The stable defective Kreweras number formula

About 2 years old · traced to

Let m,d,nm,d,n be nonnegative integers, and let 4λ⊢m4\lambda\vdash m be a partition of mm of length kk. For each ii, let 4μi(λ)4\mu_i(\lambda) be the number of parts of 4λ4\lambda of size ii. For n≥d+m−1n\geq d+m-1, the defective Kreweras number conjecture.

Krewd,n(λ)=m+dkm+d1m+d+1(m+d+1m+d+1−k,μ1(λ),μ2(λ),…,μm(λ)).\mathrm{Krew}_{d,n}(\lambda)=\frac{m+dk}{m+d}\frac{1}{m+d+1}{m+d+1\choose m+d+1-k,\mu_1(\lambda),\mu_2(\lambda),\ldots,\mu_m(\lambda)}.

This conjectured formula is a stable-range analogue of the formula for the ordinary Kreweras numbers. The supplied text gives no evidence that the conjecture has been proved or disproved, so its status remains open.

References

Primary source

Rebecca E. Garcia, Pamela E. Harris, Alex Moon, Aaron Ortiz, Lauren J. Quesada, Cynthia Marie Rivera Sánchez, Dwight Anderson Williams and Alexander N. Wilson, “The Defective Parking Space and Defective Kreweras Numbers”, arXiv:2405.14635 (2026).

Progress summary

Refreshed
Claimed progress

The conjecture remains unverified: a submitted argument claims a proof, but no independent public confirmation was found.

The 2024 paper introducing these numbers states the stable-range formula as a conjecture and explicitly solicits either a proof or a counterexample. No published or independently verified resolution was found.

Community submission (unverified)

A submitted proof argues that a marked cycle lemma, followed by a bijective counting argument, establishes the conjectured formula for all d≥0d\geq 0 and n≥m+d−1n\geq m+d-1, including its independence from nn in the stable range. The argument has not been independently verified.

Current status (as of August 2026): The conjecture is settled only by an unverified submitted proof claim; no independently confirmed proof or counterexample is recorded.

Sources

Solutions 1

ProofThis solution needs a summarySee full solutionHide full solution

Exact stable formula for every defective Kreweras number

Let m,n≥1m,n\geq1, let λ⊢m\lambda\vdash m have length kk, and put

μj=#{i:λi=j}(1≤j≤m).\mu_j=\#\{i:\lambda_i=j\} \qquad(1\leq j\leq m).

For a nondecreasing preference list x=(x1,…,xm)\mathbf{x}=(x_1,\ldots,x_m), define its predefect by

pdft⁡(x)=max⁡1≤i≤m(xi−i).\operatorname{pdft}(\mathbf{x}) =\max_{1\leq i\leq m}(x_i-i).

Following Definition 3.5 of the cited paper, Krew⁡d,n(λ)\operatorname{Krew}_{d,n}(\lambda) counts nondecreasing lists in [n+1]m[n+1]^m whose positive multiplicities, arranged in decreasing order, form λ\lambda and which satisfy

max⁡{pdft⁡(x),0}=d.\max\{\operatorname{pdft}(\mathbf{x}),0\}=d.

We prove that, for every d≥0d\geq0 and every n≥m+d−1n\geq m+d-1,

Krew⁡d,n(λ)=m+dkm+d 1m+d+1(m+d+1m+d+1−k,μ1,…,μm).\boxed{\displaystyle \operatorname{Krew}_{d,n}(\lambda) =\frac{m+dk}{m+d}\, \frac{1}{m+d+1} \binom{m+d+1} {m+d+1-k,\mu_1,\ldots,\mu_m}.}

In particular, the value is independent of nn throughout the entire conjectured stable range.

The marked cycle lemma

We use the following classical cycle lemma. Suppose that b1,…,bNb_1,\ldots,b_N are integers satisfying

bi≤1,∑i=1Nbi=r>0.b_i\leq1, \qquad \sum_{i=1}^N b_i=r>0.

Among the NN cyclic shifts, counted by their marked starting positions even when the word is periodic, exactly rr have every nonempty partial sum strictly positive.

For completeness, set

P0=0,Pj=∑i=1jbi,M=min⁡0≤j<NPj.P_0=0, \qquad P_j=\sum_{i=1}^j b_i, \qquad M=\min_{0\leq j<N}P_j.

For each integer qq with M≤q≤M+r−1M\leq q\leq M+r-1, let jq<Nj_q<N be the last index satisfying Pjq=qP_{j_q}=q. Such an index exists: after reaching its minimum, the walk eventually reaches PN=rP_N=r, and upward increments are at most 11. No subsequent value PiP_i with jq<i≤Nj_q<i\leq N can be at most qq; otherwise, before reaching r>qr>q, the walk would have to revisit qq, contradicting the choice of jqj_q. Every wrapped value satisfies

r+Pi≥r+M>q.r+P_i\geq r+M>q.

Consequently, the cyclic shift starting immediately after jqj_q has strictly positive partial sums. Conversely, if a shift starting after jj has this property, then jj is the last occurrence of PjP_j, and a minimum must occur no later than jj. Applying strict positivity to the corresponding wrapped minimum gives

M≤Pj<M+r.M\leq P_j<M+r.

Thus j=jqj=j_q for exactly one of the rr indicated levels. This proves the marked cycle lemma.

Cumulative enumeration by occupancy words

For d≥0d\geq0, let Fd(λ)F_d(\lambda) denote the number of nondecreasing lists having multiplicity partition λ\lambda and satisfying

pdft⁡(x)≤d.\operatorname{pdft}(\mathbf{x})\leq d.

Every such list satisfies

xi≤i+d,xm≤m+d.x_i\leq i+d, \qquad x_m\leq m+d.

Therefore, whenever n≥m+d−1n\geq m+d-1, the requirement xi∈[n+1]x_i\in[n+1] imposes no additional restriction. Put

N=m+d+1,r=d+1,αt=#{i:xi=t}(1≤t≤N).N=m+d+1, \qquad r=d+1, \qquad \alpha_t=\#\{i:x_i=t\} \quad(1\leq t\leq N).

The multiset of the αt\alpha_t consists of the kk parts of λ\lambda and N−kN-k zeros. In particular,

∑t=1Nαt=m,αN=0.\sum_{t=1}^N\alpha_t=m, \qquad \alpha_N=0.

Write

At=∑j=1tαj,St=At−t.A_t=\sum_{j=1}^t\alpha_j, \qquad S_t=A_t-t.

For 0≤t<N0\leq t<N, the inequalities xi≤i+dx_i\leq i+d for every ii are equivalent to

At≥t−d,or equivalentlySt≥−d>−r.A_t\geq t-d, \qquad\text{or equivalently}\qquad S_t\geq-d>-r.

Indeed, if At<t−dA_t<t-d, then the (t−d)(t-d)th list entry is greater than tt, violating its required upper bound. Conversely, if xi>i+dx_i>i+d, taking t=i+d<Nt=i+d<N gives At<i=t−dA_t<i=t-d. At the endpoint,

SN=m−N=−r.S_N=m-N=-r.

Now reverse the occupancy word and set

bj=1−αN+1−j(1≤j≤N).b_j=1-\alpha_{N+1-j} \qquad(1\leq j\leq N).

These integers satisfy bj≤1b_j\leq1 and ∑j=1Nbj=r\sum_{j=1}^N b_j=r. Moreover, their partial sums satisfy

∑j=1sbj=r+SN−s(1≤s≤N).\sum_{j=1}^s b_j =r+S_{N-s} \qquad(1\leq s\leq N).

Thus x\mathbf{x} has predefect at most dd exactly when every partial sum of the reversed word is strictly positive. Strict positivity of the first partial sum also forces αN=0\alpha_N=0, so no additional terminal restriction is needed.

There are

(NN−k,μ1,…,μm)\binom{N}{N-k,\mu_1,\ldots,\mu_m}

distinct occupancy words with the prescribed multiplicity multiset. The marked cycle lemma and double counting of pairs consisting of a word and a marked cyclic starting position therefore give

Fd(λ)=d+1m+d+1(m+d+1m+d+1−k,μ1,…,μm).\boxed{\displaystyle F_d(\lambda) =\frac{d+1}{m+d+1} \binom{m+d+1}{m+d+1-k,\mu_1,\ldots,\mu_m}.}

Counting marked starts makes this argument valid without any aperiodicity assumption, including when some parts of λ\lambda coincide.

Extraction of the exact predefect

Set F−1(λ)=0F_{-1}(\lambda)=0. For every d≥0d\geq0,

Krew⁡d,n(λ)=Fd(λ)−Fd−1(λ).\operatorname{Krew}_{d,n}(\lambda) =F_d(\lambda)-F_{d-1}(\lambda).

For d=0d=0, this immediately gives

Krew⁡0,n(λ)=1m+1(m+1m+1−k,μ1,…,μm).\operatorname{Krew}_{0,n}(\lambda) =\frac{1}{m+1} \binom{m+1}{m+1-k,\mu_1,\ldots,\mu_m}.

For d≥1d\geq1, the cumulative formula yields

Fd(λ)−Fd−1(λ)=(d+1)(m+d)!(m+d+1−k)!∏j=1mμj!−d(m+d−1)!(m+d−k)!∏j=1mμj!=(m+dk)(m+d−1)!(m+d+1−k)!∏j=1mμj!=m+dkm+d 1m+d+1(m+d+1m+d+1−k,μ1,…,μm),\begin{aligned} F_d(\lambda)-F_{d-1}(\lambda) &=\frac{(d+1)(m+d)!} {(m+d+1-k)!\prod_{j=1}^m\mu_j!} -\frac{d(m+d-1)!} {(m+d-k)!\prod_{j=1}^m\mu_j!}\\ &=\frac{(m+dk)(m+d-1)!} {(m+d+1-k)!\prod_{j=1}^m\mu_j!}\\ &=\frac{m+dk}{m+d}\, \frac{1}{m+d+1} \binom{m+d+1} {m+d+1-k,\mu_1,\ldots,\mu_m}, \end{aligned}

which is precisely the conjectured formula.

The stability threshold cannot be lowered uniformly. For λ=(1m)\lambda=(1^m), all entries are strictly increasing, so xi−ix_i-i is nondecreasing and exact predefect dd requires

xm=m+d.x_m=m+d.

There are

Krew⁡d,n(1m)=(m+d−1m−1)>0\operatorname{Krew}_{d,n}(1^m) =\binom{m+d-1}{m-1}>0

in the stable range, whereas none of these lists exists if n=m+d−2≥1n=m+d-2\geq1.

The parameter dd here is the predefect parameter in Definition 3.5, not the actual number of cars failing to park when m≠nm\neq n. The source assumes m,n≥1m,n\geq1; the displayed rational formula is not defined for the empty-partition case m=d=0m=d=0.

This proves Conjecture 3.15 of Rebecca E. Garcia, Pamela E. Harris, Alex Moon, Aaron Ortiz, Lauren J. Quesada, Cynthia Marie Rivera Sánchez, Dwight Anderson Williams II, and Alexander N. Wilson, The defective parking space and defective Kreweras numbers, arXiv:2405.14635v4, published in Discrete Mathematics 349 (2026), Article 115164, doi:10.1016/j.disc.2026.115164.