Asymptotic size conjecture for the constructed cone

About 8 years old · traced to

Let n=4k+7n=4k+7 with k≥2k\geq 2, and let Cone⁡=Cone⁡(Cone⁡2k+1∪Cone⁡2k+3)\operatorname{Cone}=\operatorname{Cone}(\operatorname{Cone}_{2k+1}\cup\operatorname{Cone}_{2k+3}). Cone asymptotic-size conjecture. Is

lim⁡n→∞∣Cone⁡∣2n−2=1?\lim_{n\to\infty}\frac{|\operatorname{Cone}|}{2^{n-2}}=1?

Numerical checks through k=249k=249 found ratios below 11 and increasingly close to it. The conjecture asks whether this observed limiting behavior holds asymptotically.

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 to prove the expected limiting size, but the proof has not been independently checked.

The conjecture asks whether the constructed cone has asymptotic size 2n−22^{n-2} when n=4k+7n=4k+7 tends to infinity. The original source records numerical ratios below 11 and approaching 11, but no published resolution.

Community submission (unverified)

Posted August 26, 2026. A submitted calculation claims the exact identity ∣Cone⁡∣=2n−2−3(4k2k)|\operatorname{Cone}|=2^{n-2}-3\binom{4k}{2k} for n=4k+7n=4k+7. If correct, the normalized deficit tends to zero, proving the conjecture and strict inequality for every k≥2k\ge 2. The displayed derivation is truncated and remains unverified.

Current status (as of August 2026): The conjecture remains unverified; a community submission claims an identity implying the limit, but no independently confirmed proof is recorded.

Sources

Solutions 1

This solution needs a summarySee full solutionHide full solution

Exact deficit and the asymptotic size of the cone

Statement

For k≥2k\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

Cone⁡=Cone⁡(Cone⁡2k+1∪Cone⁡2k+3)⊆Pn(∗).\operatorname{Cone} =\operatorname{Cone}(\operatorname{Cone}_{2k+1} \cup\operatorname{Cone}_{2k+3}) \subseteq\mathcal P_n(*) .

The authors conjecture that

lim⁡k→∞∣Cone⁡∣2n−2=1.\lim_{k\to\infty} \frac{|\operatorname{Cone}|}{2^{n-2}}=1.

We prove the stronger exact identity

∣Cone⁡∣=2n−2−3(4k2k).(1)\boxed{ |\operatorname{Cone}|=2^{n-2}-3\binom{4k}{2k}.} \tag{1}

Thus the source's observed strict inequality ∣Cone⁡∣<2n−2|\operatorname{Cone}|<2^{n-2} holds for every k≥2k\ge2, and the normalized deficit tends to zero.

The four source layers

For an integer tt, write

Bt=(mr+t).B_t=\binom{m}{r+t}.

Since m=2rm=2r, binomial symmetry gives B−t=BtB_{-t}=B_t. The source's layer description gives

Lr+1=B1,Lr+3=7B0+21B1+7B2+B3,Lr+5=B5+7B4+21B3+35B2+35B1+21B0,(2)\begin{aligned} L_{r+1}&=B_1,\\ L_{r+3}&=7B_0+21B_1+7B_2+B_3,\\ L_{r+5}&=B_5+7B_4+21B_3+35B_2+35B_1+21B_0, \end{aligned} \tag{2}

and every odd layer of size at least r+7r+7. Hence, if

T7=∑j≥r+7\j odd(nj),T_7=\sum_{\substack{j\ge r+7\j\text{ odd}}}\binom nj,

then

∣Cone⁡∣=Lr+1+Lr+3+Lr+5+T7.(3)|\operatorname{Cone}|=L_{r+1}+L_{r+3}+L_{r+5}+T_7. \tag{3}

The odd upper tail

Let

T5=∑j≥r+5\j odd(nj).T_5=\sum_{\substack{j\ge r+5\j\text{ odd}}}\binom nj.

The integer n=2r+7n=2r+7 is odd. Complementation therefore shows that the full upper half, starting at r+4r+4, has size 2n−12^{n-1}. The alternating-tail identity

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

with the even integer s=r+4s=r+4 says that, within this upper half, the number of even sets minus the number of odd sets is (n−1r+3)\binom{n-1}{r+3}. Consequently

T5=2n−2−12(n−1r+3),T7=T5−(nr+5).(5)T_5=2^{n-2}-\frac12\binom{n-1}{r+3}, \qquad T_7=T_5-\binom n{r+5}. \tag{5}

Two Vandermonde collections

First, n−1=m+6n-1=m+6, so Vandermonde's identity and B−t=BtB_{-t}=B_t give

12(n−1r+3)=10B0+15B1+6B2+B3.(6)\frac12\binom{n-1}{r+3} =10B_0+15B_1+6B_2+B_3. \tag{6}

Second, n=m+7n=m+7, and another Vandermonde expansion gives

(nr+5)=∑i=07(7i)B5−i=B5+7B4+21B3+36B2+42B1+21B0.(7)\begin{aligned} \binom n{r+5} &=\sum_{i=0}^{7}\binom7i B_{5-i}\\ &=B_5+7B_4+21B_3+36B_2+42B_1+21B_0. \end{aligned} \tag{7}

Comparing (7) with the third line of (2),

Lr+5−(nr+5)=−B2−7B1.(8)L_{r+5}-\binom n{r+5}=-B_2-7B_1. \tag{8}

Now substitute (2), (5), (6), and (8) into (3). The difference from 2n−22^{n-2} is

∣Cone⁡∣−2n−2=B1+(7B0+21B1+7B2+B3)−B2−7B1−(10B0+15B1+6B2+B3)=−3B0.\begin{aligned} |\operatorname{Cone}|-2^{n-2} ={}&B_1+(7B_0+21B_1+7B_2+B_3)\\ &{}-B_2-7B_1-(10B_0+15B_1+6B_2+B_3)\\ =&-3B_0. \end{aligned}

This is exactly (1).

Taking the limit

Because n−2=4k+5n-2=4k+5, equation (1) gives

∣Cone⁡∣2n−2=1−332(4k2k)24k.(9)\frac{|\operatorname{Cone}|}{2^{n-2}} =1-\frac3{32}\frac{\binom{4k}{2k}}{2^{4k}}. \tag{9}

The remaining factor tends to zero elementarily. Indeed,

(4k2k)24k=∏j=12k(1−12j)≤exp⁡ ⁣(−12∑j=12k1j)⟶0.\frac{\binom{4k}{2k}}{2^{4k}} =\prod_{j=1}^{2k}\left(1-\frac1{2j}\right) \leq \exp\!\left(-\frac12\sum_{j=1}^{2k}\frac1j\right) \longrightarrow0.

Taking the limit in (9) proves the conjecture.

Lean: https://github.com/antoshashakov/Principia-Math-In-Progress/blob/main/mathdb-open-problems/problems/338641/Problem338641.lean

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

Models used: GPT 5.6 Sol, Fable