Chromatic Coloring of Distance Graphs V

Yegnanarayanan Venkataraman1, George Barnabas1, Bryan Freyberg2
1Department of Mathematics, Kalasalingam Academy of Research and Education, Krishnankoil, 626126, Tamilnadu, India
2Department of Mathematics and Statistics, University of Minnesota Duluth, 1117 University Drive, Duluth, 55811, MN, USA

Abstract

Let G=(V,E) be any graph. If there exists an injection f:VZ, such that |f(u)f(v)| is prime for every uvE, then we say G is a prime distance graph (PDG). The problem of characterizing the family of all prime distance graphs (PDGs) with chromatic number 3 or 4 is challenging. In the fourth part of this series of articles, we determined which fans are PDGs and which wheels are PDGs. In addition, we showed: (1) a chain of n mutually isomorphic PDGs is a PDG, and (2) the Cartesian product of a PDG and a path is a PDG. In this part of the series, we improve (1) by showing that there exists a chain of n arbitrary PDGs which is a PDG. We also show that the following graphs are PDGs: (a) any graph with at most three cycles, (b) the one-point union of cycles, and (c) a family of graphs consisting of paths with common end vertices.

Keywords: Chromatic number, Distance graph, Prime distance graph, Prime distance labeling

1. Introduction

By a graph G we mean a pair G=(V,E), where V is any arbitrary set and E is a subset of the set of all 2-element subsets of V. The elements of V and E are called, respectively, the vertices and edges of G. If |V(G)|< then G is said to be finite. The graphs considered here are simple with finite or countably infinite vertex sets. By a distance graph Z(D), we mean a graph, whose vertex set is the set of integers Z, and two vertices u and v form an edge if and only if |uv|D. If D=P, the set of primes, then we call Z(P), the prime distance graph.

Given a graph G with vertex set V and edge set E and a set X of colors, an ordinary coloring is an assignment of elements of X to the elements of V in any order. A coloring f:VX is called proper if f(u)f(v) for all uvE. Such a coloring is called a proper k-coloring if |{f(u):uV(G)}|=k. The least k for which G has a proper k-coloring is called the chromatic number of G and is denoted by χ(G). One can consult [1-15] for more on this topic.

Eggleton, Erdős, and Skilton attempted the task of finding the chromatic number of the prime distance graph Z(D) where DP. They found that χ(Z(P))=4 [9]. The problem of computing χ(Z(D)) where DP is still open. Voigt and Walther in [16], Kemnitz and Kolberg in [2], and Yegnanarayanan et al. [17-21] have contributed more on this topic.

Observe that f:ZZ4Z where f(v)=vmod4 is a 4-coloring of Z(D) for any DP. This is because if f(u)=f(v), then uv(mod4) and |uv|0(mod4). Thus |uv|P and χ(Z(D))4. As χ(G) is a monotonic function, if D1D2 then χ(D1)χ(D2). So, if DP then the tough task is to classify Z(D) as per its chromatic number.

We say a graph G is of Class i if and only if χ(G)=i. It is obvious that Z() is the only distance graph that is Class 1, and Z(D) is Class 2 whenever D is a singleton set. It is already known from the results available in the literature that if |D| 2 then Z(D) is Class 2 if and only if 2 D. Similarly, if 2 D and 3 D, then Z(D) is Class 3. If |D|=3 with {2,3,5} D then Z(D) is Class 4 [6]. Also, if D={2,3,p} with p5, then Z(D) is Class 3. If 2 or 3 is contained in a subset DP, then χ(D)=3 or 4. It is an unsolved problem to classify these sets D whether they belong to Class 3 or 4 (see [6,9,22,23]).

Let G=(V,E) be any graph. If there exists an injection f:VZ, such that |f(u)f(v)| is prime for every uvE, then we say f a prime distance labeling of G and G is a prime distance graph. The value of |f(u)f(v)| is referred to as the induced label of the edge uv. For the sake of brevity, if this induced label is prime, we say the edge uv is prime.

It is known that a bipartite graph is a prime distance graph [24]. So every graph G with χ(G)=2 is a prime distance graph. In particular, trees are prime distance graphs. It is still open to find a characterization for graphs with χ(G)=3 or 4. In this paper we demonstrate the existence of several new collections of families of prime distance graphs with χ(G)=3 or 4.

Laison, Starr, and Walker showed that not only is every bipartite graph G a prime distance graph, but a prime distance labeling f can be found such that f(v)=0 for any particular vertex vV(G) and f(u)0 for all uv [24]. They showed the same is true for cycles.

Theorem 1 (Laison, et al., [24]). Let G be a bipartite graph or a cycle and vV(G) be any vertex. There exists a prime distance labeling f:V(G)N such that f(v)=0.

Next we share some observations that are used in the constructions throughout this paper. The following three observations are from [25].

Observation 2. Let G be any graph. An injection f:V(G)Z is a prime distance labeling of G if and only if f is a prime distance labeling of G.

Observation 3. If G is a prime distance graph and H is a subgraph of G, then H is also a prime distance graph.

Observation 4. Let G be any graph, f:V(G)Z an injection, and c any real number. The function f is a prime distance labeling of G if and only if f+c is a prime distance labeling of G.

We call the operation from the previous observation lifting f by c.

Corollary 1. If G1 and G2 are both prime distance graphs, then the disjoint union G1G2 is also a prime distance graph.

Proof. Let f and g be prime distance labelings of G1 and G2 respectively and c be the maximum label of f. Label the vertices by (v)={f(v)if \;vV(G1)g(v)+cif \;vV(G2) Since is injective, it is a prime distance labeling of G1G2 by Observation 4. ◻

Let Pn=u1u2u3un be the path on n vertices and Cn=u1u2u3unu1 be the cycle on n vertices.

2. New Results

Lemma 1. Let G be a prime distance graph. Then a graph G+ obtained from G by adding a pendant edge to G is also a prime distance graph.

Proof. Let V(G+)=V(G){u} and uv be the pendant edge of G+ where vV(G). Suppose f is a prime distance labeling of G and p is the smallest prime such that p>max{f(x):xV(G)}. Then a prime distance labeling of G+ is (x)={f(x)if \;xuf(v)+pif \;x=u since is clearly an injection and the induced edge labels are prime. ◻

Notice that Lemma 1 can be repeatedly applied to make larger graphs from a given prime distance graph G0. Let G be such a graph and vV(G) and vV(G0) such that deg(v)=1. Then it is possible to construct the prime distance labeling f from repeatedly applying Lemma 1 such that f(v)>f(u) for every uV(G). This will be important in the proof of Theorem 6.

2.1 Graphs Containing two or Less Cycles

A graph which contains no cycles is necessarily bipartite, and thus a prime distance graph by Theorem 1. A graph G is unicyclic if it contains exactly one cycle. If the cycle contained by G is even, then G belongs to Class 2. Otherwise, G belongs to Class 3. Our next result shows all unicyclic graphs are prime distance graphs.

Theorem 5. Every unicyclic graph G is a prime distance graph. Furthermore, for any vertex v lying on the cycle of G, there exists a prime distance labeling f:V(G)N such that f(v)=0.

Proof. Let G be a unicyclic graph that contains t connected components. If t=1, the vertices of G on the cycle may be labeled by Theorem 1. The result now follows by inductively applying Lemma 1.

On the other hand, if t2, label the connected component containing the cycle as in the previous case. The remaining components constitute a bipartite subgraph H of G. Therefore, H admits a distance prime labeling f:V(H)Z which may be suitably lifted by Observation 4 to obtain the result. ◻

A graph G is bicyclic if it contains exactly two cycles. If both cycles contained in G are even, then G belongs to Class 2. Otherwise, G belongs to Class 3.

Theorem 6. Every bicyclic graph is a prime distance graph.

Proof. Let G be a bicyclic graph containing cycles Cm and Cn. We establish two cases based on the structure of G.
Cm and Cn are not in the same connected component.

In this case, G is the disjoint union of two unicyclic graphs. Therefore, the prime distance labeling exists by Theorem 5 and Corollary 1.

Cm and Cn belong to the same connected component.

We may assume G is the 1-point intersection of two unicyclic subgraphs H1 and H2 with common vertex v such that Cm is contained in H1, Cn is contained in H2, degH1(v)=1, and v is contained in Cn. By Theorem 5 and observation 4 a nonnegative prime distance labeling f of H1 exists such that f(v)=c>f(u) for all uv. Lifting f by c gives a prime distance labeling f of H1 such that f(v)=0 and f(u)0 for all uV(H1).

Let g be a nonnegative prime distance labeling of H2 such that g(v)=0. Such a labeling exists by Theorem 5. We combine the labelings f and g to produce a labeling as follows. Let (u)={f(u)if \;uV(H1)g(u)if \;uV(H2) It is clear that is injective since f(u)0 for all uV(H1), g(u)0 for all uV(H2), and V(H1)V(H2)={v} and f(v)=g(v)=0. Further, every edge is prime since f and g are prime distance labelings. ◻

2.2 Graphs Containing more than Two Cycles

The graph of k edge-disjoint cycles Cn1,Cn2,,Cnk glued together at a common vertex is called a one point union of cycles. See Figure 1 for an example. Before investigating this family of graphs, we state a conjecture of de Polignac.

Conjecture 1 (de Polignac, 1849). For every positive integer n, there are infinitely many primes that differ by 2n.

Notice that the famous Twin Prime Conjecture corresponds to the case n=1 in de Polignac’s Conjecture. Laison, Starr, and Walker showed that the one point union of n 3-cycles is a prime distance graph if and only if the Twin Prime Conjecture is true [24]. We prove some analogous but broader results using de Polignac’s Conjecture next.

Theorem 7. If de Polignac’s Conjecture is true, then the one point union of cycles Cn1,Cn2,,Cnk is a prime distance graph for any ni3 and k1.

Proof. Let pi and pi+2(ni2) be mutually distinct primes for i=1,2,,k such that pi+1pi+2(ni2). If de Polignac’s Conjecture is true, we may find such primes. Label the vertices of each cycle Cni by 0,pi,pi+2,pi+4,,pi+2(ni2),0 consecutively around the cycle such that the vertex labeled 0 is common to all cycles.

The lengths of the edges in Cni form the set {2,pi,pi+2(ni2)}. Since every member of this set is prime and the labeling described is injective, we have completed the proof. ◻

A graph is tricyclic if it contains exactly 3 cycles. If a graph G is tricyclic, then the Class of G is 2 if all three of the cycles are even, and is 3 otherwise. For any n4, a 1-chorded n-cycle is a tricyclic graph with vertex set V={u1,u2,,un} and edge set E={u1u2,u2u3,,unu1}{u1uc} for some c{2,n}. The edge u1uc is called the chord.

For practical reasons, the next theorem only requires the truth of a much weaker version of Polignac’s conjecture: the existence of a prime gap of size 2n for sufficiently large primes.

Theorem 8. If de Polignac’s Conjecture is true, then every 1-chorded n-cycle is a prime distance graph.

Proof. Label the cycle u1ucuc+1unu1 using any prime distance labeling f of Cnc+2 such that f(u1)=0 and f(uc)=2 (see Theorem 3 in [24] for example). Let q=max{f(ui):i=1,c,c+1,,n} and label the cycle u1u2ucu1 in order by 0,p,p+2,p+4,,p+2(c3),2,0 for any p>q such that both p and p+2(c4) are both prime. Such a pair can be found if Conjecture 1 is true. Since f is a prime distance labeling, we need only check that the edges on the cycle u1u2ucu1 are all prime. Since these labels all belong to the set {2,p,p+2(c4)}, we have proven the theorem. ◻

Theorem 9. Every disconnected tricyclic graph is a prime distance graph.

Proof. If G is a disconnected tricyclic graph, then it is either the disjoint union of three unicyclic graphs or a bicyclic graph with a unicyclic graph. Therefore, the result follows from Theorems 5 and 6 and Corollary 1. ◻

Theorem 10. If de Polignac’s Conjecture is true, then every connected tricyclic graph is a prime distance graph.

Proof. Let G be a connected tricyclic graph. Then G either contains a 1-chorded cycle or the one point union of three cycles. Therefore, the labeling can be carried out by Theorems 8 or 7, respectively. Labeling the remaining vertices in V(G) by Lemma 1 concludes the proof. ◻

2.2.1 Graphs containing more than three cycles

For any set of graphs G={G1,G2,,Gn} and integer n2, we define an n-chain graph, Gn as any edge-disjoint union of G1,G2,,Gn such that |V(Gi)V(Gi+1)|=1 for every i=1,2,,n1 and |V(Gi)V(Gj)|=0 otherwise. Notice that two n-chains of the same set of n graphs are not necessarily isomorphic. The authors showed that if GiG for all i=1,2,,n where G is a prime distance graph, then there exists a corresponding n-chain of G that is a prime distance graph [25]. Here, we prove a much stronger result.

Theorem 11. Let G={G1,G2,,Gn} be any set of prime distance graphs. Then there exists an n-chain Gn that is a prime distance graph.

Proof. Let fi:V(Gi)N be a prime distance labeling of Gi and uiV(Gi) be a vertex such that fi(ui)=0. We can make this assumption by Observation 4. Denote by viV(Gi) the vertex with the largest label, ρi. Form an n-chain G by gluing vi to ui+1 for 1in1. Then it is easy to see that the labeling f(x)={f1(x) if xV(G1)fi(x)+ρi1 if xV(Gi) and 2in is a prime distance labeling of G. ◻

One possible generalization of the 1-chorded graphs can be defined as follows. Let {n1,n2,,nk} with 0n1n2nk be any set of k non-decreasing integers. We define the rainbow graph R(n1,n2,,nk) as the union of k paths of the form ux1ix2ixniiv for 1ik. Since we are only considering simple graphs, we require n21. Simply put, R(n1,n2,,nk) is isomorphic to the graph obtained by gluing together the left endpoints and gluing together the right endpoints of k paths. The lengths of these paths may differ. We call the vertices u and v the rainbow ends. Notice that R(0,n2,n3) is isomorphic to a 1-chorded Cn2+n3+2. See Figure 2 for an example of R(1,3,5).

Theorem 12. If de Polignac’s Conjecture is true, then every rainbow graph R(n1,n2,,nk) is a prime distance graph.

Proof. For 1ik, let pi be a prime such that pi+2(ni2) is also prime. If de Polignac’s Conjecture is true, not only does such a set of primes exist, but we may further assume that the primes are all distinct (pi may be chosen so that pi>pi1+2(ni11)). For 1ik, label the path ux1ix2ixniiv with 0,pi,pi+2,,pi+2(ni1),2 in order. It is easy to check that the induced edge labels comprise the set {pi,pi+2(ni2):1ik}{2}. Since every member of this set is prime, the proof is complete. ◻

Let R={R1,R2,,Rn} be a set of n rainbow graphs. We define the rainbow range RR(R1,R2,,Rn) as the graph obtained by chaining together the rainbows R1,R2,,Rn by gluing the v vertex of Ri to the u vertex of Ri+1 for 1in1. Notice that the next result does not necessarily follow from Theorem 11.

Theorem 13. If de Polignac’s Conjecture is true, then every rainbow range RR(R1,R2,,Rn) is a prime distance graph for any n1.

Proof. Let GRR(R1,R2,,Rn) and f1:V(R1)Z be a prime distance labeling of R1 as given in the proof of Theorem 12 with f1(u)=0,f1(v)=2, and maximum label ρ1. Apply f1 in the natural way to the subgraph isomorphic to R1 in G.

Again, use the proof of Theorem 12 to find a prime distance labeling f2 of R2 such that f2(u)=0,f2(v)=2, and f2(x)>ρ1 for all vertices x{u,v}. Then apply the labeling f2+2 to the subgraph R2 in G. Continue in this way to label the subgraphs R3,,Rn. Since lifting each labeling by 2 preserves the induced edge labels on each subgraph isomorphic to Ri for 1in, and introduces no conflicts in the labels of the ends of the rainbows, we see that G is a prime distance graph. ◻

3. Conclusion

We have provided many new families of graphs with chromatic number at least 3 which are prime distance graphs. One possible direction forward would be to prove Theorems 7,8,10,12 and 13 independently of de Polignac’s Conjecture (or show that the conjecture is necessary).

Acknowledgment

This research was funded by National Board of Higher Mathematics Department of Atomic Energy, Government of India, Mumbai, by their grant no. 02011/10/21NBHM- (R.P)/R&D/8007/Date: 13-07-2021. J George Barnabas and Yegnanarayanan Venkatraman gratefully acknowledges the same.

Funding

This work was supported by National Board of Higher Mathematics, Government of India.

Role of Funding Support

The National Board of Higher Mathematics-NBHM, Department of Atomic Energy, Government of India has provided scholarship support for George Barnabas to complete his doctoral thesis work.

References:

  1. Chang, G. J., Daphne, D., Liu, F. and Zhu, X., 1999. Distance graphs and T-coloring. J. Combin. Theory Ser. B, 75, pp.259–269.

  2. Chen, J. J., Huang, K. C. and Chang, G. J., 1997. Integral distance graphs. J. Graph Theory, 25, pp.287–294.

  3. de Grey, A., 2018. The chromatic number of the plane is at least 5. Geocombinatorics, 28, pp.18–31.

  4. Huele, M. J. H., 2018. Computing small unit-distance graphs with chromatic number 5. Geocombinatorics, 28, pp.32–50.

  5. Holt, R., Rinehart, W. and Winston., 1964. Combinatorial Geometry in Plane. Geneva: University of Geneva.

  6. Eggleton, R. B., 1988. New results on 3-chromatic prime distance graphs. Ars Combin., 26, pp.153–180.

  7. Deuber, W. and Zhu, X., 1997. The chromatic number of distance graphs. J. Discrete Math., 25, pp.195–204.

  8. Eggleton, R. B., Erdős, P. and Skilton, D. K., 1985. Coloring the real line. J. Combin. Theory Ser. B, 39, pp.86–100.

  9. Eggleton, R. B., Erdős, P. and Skilton, D. K., 1986. Research problem 77. J. Discrete Math., 58, p.323.

  10. Eggleton, R. B., Erdős, P. and Skilton, D. K., 1988. Update information on research problem 77. J. Discrete Math., 69, p.105.

  11. Eggleton, R. B., Erdős, P. and Skilton, D. K., 1990. Coloring prime distance graphs. Graphs Combin., 32, pp.17–32.

  12. Kemnitz, A. and Kolberg, H., 1998. Coloring of integer distance graphs. J. Discrete Math., 191, pp.113–123.

  13. Voigt, M. and Walther, H., 1994. Chromatic number of prime distance graphs. J. Discrete Appl. Math., 51, pp.197–209.

  14. Voigt, M., 1999. Coloring of distance graphs. Ars Combin., 52, pp.3–12.

  15. Zhu, X., 1998. Pattern-periodic coloring of distance graphs. J. Combin. Theory Ser. B, 73, pp.195–206.

  16. Voigt, M. and Walther, H., 1991. On the chromatic number of special distance graphs. Discrete Math., 97, pp.395–397.

  17. Yegnanarayanan, V., 2014. Chromatic number of graphs with special distance sets I. Algebra Discrete Math., 17, pp.135–160.

  18. Yegnanarayanan, V., 2002. On a question concerning prime distance graphs. J. Discrete Math., 245, pp.293–298.

  19. Yegnanarayanan, V. and Barnabas, G., 2022. Chromatic coloring of distance graph III. Proyecciones, 42, pp.175–204.

  20. Yegnanarayanan, V., Anitha, M., Rohatinovici, N. and Balas, E., 2021. Chromatic coloring of distance graph II. Journal of Tianjin University Science and Technology, 54(10), pp.56–67.

  21. Yegnanarayanan, V., 2021. Chromatic coloring of distance graphs I. International Journal of Innovative Technology and Exploring Engineering, 10, pp.31–34.

  22. Walther, H., 1990. Über eine spezielle Klasse unendlicher Graphen. In K. Wagner & R. Bodendiek (Eds.), Graphentheorie II (pp. 268–295). Mannheim: Bibliographisches Institut.

  23. Eggleton, R. B., 1987. Three unsolved problems in graph theory. Ars Combin., 23, pp.105–121.

  24. Laison, J. D., Starr, C. and Halker, A., 2013. Finite prime distance graphs and 2-odd graphs. Discrete Math., 313, pp.2281–2291. Yegnanarayanan, V., Barnabas, G. and Freyberg, B., 2024. Chromatic coloring of distance graphs IV. submitted.