Abstract
For given graphs  and any integer j, the size of the multipartite Ramsey number  is the smallest positive integer t such that any n-coloring of the edges of  contains a monochromatic copy of  in color i for some i, , where  denotes the complete multipartite graph having j classes with t vertices per each class. In this paper, we computed the size of the multipartite Ramsey numbers  for any  and , for any  and .
    MSC:
                05D10; 05C55
            1. Introduction
In this paper, we were only concerned with undirected, simple and finite graphs. We followed [] for terminology and notations not defined here. For a given graph G, we denoted its vertex set, edge set, maximum degree and minimum degree by , ,  and , respectively. For a vertex , we used  and  to denote the degree and neighbours of v in G, respectively. The neighbourhood of a vertex  are denoted by  and .
As usual, a cycle and a path on n vertices are denoted by  and , respectively. A complete graph on n vertices, denoted , is a graph in which every vertex is adjacent, or connected by an edge, to every other vertex in G. By a stripe  we mean a graph on  vertices and m independent edges. A clique is a subset of vertices such that there exists an edge between any pair of vertices in that subset of vertices. An independent set of a graph is a subset of vertices such that there exists no edges between any pair of vertices in that subset. Let C be a set of colors  and  be the edges of a graph  An edge coloring  assigns each edge in  to a color in C. If an edge coloring uses k color on a graph, then it is known as a k-colored graph. The complete multipartite graph with the partite set ,  for , denoted by . We use  to denote the set of edges between partite sets  and . The complement of a graph G, denoted by , is a graph with the same vertices as G and contains those edges which are not in G. Let  be any subset of vertices of G. Then, the induced subgraph G[T] is the graph whose vertex set is T and whose edge set consists of all of the edges in E(G) that have both endpoints in T.
Since 1956, when Erdös and Rado published the fundamental paper [], major research has been conducted to compute the size of the multipartite and bipartite Ramsey numbers. A big challenge in combinatorics is to determining the Ramsey numbers for the graphs. We refer to [] for an overview on Ramsey theory. Ramsey numbers are related to other areas of mathematics, like combinatorial designs []. In fact, exact or near-optimal values of several Ramsey numbers depend on the existence of some combinatorial designs like projective planes, which have been studied to date. Many of these connections are briefly described in [,]. There are many applications of Ramsey theory in various branches of mathematics and computer science, such as number theory, information theory, set theory, geometry, algebra, topology, logic, ergodic theory and theoretical computer science []. In particular, multipartite Ramsey numbers have applications in decision-making problems and communications []. There are many mathematicians who present the new results of multipartite Ramsey numbers every year. As a result of this vast range of applications, we were motivated to conduct research on multipartite Ramsey numbers.
For given graphs  and integer j, the size of the multipartite Ramsey number  is the smallest integer t such that any n-coloring of the edges of  contains a monochromatic copy of  in color i for some i, , where  denotes the complete multipartite graph having j classes with t vertices per each class. G is n-colorable to  if there exist a t-edge decomposition of G say , where  for each 
The existence of such a positive integer is guaranteed by a result in []. The size of the multipartite Ramsey numbers of small paths versus certain classes of graphs have been studied in [,,]. The size of the multipartite Ramsey numbers of stars versus certain classes of graphs have been studied in [,]. In [,], Burger, Stipp, Vuuren, and Grobler investigated the multipartite Ramsey numbers , where  and  are in a completely balanced multipartite graph, which can be naturally extended to several colors. Recently, the numbers  have been investigated for special classes: stripes versus cycles; and stars versus cycles, see [] and its references. In [], authors determined the necessary and sufficient conditions for the existence of multipartite Ramsey numbers  where both G and H are incomplete graphs, which also determined the exact values of the size multipartite Ramsey numbers  for all integers  and . Syafrizal et al. determined the size multipartite Ramsey numbers of path versus path [].  and  where G is a star forest, namely a disjoint union of heterogeneous stars have been studied in []. The exact values of the size Ramsey numbers  and  for  computed in [].
In [], Lusiani et al. determined the size of the multipartite Ramsey numbers of , for , where H is a path or a cycle on n vertices, and  is a star of order . In this paper, we computed the size of the multipartite Ramsey numbers  for  and , for  and  which are the new results of multipartite Ramsey numbers. Computing classic Ramsey numbers is very a difficult problem, therefore we can use multipartite and bipartite Ramsey numbers to obtain an upper bound for a classic Ramsey number. In particular, the first target of this work was to prove the following theorems:
Theorem 1. 
 where .
In [], Jayawardene et al. determined the size of the multipartite Ramsey numbers  where  and . The second goal of this work extends these results, as stated below.
Theorem 2. 
Let  and . Then
      
        
      
      
      
      
    
We estimate that this value of  holds for every . We checked the proof of the main theorems into smaller cases and lemmas in order to simplify the idea of the proof.
2. Proof of Theorem 1
In order to simplify the comprehension, let us split the proof of Theorem 1 into small parts. We begin with a simple but very useful general lower bound in the following lemma:
Lemma 1. 
 where .
Proof.  
Consider  where  with partition sets ,  for . Consider , decompose the edges of  into graphs , and , where  is a null graph and , where  is . In fact  is isomorphic to  and:
        
      
        
      
      
      
      
    
Clearly , ,  and . Since , we have , that is, , which means that  and the proof is complete.    □
Observation 1. 
Let . For any subgraph of G, say H, either H has a subgraph isomorphism to  or  has a subgraph isomorphism to .
Proof.  
Let , for  the proof is same. Without loss of generality (w.l.g.), let  and  be a partition set of  and P be a maximum path in H. If , then H has a subgraph isomorphic to , so let . If , then  has a subgraph isomorphic to . Hence, we may assume that , w.l.g., and let . Since ,  and  are in  and there is at least one edge of  in , in any case,  and the proof is complete.    □
We determined the exact value of the multipartite Ramsey number of  for  in the following lemma:
Lemma 2. 
 for .
Proof.  
Let  and  be a partition set of . Consider a three-edge coloring ,  and  of G. By Lemma 1, the lower bound holds. Now, let M be the maximum matching in . If , then the lemma holds, so let . If , then we have  and by Observation 1, the lemma holds, so let . W.l.g., we may assume that . By considering the edges between  and  and the edges between  and , we have . Hence, by Observation 1, the lemma holds.    □
In the next two lemmas, we consider  for certain values of n. In particular, we proved that  for  in Lemma 3 and  in Lemma 4.
Lemma 3. 
 for .
Proof.  
Let  for  be a partition set of . Consider a three-edge coloring ,  and  of G. By Lemma 1 the lower bound holds. Now, let M be the maximum matching in  and consider the following cases:
Case 1: . If  then  and the proof is complete. So let . W.l.g., we may assume that , hence, we have  and by Observation 1, the proof is complete.
Case 2: . In this case, if  or , then the proof is the same as case 1. So let  and w.l.g., we may assume that —considering any  and  in . In any case, we have  has a subgraph isomorphic to , hence, by Observation 1, the lemma holds. Therefore, we have . Now, through cases 1 and 2, the proof is complete.    □
Lemma 4. 
.
Proof.  
Let  for  be a partition set of . By Lemma 1, the lower bound holds. Consider a three-edge coloring (, , ) of G where . Let M be a maximum matching in , if , then the proof is same as Lemma 3. Hence, we may assume that  and w.l.g., let . By Observation 1, there is at least one edge between  and  in , say , and similarly, there is at least one edge between  and  in , say , otherwise  and the proof is complete. Now, by Observation 1, there is at least one edge between  and  in , and let  be this edge. If  (say  ), then .
Now, consider the vertex  and , since  and , it is easy to check that  and , otherwise  and the proof is complete. Similarly, we have  and . Now, by considering the edges of , it is easy to check that  and the lemma holds. Hence, we have  (say ), in this case, and we have , otherwise, if there exists at least one edge between  and  in , say e, then set  and the proof is the same. Hence, by considering the vertex  and , since  and , it is easy to check that  and by Observation 1 the proof is complete.    □
Lemma 5. 
 for each .
Proof.  
Let  for  be a partition set of  where . We will prove this Lemma by induction. For the base step of the induction, since  and , lemma holds by Lemmas 3 and 4. Suppose that  and  for each . We will show that . By contradiction, we may assume that , that is,  is three-colorable to . Consider a three-edge coloring  of G, such that ,  and . By the induction hypothesis and Lemma 1, we have . Therefore, since  and , we have . Now, we have the following cases:
Case 1: .
Since , we have a copy of  in G. In other words, for each , there is a vertex, say , such that . W.l.g., we may assume that  would be these vertices. Since , we have  and . Hence, by the induction hypothesis, we have . We consider that the three vertices do not belong to , i.e., A. Since , we have . Now, we consider the following Claim:
Claim 1. 
 where  and .
Proof of the Claim. By contradiction, we may assume that . In other words, let , then we have:
        
      
        
      
      
      
      
    
      
        
      
      
      
      
    
        which is a contradiction implying that .
Claim 2. 
There is at least one vertex in .
Proof of the Claim. Let , then . Since , by Claim 1, if , we have  for . Now, we have:
        
      
        
      
      
      
      
    
Hence, we have  and thus . If  then we have:
        
      
        
      
      
      
      
    
Hence, . Therefore, .
By Claim 2, let . Since , we have . Hence, by Observation 1, we again have a contradiction.
Case 2: .
In this case, by Claim 1 we have . Since  and , by the induction hypothesis, we have . Now, we have the following claim:
Claim 3. 
.
Proof of the Claim. Let . Since  and , we have  and therefore, , that is, .
By Claim 3, we have . W.l.g., we may assume that  has three vertices, since , and we have . We consider the three vertices belonging to , and now, we have the following subcases:
Subcase 2-1:  for only one . W.l.g. we may assume that  and . Since  and  we have  and , otherwise,  and by Observation 1; a contradiction. W.l.g. we may assume that . Consider . Since , if  is a maximum matching in , then , otherwise we have ; a contradiction again. Since  and , we have  or ; also a contradiction.
Subcase 2-2:  for each . W.l.g., we may assume that  and . Hence . Since , we have  for each . W.l.g., we may assume that ,  and . If  and  then we have  and by Observation 1; a contradiction. So let  or . If , then, since , we have , that is, ; we have a contradiction again. So let  and . Since , we have . If , then we have . So let  and . Since  there is at least one edge, say . W.l.g., we may assume that  and . Since  and  we have . Hence, we had a ; a contradiction.
Subcase 2-3:  for only one . W.l.g., we may assume that  and . Hence, we have . Since , we have , that is, . W.l.g., we may assume that  where  and . Now, we have the following claim:
Claim 4. 
.
Proof of the Claim. By contradiction, w.l.g., we may assume that . Since , we have . Consider  and . Hence,  and  for each ; we have a contradiction to subcase 2-2.
Now, by Claim 4, we have . In this case, by Observation 1, we have  or ; we have a contradiction again.
Therefore, by Cases 1 and 2, we have  for .    □
Now, by Lemmas 1 and 5, we have the following lemma:
Lemma 6. 
 for .
In the next two lemmas, we consider  for each values of  and . In particular, we proved that  for  and . We started with the following lemma:
Lemma 7. 
Let  and . Given that , it follows that 
Proof.  
Let  and . For  let  be partition set of  where . Assume that  is true. To prove . Consider three-edge coloring  of G. Suppose that , we prove that  or . Let  be the maximum matching in . Hence, by the assumption, , that is . Now, we have the following claim:
Claim 5. 
.
Proof of the Claim. Consider the following cases:
Case 1: Let . In this case, we have:
        
      
        
      
      
      
      
    
Hence:
        
      
        
      
      
      
      
    
Case 2: Let  where . In this case, we have:
Hence we have:
By Claim 5, G contains three vertices, say  and z in . Consider the vertex set  and let . Now, we have the following cases:
Case 1: Let ,  and , where  for  are distinct partition sets of . Note that all vertices of A are adjacent to each other in . Since , we have . Consider the partition  for . Since , if  for at least one , then we have  and the proof is complete by Observation 1. Now, let  for each . Hence, for  there exists a vertex, say u such that . Consider . If , then we have  and by Observation 1, the proof is complete. Therefore, let . W.l.g., we may assume that . In this case, we have . On the contrary, let  and set . Clearly  is a match where , which contradicts the maximality of . Hence, we have . Therefore, we have  and, by Observation 1, the proof is complete.
Case 2: Let  and  where  are distinct partition sets of G. W.l.g., let  and . Consider the partition . Since , if , then we have  and by Observation 1, the proof is complete. So let  for each . Now, we have the following claim.
Claim 6. 
Let , and w.l.g. let . If , then . If , then  has the same neighbor in .
Proof of the Claim. Let . W.l.g., we may assume that . By contradiction, let , w.l.g., let . In this case, we set . Clearly  is a match with , which contradicts the maximality of . Thus, let  for , if  has a different neighbor, then the proof is same.
Claim 7. 
There is at least one edge, say , such that .
Proof of the Claim. If , then there is at least one edge, say , such that . Otherwise, we have  where , hence, by Observation 1; we have a contradiction. So, let . In this case, if , then the proof is same. Now, let . We have , that is, . Hence, there is at least one vertex, say ; a contradiction to .
By Claim 7, there is at least one edge, say , such that . W.l.g., let  such that , also, w.l.g., assume that . If , then by Claim 7, we have . Hence, we have . So, let , in this case, by Claim 7, we have , and if x or y is this vertex, then ; otherwise, . In any case, by Observation 1, the proof is complete.
Case 3: Let  where  is a partition set of , say . If there exists a vertex, say , where , then the proof is the same as Case 2. Hence, let . Since , there exists an edge, say , such that . Consider the neighbors of vertices v and u in . W.l.g., let . If , then we have , so let . In this case, by Claim 7, we had . Hence, w.l.g., we may assume that  and  be in  and . Now, set  and , the proof is the same as Case 2 and the proof is complete.
According to the Cases  and 3 we have .    □
The results of Lemmas 1, 2, 6 and 7, concludes the proof of Theorem 1.
3. Proof of Theorem 2
In this section, we investigate the size multipartite Ramsey numbers  for  and . In order to simplify the comprehension, let us split the proof of Theorem 2 into small parts. For , since the bipartite graph has no odd cycle, we have . For other cases, we start with the following proposition:
Proposition 1. 
 where .
Proof.  
  
    
      
      
    
  
  
  
    
      
      
    
  
  
Clearly, . Consider  with the partition set  for . Let G be a subgraph of . For , if , then proof is complete, so let . In this case, we have , hence , that is, . For  by contradiction, we may assume that , that is,  is 2-colorable to , say  and . Since  [], and , we have . Let  and  for . If there exists  such that , say , then we have  and . Let . Since , for each  in ,  cannot be adjacent to  and  for . Hence, we have  for each . One can easily check that in any case, we have ; a contradiction, hence, let  for each . Set . Now, we have the following cases:
Case 1: . Let . In this case, we have . Consider the vertex set . Since , we have . Hence, . W.l.g., let . By similarity, we have  and , see Figure 1. In any case, we have ; a contradiction again.
      
    
    Figure 1.
      .
  
Case 2: . W.l.g., let  for . In this case, we have . W.l.g., let . Since  and , we have . If , then considering Figure 2a, the proof is the same as case 1. So let . W.l.g., let , , . In this case, consider Figure 2b and the proof is the same as case 1. Hence, in any case, we have ; again a contradiction.
      
    
    Figure 2.
      (a) , (b) .
  
By Cases 1 and 2, we have . Thus, the proof is complete and the proposition holds.    □
We determine the exact value of the multipartite Ramsey number  for  in the following lemma:
Lemma 8. 
For each  we have .
Proof.  
First, we show that . Consider the coloring given by  where  and . Since  and  is bipartite, we have  and , that is, . For the upper bound, consider  with partite sets  for . We will prove this by induction. For , by Proposition 1, the lemma holds. Suppose that  for each . We will show that , as follows: by contradiction, we may assume that , that is,  is 2-colorable to , say  and . Let . Hence, by the induction hypothesis, we have . Therefore, since  and , we have . If there exists i and j such that , then we have ; a contradiction. Hence, we have  for . Let . Hence, we have . Since , we have . Since , one can easily check that, in any case, we have , where, . If , one can easily observe that we have ; a contradiction again. So let  and consider the following cases:
Case 1:  for only one i, that is, . W.l.g., let  and . Then, we have  and . Since , we have , that is, there exists at least two edges, say  and  in , where . W.l.g., let  for . One can easily check that , otherwise, we have ; a contradiction. Since  and , we have . Hence, we have . W.l.g., we may assume that . In this case, we have ; a contradiction again.
Case 2: . W.l.g., let  and . Let  and , that is, we have  and . That is, there exists at least one edge, say , where  and . W.l.g., let . One can easily check that . Hence, we have  and the proof is same as case 1.
By cases 1 and 2, we have the assumption that  does not hold. Now we have  for each . This completes the induction step and the proof.    □
Lemma 9. 
For  and , we have .
Proof.  
To show that , assume that . Consider the coloring given by  where  such that  and . Since  is bipartite, we have , and
        
      
        
      
      
      
      
    
      
        
      
      
      
      
    
Since , we have . Hence, we have . Since , we have  for .    □
Lemma 10. 
.
Proof.  
By Lemma 9, we have . For the upper bound, consider the coloring given by  such that . Since , we have . Let ; hence, we have . W.l.g., let  be these vertices. If , then we have . So let . Consider the edge , and it is easy to show that  for some , otherwise, we have . In any case, one can easily check that ; which is a contradiction. Thus, we obtain .    □
Lemma 11. 
For  we have .
Proof.  
By Lemma 9, we have . To prove , consider  with partite set  for , where . We will prove this by induction. For  by Lemma 10, the lemma holds. Now, we consider the following cases:
Case 1: , where . Suppose that  for each . We will show that  as follows: by contradiction, we may assume that , that is,  is 2-colorable to , say  and . Let  for . Hence, by the induction hypothesis, we have . Therefore, since  and , we have . If there exists , where , then ; a contradiction. Now, we have . Since  and , we have , that is, there exists two vertices, say  and  in . Since , we have , where . Hence, we have the following claim:
Claim 8. 
Let  and w.l.g., we may assume that . If  then . If  then . If  then  and  have the same neighbor in S.
Proof of the Claim. By contradiction. We may assume that  and , in this case, we set . Clearly,  is a match with , which contradicts the . If  and  has a different neighbor, then the proof is same.
Since  and . If , say , then there is at least one edge, say  such that . Otherwise, we have ; we again have a contradiction. W.l.g., let . Now, by Claim 8 we have . One can easily check that in any case, we have ; again a contradiction. So w.l.g., let  and . In this case, since , we have ; a contradiction again.
Case 2:  where , . Suppose that  for . We show that  as follows: by contradiction, we may assume that , that is,  is 2-colorable to , say  and . Let . By the induction hypothesis, we have . Therefore, since  and , we have  and thus, we have the following claim:
Claim 9. 
There exist two edges, say  and  in , such that  and  are in different partites.
Proof of the Claim. W.l.g., assume that  and . By contradiction, assume that , that is, . Since  and , we have . Since , . W.l.g., let  where . W.l.g., we may assume that . Hence, we have . In other words, take  and  and the proof is complete. Hence, we have  for , in other words, if there exists , then set  and  and the proof is complete. Therefore, for each  we have  which means that ; a contradiction to .
Now, by Claim 9 there exist two edges, say  and  in , such that  and  are in different partite. W.l.g., let  and , since are these edges, and let  for . Set , hence, we have . Since , we have . Therefore, by the induction hypothesis, we have . Let , set  hence , that is, ; again a contradiction. Hence, the assumption that  does not hold and we have . This completes the induction step and the proof is complete. By Cases 1 and 2, we have  for .    □
The results of Proposition 1 as well as Lemmas 8 and 11 concludes the proof of Theorem 2.
4. Concluding Remarks and Further Works
There are several papers in which the multipartite Ramsey numbers have been studied. In this paper, as a first target, we compute the size of the multipartite Ramsey number  for . To approach this purpose, we prove four lemmas as follows:
- where ;
 - for ;
 - for ;
 - Let and . Given that , it follows that
 
We computed the size of the multipartite Ramsey numbers , for  and  as the second purpose of this paper. This extended the result of []. To approach this purpose, we proved the following:
- where ;
 - For each we have ;
 - For we have ; We estimated our result for which holds for every , so it could be a good problem to work on.
 
In addition, one can compute  and also , for  and  in the future, using the idea of proofs in this paper.
Author Contributions
These authors contributed equally to this work. All authors have read and agreed to the published version of the manuscript.
Funding
This research received no external funding.
Institutional Review Board Statement
Not applicable.
Informed Consent Statement
Not applicable.
Conflicts of Interest
The authors declare that there is no conflict of interest.
References
- Bondy, J.A.; Murty, U.S.R. Graph Theory with Applications; American Elsevier Publishing Co., Inc.: New York, NY, USA, 1976. [Google Scholar]
 - Erdös, P.; Rado, R. A partition calculus in set theory. Bull. Amer. Math. Soc. 1956, 62, 427–489. [Google Scholar] [CrossRef]
 - Graham, R.L.; Rothschild, B.L.; Spencer, J.H. Ramsey Theory; Wiley Series in Discrete Mathematics and Optimization; John Wiley and Sons: Hoboken, NJ, USA, 1990. [Google Scholar]
 - Erdös, P.; Graham, R.L. On Partition Theorems for Finite Graphs. Infinite and Finite Sets. Colloq. Math. Soc. János Bolyai N.-Holl. Amsterdam. 1975, 10. Available online: http://citeseerx.ist.psu.edu/viewdoc/summary?doi=10.1.1.210.8857 (accessed on 30 January 2021).
 - Parsons, T. Ramsey Graph Theory; Beineke, L.W., Wilson, R.J., Eds.; Selected Topics in Graph Theory; Academic Press: London, UK, 1978; pp. 361–384. [Google Scholar]
 - Rosta, V. Ramsey theory applications. Electron. J. Comb. 2004, 1000, DS13. [Google Scholar] [CrossRef]
 - Roberts, F.S. Applications of Ramsey theory. Discret. Appl. Math. 1984, 9, 251–261. [Google Scholar] [CrossRef]
 - Łuczak, T.; Polcyn, J. The multipartite Ramsey number for the 3-path of length three. Discret. Math. 2018, 341, 1270–1274. [Google Scholar] [CrossRef]
 - Sy, S. On the size multipartite Ramsey numbers for small path versus cocktail party graphs. Far East J. Appl. Math. 2011, 55, 53–60. [Google Scholar]
 - Jayawardene, C.; Baskoro, E.T.; Samarasekara, L.; Sy, S. Size multipartite Ramsey numbers for stripes versus small cycles. Electron. J. Graph Theory Appl. (EJGTA) 2016, 4, 157–170. [Google Scholar] [CrossRef][Green Version]
 - Perondi, P.H.; Carmelo, E.L.M. Set and size multipartite Ramsey numbers for stars. Discrete Appl. Math. 2018, 250, 368–372. [Google Scholar] [CrossRef]
 - Lusiani, A.; Baskoro, E.T.; Saputro, S.W. On size multipartite Ramsey numbers for stars versus paths and cycles. Electron. J. Graph Theory Appl. (EJGTA) 2017, 5, 43–50. [Google Scholar] [CrossRef]
 - Burger, A.P.; Vuuren, J.H.V. Ramsey numbers in complete balanced multipartite graphs. II. Size numbers. Discret. Math. 2004, 283, 45–49. [Google Scholar] [CrossRef]
 - Burger, A.P.; Grobler, P.J.P.; Stipp, E.H.; van Vuuren, J.H. Diagonal Ramsey numbers in multipartite graphs. Util. Math. 2004, 66, 137–163. [Google Scholar]
 - Lusiani, A.; Baskoro, E.T.; Saputro, S.W. On size multipartite Ramsey numbers for stars. Indones. J. Comb. 2020, 3, 109–115. [Google Scholar] [CrossRef]
 - Sy, S.; Baskoro, E.T.; Uttunggadewa, S. The size multipartite Ramsey numbers for paths. J. Comb. Math. Comb. Comput. 2005, 55, 103. [Google Scholar]
 - Lusiani, A.; Baskoro, E.T.; Saputro, S.W. On Size Bipartite and Tripartite Ramsey Numbers for The Star Forest and Path on 3 Vertices. J. Math. Fundam. Sci. 2020, 52, 1–16. [Google Scholar] [CrossRef]
 - Jayawardene, C. Size Multipartite Ramsey Numbers for Small Paths vs. K2, n. Ann. Pure Appl. Math. 2019, 19, 7–17. [Google Scholar] [CrossRef]
 
Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.  | 
© 2021 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/).