The recursive profile conjecture for distance games on complete bipartite graphs

About 5 years old · traced to

Let Km,nK_{m,n} be the complete bipartite graph with part sizes mm and nn, and let P\textscCol,Km,n(1)P_{\textsc{Col},K_{m,n}}(1) denote the number of positions in Col (equivalently, by the bipartite-graph theorem, in Snort) on this graph. The initial terms are those in the stated table, and let cmc_m be given by OEIS sequence A260217, with c2=4c_2=4, c3=24c_3=24, c4=100c_4=100, c5=360c_5=360, and c6=1204c_6=1204. Recursive profile conjecture. The numbers of positions satisfy

P\textscCol,Km,n(1)=5P\textscCol,Km,n−1(1)−6P\textscCol,Km,n−2(1)+cm.P_{\textsc{Col},K_{m,n}}(1)=5P_{\textsc{Col},K_{m,n-1}}(1)-6P_{\textsc{Col},K_{m,n-2}}(1)+c_m.

This conjecture is based on computed data for complete bipartite graphs and predicts a uniform second-order recurrence in nn, with an mm-dependent inhomogeneous term. Its resolution would provide formulas for the position counts of Col and Snort on this family of bipartite boards.

References

Primary source

Svenja Huntemann and Lexi A. Nash, “The Polynomial Profile of Distance Games on Paths and Cycles”, arXiv:2111.09349 (2021).

Progress summary

Refreshed
Claimed progress

The conjecture remains unproved in public sources, but an unverified submission claims a stronger explicit formula that would settle it.

Huntemann and Nash formulated the conjecture for position counts of Col, equivalently Snort, on complete bipartite graphs. It predicts a uniform recurrence in one part size, supported by computed data but not proved in the published account.

Known results

  • The conjecture is stated as Conjecture 4.74.7, with the constants cmc_m linked to OEIS A260217.
  • Col and Snort have equal position counts on every bipartite graph.
  • The source identifies proving the recurrence and finding complete-bipartite generating functions as open goals.

Community submission (unverified)

A submitted proof argues that the recurrence holds for every m,nm,n and claims full bivariate profiles for both games. For Col it gives an explicit decomposition using A=1+x+yA=1+x+y, B=1+xB=1+x, and C=1+yC=1+y; this would imply the stated recurrence, but the argument has not been independently verified.

Current status (as of August 2026): The published source records the conjecture as open, while an unverified community submission claims a stronger proof and explicit profiles.

Sources

Solutions 1

ProofThis solution needs a summarySee full solutionHide full solution

Complete bivariate profiles and the conjectured recurrence for Col and Snort

Svenja Huntemann and Lexi A. Nash formulate Conjecture 1 in their published paper The Polynomial Profile of Distance Games on Paths and Cycles, INTEGERS 22 (2022), Article G4. For the complete bipartite graph Km,nK_{m,n}, their conjecture concerns the numbers of legal positions in Col and Snort, with an inhomogeneous recurrence whose constants are identified with OEIS A260217. The paper defines a profile by counting legal positions without assuming alternating play. We prove the conjecture for every m,nm,n and, more strongly, determine the full bivariate polynomial profiles of both games.

Let UU and VV denote the bipartition classes, with ∣U∣=m|U|=m and ∣V∣=n|V|=n. Every vertex is either unoccupied, blue, or red. Give a position with bb blue vertices and rr red vertices the weight xbyrx^b y^r, and write Pm,nCol(x,y)P^{\mathrm{Col}}_{m,n}(x,y) and Pm,nSnort(x,y)P^{\mathrm{Snort}}_{m,n}(x,y) for the sums of these weights over the legal positions. Set

A=1+x+y,B=1+x,C=1+y.(1)A=1+x+y,\qquad B=1+x,\qquad C=1+y. \tag{1}

In Col, adjacent vertices cannot have the same color. In Snort, adjacent vertices cannot have opposite colors.

1. Exact bivariate Col profile. Classify a legal position according to the set of colors appearing in UU.

  • If UU is entirely unoccupied, the vertices of VV can be colored independently, contributing AnA^n.
  • If blue is the unique color appearing in UU, the nonempty blue subset contributes Bm−1B^m-1, and each vertex of VV is either unoccupied or red. The contribution is (Bm−1)Cn(B^m-1)C^n.
  • If red is the unique color appearing in UU, the contribution is (Cm−1)Bn(C^m-1)B^n.
  • If both colors appear in UU, every vertex of VV must be unoccupied. Inclusion-exclusion gives the contribution Am−Bm−Cm+1A^m-B^m-C^m+1.

These four cases are disjoint and exhaustive. Therefore

Pm,nCol(x,y)=An+(Bm−1)Cn+(Cm−1)Bn+Am−Bm−Cm+1.(2)P^{\mathrm{Col}}_{m,n}(x,y) =A^n+(B^m-1)C^n+(C^m-1)B^n +A^m-B^m-C^m+1. \tag{2}

Equivalently, in symmetric form,

Pm,nCol(x,y)=Am+An+BmCn+CmBn−Bm−Cm−Bn−Cn+1.(3)P^{\mathrm{Col}}_{m,n}(x,y) =A^m+A^n+B^mC^n+C^mB^n -B^m-C^m-B^n-C^n+1. \tag{3}

The argument includes the degenerate cases m=0m=0 or n=0n=0.

2. Exact bivariate Snort profile. The same four-way classification applies. If only blue appears in UU, vertices of VV may be blue or unoccupied, contributing (Bm−1)Bn(B^m-1)B^n. If only red appears in UU, the corresponding contribution is (Cm−1)Cn(C^m-1)C^n. The empty and two-color cases are unchanged. Consequently,

Pm,nSnort(x,y)=An+(Bm−1)Bn+(Cm−1)Cn+Am−Bm−Cm+1.(4)P^{\mathrm{Snort}}_{m,n}(x,y) =A^n+(B^m-1)B^n+(C^m-1)C^n +A^m-B^m-C^m+1. \tag{4}

Thus

Pm,nSnort(x,y)=Am+An+Bm+n+Cm+n−Bm−Cm−Bn−Cn+1.(5)P^{\mathrm{Snort}}_{m,n}(x,y) =A^m+A^n+B^{m+n}+C^{m+n} -B^m-C^m-B^n-C^n+1. \tag{5}

In general, (3) and (5) are different bivariate polynomials. However, putting x=y=tx=y=t makes B=C=1+tB=C=1+t, so both become the same univariate position polynomial:

Pm,n(t)=(1+2t)m+(1+2t)n+2(1+t)m+n−2(1+t)m−2(1+t)n+1.(6)P_{m,n}(t) =(1+2t)^m+(1+2t)^n+2(1+t)^{m+n} -2(1+t)^m-2(1+t)^n+1. \tag{6}

This equality also agrees with the known bipartite Col--Snort profile equivalence recalled in the source paper: interchanging blue and red on one bipartition class converts one legality rule into the other while preserving the total number of occupied vertices. It need not preserve the separate color counts, explaining the distinction between (3) and (5).

3. Complete enumeration and the conjectured recurrence. Evaluating (6) at t=1t=1 gives, for all m,n≥0m,n\geq 0,

Fm,n=Pm,n(1)=3m+3n+2m+n+1−2m+1−2n+1+1.(7)F_{m,n}=P_{m,n}(1) =3^m+3^n+2^{m+n+1}-2^{m+1}-2^{n+1}+1. \tag{7}

For example,

F1,1=7,F2,2=35,F2,3=77,F3,3=151,F4,4=611.(8)F_{1,1}=7,\qquad F_{2,2}=35,\qquad F_{2,3}=77, \qquad F_{3,3}=151,\qquad F_{4,4}=611. \tag{8}

For fixed mm, rearrange (7) as

Fm,n=3n+2(2m−1)2n+Dm,Dm=3m−2m+1+1.(9)F_{m,n}=3^n+2(2^m-1)2^n+D_m, \qquad D_m=3^m-2^{m+1}+1. \tag{9}

The operator fn↦fn−5fn−1+6fn−2f_n\mapsto f_n-5f_{n-1}+6f_{n-2} annihilates both 3n3^n and 2n2^n, and sends the constant DmD_m to 2Dm2D_m. Hence, for every m≥0m\geq 0 and n≥2n\geq 2,

Fm,n=5Fm,n−1−6Fm,n−2+cm,(10)F_{m,n}=5F_{m,n-1}-6F_{m,n-2}+c_m, \tag{10}

where

cm=2Dm=2⋅3m−2m+2+2.(11)c_m=2D_m=2\cdot 3^m-2^{m+2}+2. \tag{11}

The first nonzero constants are precisely

c2=4,c3=24,c4=100,c5=360,c6=1204.(12)c_2=4,\qquad c_3=24,\qquad c_4=100, \qquad c_5=360,\qquad c_6=1204. \tag{12}

The preexisting OEIS formula is a(j)=2⋅3j−1−2j+1+2a(j)=2\cdot 3^{j-1}-2^{j+1}+2. Its correct index shift in the conjecture is therefore

cm=a(m+1)=A260217(m+1).(13)c_m=a(m+1)=\mathrm{A260217}(m+1). \tag{13}

The closed form also supplies the initial conditions

Fm,0=3m,Fm,1=3m+2m+1.(14)F_{m,0}=3^m,\qquad F_{m,1}=3^m+2^{m+1}. \tag{14}

Thus (10)--(14) establish every coefficient and initial condition in Conjecture 1, with no restriction on the sizes of the two bipartition classes.

4. Stronger polynomial recurrence and generating function. The same argument works before specialization. Let

a=1+2t,b=1+t,Dm(t)=am−2bm+1.(15)a=1+2t,\qquad b=1+t,\qquad D_m(t)=a^m-2b^m+1. \tag{15}

Equation (6) becomes

Pm,n(t)=an+2(bm−1)bn+Dm(t).(16)P_{m,n}(t)=a^n+2(b^m-1)b^n+D_m(t). \tag{16}

Since (1−a)(1−b)=2t2(1-a)(1-b)=2t^2, it follows that, for n≥2n\geq 2,

Pm,n(t)=(2+3t)Pm,n−1(t)−(1+3t+2t2)Pm,n−2(t)+2t2((1+2t)m−2(1+t)m+1).(17)P_{m,n}(t) =(2+3t)P_{m,n-1}(t) -(1+3t+2t^2)P_{m,n-2}(t) +2t^2\bigl((1+2t)^m-2(1+t)^m+1\bigr). \tag{17}

Moreover, the entire fixed-mm profile sequence has the explicit ordinary generating function

∑n≥0Pm,n(t)zn=11−(1+2t)z+2((1+t)m−1)1−(1+t)z+(1+2t)m−2(1+t)m+11−z.(18)\sum_{n\geq 0}P_{m,n}(t)z^n =\frac{1}{1-(1+2t)z} +\frac{2((1+t)^m-1)}{1-(1+t)z} +\frac{(1+2t)^m-2(1+t)^m+1}{1-z}. \tag{18}

In particular, the conjectured numerical recurrence is the specialization t=1t=1 of the stronger complete polynomial identity (17), while (3) and (5) determine the separate two-color profiles themselves.