proves the conjecture; the argument is unverified and has no independent source.

Current status (as of August 2026): The original conjecture remains open in the published record; a community submission claims a complete proof, but that claim is unverified.

Sources

Solutions 1

Exact excess and the asymptotic size of D\mathcal D

Statement

For k2k\ge 2, put

n=4k+7,m=4k,r=2k.n=4k+7,\qquad m=4k,\qquad r=2k.

Section 4 of Bovdi and Leung's construction gives a maximal commutative algebraic system DPn()\mathcal D\subseteq\mathcal P_n(*) with

Dr+3=7(mr)+21(mr+1)+7(mr+2)+(mr+3),Dj=(nj)(jr+5 odd),(1)\begin{aligned} |\mathcal D_{r+3}| &=7\binom mr+21\binom m{r+1} +7\binom m{r+2}+\binom m{r+3},\\ |\mathcal D_j|&=\binom nj \qquad(j\ge r+5\text{ odd}), \end{aligned} \tag{1}

and no other layers. They conjecture that

limkD2n2=1.\lim_{k\to\infty}\frac{|\mathcal D|}{2^{n-2}}=1.

We prove the stronger exact identity

D2n2=2(4k+3)(2k1)(2k+1)(2k+2)(4k2k).(2)\boxed{ |\mathcal D|-2^{n-2} =\frac{2(4k+3)(2k-1)}{(2k+1)(2k+2)} \binom{4k}{2k}.} \tag{2}

In particular, the excess is positive for every k2k\ge2, and its ratio to 2n22^{n-2} tends to zero. This proves the conjecture.

The odd upper tail

Let

T=jr+5\j odd(nj).T=\sum_{\substack{j\ge r+5\j\text{ odd}}}\binom nj.

Because n=2r+7n=2r+7 is odd, complement symmetry gives

j=r+4n(nj)=2n1.(3)\sum_{j=r+4}^{n}\binom nj=2^{n-1}. \tag{3}

We also use the elementary alternating-tail identity

j=sn(1)j(nj)=(1)s(n1s1).(4)\sum_{j=s}^{n}(-1)^j\binom nj =(-1)^s\binom{n-1}{s-1}. \tag{4}

Here s=r+4=2k+4s=r+4=2k+4 is even. Thus the difference between the even and odd terms in the upper half is

j=r+4n(1)j(nj)=(n1r+3).\sum_{j=r+4}^{n}(-1)^j\binom nj=\binom{n-1}{r+3}.

The odd terms in that upper half begin at r+5=2k+5r+5=2k+5, exactly the terms defining TT. Combining this observation with (3) gives

T=2n212(n1r+3).(5)T=2^{n-2}-\frac12\binom{n-1}{r+3}. \tag{5}

The exceptional layer

Write

Bj=(mr+j).B_j=\binom m{r+j}.

The exceptional layer in (1) is

F=7B0+21B1+7B2+B3.(6)F=7B_0+21B_1+7B_2+B_3. \tag{6}

Since n1=m+6n-1=m+6, Vandermonde's identity and Bj=BjB_{-j}=B_j give

12(n1r+3)=12(m+6r+3)=12i=06(6i)B3i=10B0+15B1+6B2+B3.(7)\begin{aligned} \frac12\binom{n-1}{r+3} &=\frac12\binom{m+6}{r+3}\\ &=\frac12\sum_{i=0}^{6}\binom6i B_{3-i}\\ &=10B_0+15B_1+6B_2+B_3. \tag{7} \end{aligned}

Equations (5)--(7) therefore imply

D2n2=F12(n1r+3)=3B0+6B1+B2.(8)|\mathcal D|-2^{n-2}=F-\frac12\binom{n-1}{r+3} =-3B_0+6B_1+B_2. \tag{8}

The adjacent-binomial ratios are

B1B0=2k2k+1,B2B0=2k(2k1)(2k+1)(2k+2).\frac{B_1}{B_0}=\frac{2k}{2k+1},\qquad \frac{B_2}{B_0}=\frac{2k(2k-1)}{(2k+1)(2k+2)}.

Substitution into (8) yields

D2n2=(3+12k2k+1+2k(2k1)(2k+1)(2k+2))(4k2k)=2(4k+3)(2k1)(2k+1)(2k+2)(4k2k),\begin{aligned} |\mathcal D|-2^{n-2} &=\left( -3+\frac{12k}{2k+1} +\frac{2k(2k-1)}{(2k+1)(2k+2)} \right)\binom{4k}{2k}\\ &=\frac{2(4k+3)(2k-1)}{(2k+1)(2k+2)} \binom{4k}{2k}, \end{aligned}

which is (2).

Taking the limit

The rational prefactor in (2) tends to 44, while the standard central-binomial estimate gives

(4k2k)24k=O(k1/2)0.\frac{\binom{4k}{2k}}{2^{4k}}=O(k^{-1/2})\longrightarrow0.

Since 2n2=24k+5=3224k2^{n-2}=2^{4k+5}=32\,2^{4k}, division of (2) by 2n22^{n-2} proves

D2n2=1+O(k1/2)1.\frac{|\mathcal D|}{2^{n-2}} =1+O(k^{-1/2})\longrightarrow1.

This resolves the asymptotic-size conjecture.

Lean: https://github.com/antoshashakov/Principia-Math-In-Progress/blob/main/mathdb-open-problems/problems/338642/Problem338642.lean

Solved by the Principia Math harness. Check out our work at principia-math.com

Models used: GPT 5.6 Sol, Fable