Integer-root conjecture for total domination polynomials

About 10 years old · traced to

Let G=(V,E)G=(V,E) be a simple graph, let

Dt(G;x)=∑i=1ndt(G,i)xiD_t(G;x)=\sum_{i=1}^{n} d_t(G,i)x^i

be its total domination polynomial, where nn is the order of GG and dt(G,i)d_t(G,i) counts the total dominating sets of cardinality ii. A root of Dt(G,x)D_t(G,x) is called a total domination root. The integer-root conjecture. If rr is an integer root of Dt(G,x)D_t(G,x), then

r∈{−3,−2,−1,0}.r\in\{-3,-2,-1,0\}.

This proposes a restriction on the possible integer total domination roots. Earlier work established the narrower possibilities −2-2 and 00 for domination-polynomial roots under a minimum-degree hypothesis, but the corresponding unrestricted claim for total domination polynomials remains open.

References

Primary source

Saeid Alikhani and Nasrin Jafari, “On the roots of total domination polynomial of graphs”, arXiv:1605.02222 (2016).

Progress summary

Refreshed
Claimed progress

An unverified submission claims the conjecture is false by constructing a graph whose total-domination polynomial has the forbidden integer root −4-4.

Alikhani and Jafari formulated the conjecture in 2016: every integer root of a total-domination polynomial should lie among {−3,−2,−1,0}\{-3,-2,-1,0\}. The unrestricted statement remains distinct from corresponding results for ordinary domination polynomials.

Known results

  • Alikhani and Jafari (2016): if the minimum degree satisfies δ(G)≥2n/3\delta(G)\geq 2n/3, every integer root lies in {−3,−2,−1,0}\{-3,-2,-1,0\}.
  • Alikhani and Jafari (2019): established further restrictions on the number and location of total-domination roots, but did not resolve the integer-root conjecture.
  • Related ordinary-domination work gives an order-3333 graph with root −4-4, but does not itself address total domination.

Community submission (unverified)

A submitted argument claims a connected simple bipartite graph with 6666 vertices and 105105 edges, minimum degree 22, whose total-domination polynomial has root −4-4 with multiplicity exactly 22. It proposes transferring an ordinary-domination counterexample through a closed-neighborhood bipartite double-graph construction; no independent verification was found.

Current status (as of August 2026): The minimum-degree case is proved, while the unrestricted conjecture remains open because the submitted −4-4 counterexample is unverified.

Sources

Solutions 1

CounterexampleThis solution needs a summarySee full solutionHide full solution

A connected bipartite counterexample to the integer-root conjecture for total domination polynomials

Result. The conjecture is false. There exists a connected, simple, bipartite graph on 66 vertices and 105 edges, with minimum degree 2, whose total domination polynomial has the integer root −4-4 with multiplicity exactly two.

The statement being disproved is Conjecture 3.10 of S. Alikhani and N. Jafari, On the roots of total domination polynomial of graphs, arXiv:1605.02222; see also the published version by N. Jafari and S. Alikhani, Journal of Discrete Mathematical Sciences and Cryptography 23 (2020), 795–807, doi:10.1080/09720529.2019.1616908. The ordinary domination counterexample used as an ingredient is due to S. Alikhani and M. Griswold, On the Integer Domination Root Conjecture, arXiv:2608.00109, Theorem 2.1. Their theorem concerns ordinary domination polynomials; the argument below transfers it to the distinct conjecture about total domination.

1. A universal ordinary-to-total domination identity

Let H=(V,E)H=(V,E) be any finite simple graph. A subset S⊆VS\subseteq V dominates HH precisely when

NH[v]∩S≠∅(v∈V),N_H[v]\cap S\neq\varnothing\qquad(v\in V),

where NH[v]={v}∪NH(v)N_H[v]=\{v\}\cup N_H(v) is the closed neighborhood. Its ordinary domination polynomial is

D(H;x)=∑S dominates Hx∣S∣.D(H;x)=\sum_{S\text{ dominates }H}x^{|S|}.

Construct the bipartite closed-neighborhood double graph B(H)B(H) with two disjoint copies VLV_L and VRV_R of VV, inserting the edge uLvRu_Lv_R exactly when

u=vor{u,v}∈E.u=v\qquad\text{or}\qquad\{u,v\}\in E.

In particular, the diagonal pairs vLvRv_Lv_R are ordinary edges between distinct vertices, not loops. The graph is simple, bipartite, and has

∣V(B(H))∣=2∣V∣,∣E(B(H))∣=∣V∣+2∣E∣,deg⁡B(H)(vL)=deg⁡B(H)(vR)=1+deg⁡H(v).|V(B(H))|=2|V|, \qquad |E(B(H))|=|V|+2|E|, \qquad \deg_{B(H)}(v_L)=\deg_{B(H)}(v_R)=1+\deg_H(v).

Every T⊆V(B(H))T\subseteq V(B(H)) can be written uniquely as

T=(SL)L∪(SR)R,SL,SR⊆V.T=(S_L)_L\cup(S_R)_R, \qquad S_L,S_R\subseteq V.

A left vertex vLv_L has a neighbor in TT if and only if

NH[v]∩SR≠∅,N_H[v]\cap S_R\neq\varnothing,

and a right vertex vRv_R has a neighbor in TT if and only if

NH[v]∩SL≠∅.N_H[v]\cap S_L\neq\varnothing.

Consequently, TT is a total dominating set of B(H)B(H) if and only if SLS_L and SRS_R are two independently chosen ordinary dominating sets of HH. Preservation of their cardinalities yields the polynomial identity

Dt(B(H);x)=D(H;x)2.\boxed{\displaystyle D_t(B(H);x)=D(H;x)^2.}

Thus every ordinary domination root of every finite graph is a total domination root of a bipartite graph, with twice its original multiplicity. If HH is connected, then B(H)B(H) is connected: each edge uv∈E(H)uv\in E(H) gives the path uL−vR−vLu_L-v_R-v_L, and every right vertex is attached to its matching left vertex.

2. The explicit 33-vertex ingredient

Take the connected graph H=G33H=G_{33} of Alikhani and Griswold, with vertex set {0,1,…,32}\{0,1,\ldots,32\} and edge set

E(H)={{0,1},{0,3},{0,4},{0,12},{1,2},{1,12},{1,18},{2,3},{2,6},{2,8},{2,10},{2,18},{2,21},{3,21},{3,24},{4,5},{6,7},{8,9},{10,11},{12,13},{12,14},{12,15},{12,16},{12,17},{18,19},{18,20},{21,22},{21,23},{24,25},{24,26},{26,27},{27,28},{28,29},{28,31},{29,30},{31,32}}.\begin{aligned} E(H)=\{&\{0,1\},\{0,3\},\{0,4\},\{0,12\},\{1,2\},\{1,12\},\{1,18\},\\ &\{2,3\},\{2,6\},\{2,8\},\{2,10\},\{2,18\},\{2,21\},\{3,21\},\{3,24\},\\ &\{4,5\},\{6,7\},\{8,9\},\{10,11\},\{12,13\},\{12,14\},\{12,15\},\\ &\{12,16\},\{12,17\},\{18,19\},\{18,20\},\{21,22\},\{21,23\},\\ &\{24,25\},\{24,26\},\{26,27\},\{27,28\},\{28,29\},\{28,31\},\\ &\{29,30\},\{31,32\}\}. \end{aligned}

Their Theorem 2.1 establishes the exact factorization

D(H;x)=x11(x+2)(x+4)(x2+3x+1)Q18(x),D(H;x)=x^{11}(x+2)(x+4)(x^2+3x+1)Q_{18}(x),

where

Q18(x)=x18+24x17+269x16+1871x15+9049x14+32314x13+88298x12+188808x11+320426x10+435134x9+474341x8+414314x7+287750x6+156593x5+65228x4+20061x3+4295x2+573x+36.\begin{aligned} Q_{18}(x)={}&x^{18}+24x^{17}+269x^{16}+1871x^{15}+9049x^{14} +32314x^{13}+88298x^{12}\\ &+188808x^{11}+320426x^{10}+435134x^9+474341x^8 +414314x^7\\ &+287750x^6+156593x^5+65228x^4+20061x^3 +4295x^2+573x+36. \end{aligned}

For completeness, this factorization can also be checked directly from the displayed edge set using a finite mathematical identity. Let CC be the set of non-leaf vertices, and let tvt_v count the leaves adjacent to v∈Cv\in C. Here

C={0,1,2,3,4,6,8,10,12,18,21,24,26,27,28,29,31},C=\{0,1,2,3,4,6,8,10,12,18,21,24,26,27,28,29,31\},

and, in that order,

(tv)v∈C=(0,0,0,0,1,1,1,1,5,2,2,1,0,0,0,1,1).(t_v)_{v\in C}=(0,0,0,0,1,1,1,1,5,2,2,1,0,0,0,1,1).

For S⊆CS\subseteq C, every leaf adjacent to an unchosen vertex must itself be chosen; the leaves adjacent to a chosen vertex are free. An unchosen non-leaf vertex with no adjacent leaf must have a neighbor in SS. Therefore the complete ordinary domination polynomial is exactly

D(H;x)=∑S⊆CNH(v)∩S≠∅for every v∈C∖S with tv=0x∣S∣∏v∈S(1+x)tv∏v∈C∖Sxtv.D(H;x)= \sum_{\substack{S\subseteq C\\ N_H(v)\cap S\neq\varnothing\;\text{for every }v\in C\setminus S\text{ with }t_v=0}} x^{|S|} \prod_{v\in S}(1+x)^{t_v} \prod_{v\in C\setminus S}x^{t_v}.

Expanding this identity gives precisely the factorization above. In particular,

D(H;−4)=0,Q18(−4)=26236000≠0,D(H;-4)=0, \qquad Q_{18}(-4)=26236000\neq0,

so the root −4-4 of D(H;x)D(H;x) is simple.

3. The total-domination counterexample

Apply the universal identity to B=B(H)B=B(H). Since ∣V(H)∣=33|V(H)|=33, ∣E(H)∣=36|E(H)|=36, and δ(H)=1\delta(H)=1, this is a connected simple bipartite graph satisfying

∣V(B)∣=66,∣E(B)∣=33+2⋅36=105,δ(B)=2.|V(B)|=66, \qquad |E(B)|=33+2\cdot36=105, \qquad \delta(B)=2.

Its total domination polynomial is

Dt(B;x)=x22(x+2)2(x+4)2(x2+3x+1)2Q18(x)2.\boxed{\displaystyle D_t(B;x)=x^{22}(x+2)^2(x+4)^2(x^2+3x+1)^2Q_{18}(x)^2. }

Hence −4-4 is an integer total domination root of multiplicity exactly two. Since

−4∉{−3,−2,−1,0},-4\notin\{-3,-2,-1,0\},

the integer-root conjecture for total domination polynomials is false, even when restricted to connected bipartite graphs of minimum degree at least two. This does not contradict the source's separate result for graphs satisfying the much stronger minimum-degree hypothesis δ(G)≥2∣V(G)∣/3\delta(G)\geq 2|V(G)|/3.