On the $g$-good-neighbor connectivity of graphs
Zhao Wang, Yaping Mao, Sun-Yuan Hsieh, Jichang Wu

TL;DR
This paper investigates the properties and bounds of the $g$-good-neighbor connectivity in graphs, characterizes specific cases, and establishes extremal results to enhance understanding of fault tolerance in interconnection networks.
Contribution
It provides new bounds, characterizations, and extremal results for the $g$-good-neighbor connectivity, advancing the theoretical understanding of fault tolerance parameters.
Findings
Bounds: $1 \,\leq\, \kappa^g(G) \leq n-2g-2$ for certain $g$.
Characterizations of graphs with $\,\kappa^g(G)=1,2$ and trees with $\,\kappa^g(T_n)=n-t$.
Extremal results for the $g$-good-neighbor connectivity.
Abstract
Connectivity and diagnosability are two important parameters for the fault tolerant of an interconnection network . In 1996, F\`{a}brega and Fiol proposed the -good-neighbor connectivity of . In this paper, we show that for , and graphs with and trees with for are characterized, respectively. In the end, we get the three extremal results for the -good-neighbor connectivity.
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsInterconnection Networks and Systems · Supercapacitor Materials and Fabrication · Graphene research and applications
On the -good-neighbor connectivity of graphs 111Supported by the National
Science Foundation of China (Nos. 11601254, 11551001, 11161037, 61763041, 11661068, and 11461054) and the Science Found of Qinghai Province (Nos. 2016-ZJ-948Q, and 2014-ZJ-907) and the Qinghai Key Laboratory of Internet of Things Project (2017-ZJ-Y21).
Zhao Wang222College of Science, China Jiliang University, Hangzhou 310018, China. [email protected], Yaping Mao333Corresponding author 444School of Mathematics and Statistis, Qinghai Normal University, Xining, Qinghai 810008, China. [email protected], Sun-Yuan Hsieh 555Department of Computer Science and Information Engineering, National Cheng Kung University, Tainan 701, Taiwan [email protected], Jichang Wu 666School of Mathematics, Shandong University, Jinan, Shandong 250100, China [email protected]
Abstract
Connectivity and diagnosability are two important parameters for the fault tolerant of an interconnection network . In 1996, Fàbrega and Fiol proposed the -good-neighbor connectivity of . In this paper, we show that for , and graphs with and trees with for are characterized, respectively. In the end, we get the three extremal results for the -good-neighbor connectivity.
Keywords: Connectivity, -good-neighbor connectivity, extremal problem.
AMS subject classification 2010: 05C40; 05C05; 05C76.
1 Introduction
With the rapid development of VLSI technology, a multiprocessor system may contain hundreds or even thousands of nodes, and some of them may be faulty when the system is implemented. As the number of processors in a system increases, the possibility that its processors may be comefaulty also increases. Because designing such systems without defects is nearly impossible, reliability and fault tolerance are two of the most critical concerns of multiprocessor systems [37].
By the definition proposed by Esfahanian [6], a multiprocessor system is fault tolerant if it can remain functional in the presence of failures. Two basic functionality criteria have received considerable attention. The first criterion for a system to be regarded as functional is whether the network logically contains a certain topological structure. This is the problem that occurs when embedding one architecture into another [15, 35]. This approach involves using system-wide redundancy and reconfiguration. The second functionality criterion considers a multiprocessor system functional if a fault-free communication path exists between any two fault-free nodes; that is, the topological structure of the multiprocessor system remains connected in the presence of certain failures. Thus, connectivity and edge connectivity are two major measurements of this criterion [35]. The connectivity of a graph , denoted by , is the minimal number of vertices whose removal from produces a disconnected graph or only one vertex; the edge connectivity of a graph , denoted by , is the minimal number of edges whose removal from produces a disconnected graph. However, these two parameters tacitly assume that all vertices that are adjacent to, or all edges that are incident to, the same vertex can potentially fail simultaneously. This is practically impossible in some network applications.
For a graph , let , , , , and denote the set of vertices, the set of edges, the size, the complement, and the diameter of , respectively. A subgraph of is a graph with , , and the endpoints of every edge in belonging to . For any subset of , let denote the subgraph induced by ; similarly, for any subset of , let denote the subgraph induced by . We use to denote the subgraph of obtained by removing all the vertices of together with the edges incident with them from ; similarly, we use to denote the subgraph of obtained by removing all the edges of from . If and , we simply write and for and , respectively. For two subsets and of we denote by the set of edges of with one end in and the other end in . If , we simply write for . The degree of a vertex in a graph , denoted by , is the number of edges of incident with . Let and be the minimum degree and maximum degree of the vertices of , respectively. The set of neighbors of a vertex in a graph is denoted by . The union of two graphs and is the graph with vertex set and edge set . If is the disjoint union of copies of a graph , we simply write .
1.1 The -extra (edge-)connectivity
Fàbrega and Fiol [8, 9] introduced the following measures. A subset of vertices is said to be a cutset if is not connected. A cutset is called an -cutset, where is a non-negative integer, if every component of has at least vertices. If has at least one -cutset, the -extra connectivity of , denoted by , is then defined as the minimum cardinality over all -cutsets of . A connected graph is said to be -extra connected if has a -extra cut.
Given a graph and a non-negative integer , the -extra edge-connectivity, written as , is the minimal cardinality of a set of edges in , if it exists, whose deletion disconnects and leaves each remaining component with more than vertices.
Note that and for any graph if is not a complete graph. Therefore, the -extra connectivity (resp. -extra edge connectivity) can be regarded as a generalization of the classical connectivity (resp. classical edge connectivity) that provides measures that are more accurate for reliability and fault tolerance for large-scale parallel processing systems. Regarding the computational complexity of the problem, based on thorough research, no polynomial-time algorithm has been presented to compute for a general graph; nor has there been any tight upper bound for [6]. However, can be computed by solving numerous network flow problems [7].
1.2 The -good-neighbor connectivity
Let be connected. A fault set is called a -good-neighbor faulty set if for every vertex in . A -good-neighbor cut of is a -good-neighbor faulty set such that is disconnected. The minimum cardinality of -good-neighbor cuts is said to be the -good-neighbor connectivity of , denoted by . A connected graph is said to be -good-neighbor connected if has a -good-neighbor cut. For more research on -good-neighbor connectivity, we refer to [16, 18, 22, 24, 28, 29, 33, 32, 36, 38, 39].
The relation of -extra connectivity and -good-neighbor connectivity was given in [23].
Theorem 1.1**.**
[23]* If is a -good-neighbor connected graph, then .*
* If is a -extra and -good-neighbor connected graph, then .*
The range of the integer can be determined immediately.
Proposition 1.1**.**
Let be a non-negative integer. If has its -good-neighbor connectivity, then
[TABLE]
and
[TABLE]
Proof.
From the definition of -good-neighbor connectivity, there exists with such that is not connected and the minimum degree of each component of is at least . Let be the components of . Then and
[TABLE]
and hence . For each , we have .
Furthermore, , and hence . ∎
The monotone property of for non-negative integer is true.
Proposition 1.2**.**
Let be a non-negative integer, and let be a connected graph. Then
[TABLE]
Proof.
From the definition of -good-neighbor connectivity, by deleting vertices in , the resulting graph is not connected and the minimum degree of each component is at least , and hence . ∎
The monotone property of is true in terms of connected graphs .
Observation 1.1**.**
Let be a connected graph. If is a spanning subgraph of , then .
But for , the above monotone property is not true.
Remark 1.1**.**
Let be a graph obtained from four cliques with and and three vertices by adding edges in . Let be a graph obtained from two cliques defined in , two subgraph such that , and three vertices by adding edges in . Clearly, is a spanning subgraph of ; see Figure . We first show that . By deleting the vertex , there are three components and the minimum degree of each component is at least , and hence , and so . Next, we show that . Suppose . Then there exists a vertex such that by deleting this vertex the resulting graph is not connected and the minimum degree of each component is at least . Note that is the unique cut vertex of . Clearly, there is a component having minimum degree , a contradiction. So . By deleting the vertices , there are two components and the minimum degree of each of them is at least , and hence . So .
1.3 Some classical problems
One of the interesting problems in extremal graph theory is the Erdös-Gallai-type problem, which is to determine the maximum or minimum value of a graph parameter with some given properties. In [1, 12], the authors investigated two kinds of Erdös-Gallai-type problems for monochromatic connection number and monochromatic vertex connection number, respectively. Motivated by these, we study two kinds of Erdös-Gallai-type problems for in this paper.
Problem 1. Given two positive integers and , compute the minimum integer such that for every connected graph of order , if then .
Problem 2. Given two positive integers and , compute the maximum integer such that for every graph of order , if then .
Another interesting problem in extremal graph theory is to study the minimum size of graphs with given parameter; see [25].
Problem 3. Given two positive integers and , compute the minimum integer , where the set of all graphs of order (that is, with vertices) with -good-neighbor connectivity .
In Section , we obtain the exact values of -extra connectivities of complete bipartite graphs, complete multipartite graphs, wheels and paths. For a connected graph of order , we show that for , and for in Section . Graphs with and trees with are characterized, respectively, in Section . In the end, we get the extremal results for the -good neighbor connectivity in Section .
2 Results for special graphs
The following upper and lower bounds are immediate.
Proposition 2.1**.**
Let be a connected graph of order , and let be a non-negative integer such that . Then
[TABLE]
Moreover, the upper and lower bounds are sharp.
Proof.
From the definition of , we have . Suppose . From the definition, we can delete vertices in such that there are at least two components and one of them has no more than vertices, a contradiction. So . Theorem 3.1 shows that the upper bound is sharp. If , then . This implies that the lower bound is sharp. ∎
The following corollary is immediate from Proposition 2.1.
Corollary 2.1**.**
Let be two integers with . If is a connected graph of order , then
[TABLE]
Moreover, the upper and lower bounds are sharp.
In the following, we obtain the exact values for -good neighbor connectivity of some special graphs.
Proposition 2.2**.**
Let be a non-negative integer.
* If is a complete bipartite graph, then and and does not exist for .*
* Let be an integer with . For complete multipartite graph , we have and*
[TABLE]
and does not exist for .
Proof.
By deleting any vertex in , the resulting graph is still a complete bipartite graph and it is connected. If we require the resulting graph is not connected, then we must delete all the vertices of one part. Then . Since , we have .
Similarly to the proof of , we can get . ∎
Proposition 2.3**.**
Let be a non-negative integer.
* If is a wheel of order , then for .*
* If be a path of order , then for .*
Proof.
From the definition of , there exists such that is not connected and the minimum degree of each component of is at least . Note that each component is a path. Then . From Proposition 2.1, we have . It suffices to show that for . Let be the center of , and , and . Choose . Since , it follows that the minimum degree of each component of is at least , and hence . So .
Similarly to the proof of , we have . From Proposition 2.1, we have . It suffices to show . Let . Choose . Since , it follows that the minimum degree of each component of is at least , and hence . So . ∎
3 Graphs with given -good-neighbor connectivity
In this section, we first characterize trees with given -good-neighbor connectivity. Next, we characterize graphs with small -good-neighbor connectivity.
3.1 Trees with given -good-neighbor connectivity
Let be a star with center and leaves , and let be stars with centers , respectively. Furthermore, let be a tree of order obtained from and by adding the edges , where , , and for each ; see Figure 1.
Lemma 3.1**.**
For , we have
[TABLE]
Proof.
Choose . Then is not connected and the minimum degree of each component is exactly . So . It suffices to show . It suffices to prove that for any and , if is not connected, then there exists a component of such that its minimum degree is exactly [math]. If , then there is an isolated vertex in the resulting graph, as desired.
Suppose . Since is not connected, it follows that there exits some such that , and hence . Then there exits some such that , and hence . Furthermore, there exits some such that , and hence . Continue this process, we have
[TABLE]
Clearly, is connected, a contradiction. So , and hence . ∎
Trees with for general and can be characterized.
Theorem 3.1**.**
Let be two integers and be a tree of order with . Then if and only if satisfies one of the following conditions.
* and ;*
* , and .*
Proof.
If and , then . Suppose , and . From Lemma 4.1, we have .
Conversely, we suppose . Then we have the following claim.
Claim 1**.**
.
Proof.
Assume, to the contrary, that . From the definition of , there exists and such that is not connected and the minimum degree of each component of is at least . Since is a tree, it follows that each component of is a subtree of , and the minimum degree of each component is at most , a contradiction. ∎
From Claim 1, we have . If , then , and hence . If , then . Then there exists and such that is not connected and the minimum degree of each component is exactly . Clearly, there exits a cut vertex in such that . Let be the components of . Since is a tree, it follows that for each . Let be the number of isolated vertices in . Since and , it follows that . Furthermore, we have the following claim.
Claim 2**.**
.
Proof.
Assume, to the contrary, that . By deleting these isolated vertices and , the minimum degree of each component of the resulting graph is at least , and hence , a contradiction. ∎
Let be the isolated vertices in . Then we have the following claim.
Claim 3**.**
For each , is a star.
Proof.
Assume, to the contrary, that there exists some such that is not a star. Then contains a , say . Let be the unique edge from to . Let be the set of pendent vertices adjacent to in . Since is not a star, it follows that there is at least one edge in . Since is not connected, it follows that except , there exists another component of order at least , say . Then
[TABLE]
a contradiction. ∎
From Claim 3, is a star for each . Let be the center of , where . Then we have the following claim.
Claim 4**.**
For each , we have .
Proof.
Assume, to the contrary, that there exists some such that . Then there exists a vertex such that . Note that is a leaf of . Then is not connected and the minimum degree of each component is at least , and hence , a contradiction. ∎
From Claim 4, for each . Then . ∎
3.2 Graphs with small -good-neighbor connectivity
Graphs with can be characterized easily.
Observation 3.1**.**
Let be two integers and let be a connected graph of order with . Then if and only if there exists a cut vertex in such that the minimum degree of each connected component of is at least .
We can also characterize graphs with .
Theorem 3.2**.**
Let be two integers and let be a connected graph of order with . Then if and only if satisfies one of the following conditions.
* and there exists a cut vertex set in such that the minimum degree of each component of is at least ;*
* , and , and hold, where*
- * for each cut vertex , there exists a component of such that its minimum degree is at most ,*
- * there exists a cut vertex such that there is exactly one component in having exactly one vertex of degree at most and the neighbors of has degree at least and the minimum degree of other vertices is at least , or there exists a cut vertex such that contains at least components, where one of the component is an isolated vertex and the minimum degree of each of the other components is at least , or there are two non-cut vertices such that is not connected and the minimum degree of each component is at least .*
Proof.
Suppose that satisfies and . Suppose that holds. Since the minimum degree of each component of is at least , it follows that . From Proposition 2.1, we have .
Suppose that holds. Since for each cut vertex , there exists a component of such that its minimum degree is at most , it follows that . If there exists a cut vertex such that there is exactly one component in having exactly one vertex of degree at most and the neighbors of has degree at least and the minimum degree of other vertices is at least , then . If there exists a cut vertex such that contains at least components, where one of the component is an isolated vertex and the minimum degree of each of the other components is at least , then . If there are two non-cut vertices such that is not connected and the minimum degree of each component is at least , . So we have .
Conversely, we suppose . From Proposition 2.1, we have . Suppose . If for each vertex cut set in , there exists a component of such that its minimum degree is at most , then , a contradiction. So there exists a vertex cut set in such that the minimum degree of each component of is at least , as desired.
Suppose . Then we have the following claim.
Claim 5**.**
.
Proof.
Assume, to the contrary, that . Since , it follows that there exists a cut vertex and the minimum degree of each component of is at least [math], and hence , which contradicts to the fact . ∎
From Claim 5, we have . Since , we have the following facts.
Fact 1**.**
For any cut vertex , there exists a component of such that its minimum degree is at most .
Fact 2**.**
There exist two vertices in such that is not connected and the minimum degree of each component of is at least .
Suppose that one of is a cut vertex of . Without loss of generality, we assume that is a cut vertex of . Let be the components of .
Claim 6**.**
At most one of has exactly one vertex.
Proof.
Assume, to the contrary, that there exist such that . Then at least one component of is a isolated vertex, which contradicts to Fact 2. ∎
From Claim 6, if one of , say , has exactly one vertex, then and holds. Suppose that each has at least two vertices.
Claim 7**.**
Exactly one of has minimum degree at most .
Proof.
Assume, to the contrary, that there exist such that and . Then there is a component of such that its minimum degree is at most , which contradicts to Fact . ∎
From Claim 7, exactly one of , say , has minimum degree at most . Then there exists a vertex of degree at most . We claim that . Assume, to the contrary, that the degree of is most in , a contradiction. Then . From Fact 1, holds.
Suppose that neither nor is a cut vertex of . From Fact 1, is not connected and the minimum degree of each component is at least . Then holds. ∎
For , , and , we show the following examples corresponding them.
Example 4.1. Let be a graph obtained from and by adding two new vertices and edges in , where . From Theorem 3.2, .
Example 4.2. Let be a graph obtained from and by adding two new vertices and edges in , where and and . From Theorem 3.2, and .
Example 4.3. Let be a graph obtained from and by adding two new vertices and edges in , where and and . From Theorem 3.2, and .
Example 4.4. Let be a graph obtained from and by adding edges in , where and and . From Theorem 3.2, and .
4 Extremal problems
We now consider the three extremal problems that we stated in the Introduction.
Suppose that are three integers such that is even and . Let be two -regular graphs such that . Let be a graph obtained from and a star with center such that .
Lemma 4.1**.**
Let be three integers with . If is even, then
[TABLE]
Proof.
Let . Then is not connected and the minimum degree of each component of is at least , and hence . Let . It suffices to show . From the definition of , there exists with such that the minimum degree of each component of is at least . Since , it follows that . If , then is connected, a contradiction. So . So , and hence . ∎
Suppose that are three integers such that is odd and . Then is odd. Let be two integers such that is even, and is odd, and , and , and . Let be a -regular graph of order . Let be a graph of order such that the degree of one vertex is exactly , and the degree of each of the other vertices is exactly . Let be a graph obtained from and a star with center such that . Similarly, we have the following lemma.
Lemma 4.2**.**
Let be three integers with . Then
[TABLE]
Let be a tree of order obtained from three stars with centers by adding two edges .
Lemma 4.3**.**
Let be two integers with . Then
[TABLE]
Proof.
Let . Then is not connected and the minimum degree of each component of is at least , and hence . Let . It suffices to show . From the definition of , there exists with such that the minimum degree of each component of is at least . If , then or . Without loss of generality, let . Then . Since is not connected and , it follows that . Then . Clearly, is connected, a contradiction. If , then , and hence . So , and hence . ∎
Theorem 4.1**.**
Let be three integers with and .
* If , then .*
* If is even and , then*
[TABLE]
* If is odd and , then*
[TABLE]
Proof.
Let . From Lemma 4.3, we have . Since we only consider connected graphs, it follows that , and hence .
Suppose that is even. Let . From Lemma 4.1, we have . It suffice to show . Let be a conneced graph of order with such that is minimized. Then exists with such that the minimum degree of each component of is at least . Then . Since is connected, it follows that , and hence .
Suppose that is odd. Let . From Lemma 4.2, we have . Similarly to the proof of , we have . ∎
Lemma 4.4**.**
Let be three integers with . Let be the graph obtained from three cliques by adding the edges in . Then
[TABLE]
Proof.
Let . Since , it follows that is not connected and each component has at least vertices, and hence . Clearly, , and hence . ∎
Theorem 4.2**.**
Let be two integers with and . Then
[TABLE]
Proof.
To show , we construct defined in Lemma 4.4. Then . Since , it follows that .
Let be a graph with vertices such that . We claim that . Assume, to the contrary, that . Then there exists a vertex set and such that the minimum degree of each component of is at least . Let be the components of . The number of edges from to in is at least since and for each . Clearly, , which contradicts to . So , and hence .
From the above argument, we have . ∎
Note that . So we have the following.
Proposition 4.1**.**
Let be three integers with and .
* If is odd and , then .*
* If is even and , then .*
5 Concluding Remark
In this paper, we focus our attention on the -good neighbor connectivity of general graphs. We have proved that for . Trees with are characterized in this paper. But the graphs with is still unknown. From Proposition 2.1, the classical is a natural lower bound of , but there is no upper bound of in terms of .
The reference list from the paper itself. Each links out to its DOI / PubMed record.
- 1[1] Q. Cai, X. Li, D. Wu, Some extremal results on the colorful monochromatic vertex-connectivity of a graph, J. Combin. Optim. 35(4) (2018), 1300–1311.
- 2[2] N.-W. Chang and S.-Y. Hsieh, Conditional diagnosability of ( n , k ) 𝑛 𝑘 (n,k) -star graphs under the PMC model, IEEE Trans. Depend. Secure Comput. 15(2) (2018), 207–216.
- 3[3] N.-W. Chang, T.-Y. Lin, and S.-Y. Hsieh, Conditional diagnosability of k 𝑘 k -ary n 𝑛 n -cubes under the PMC model, ACM Trans. Des. Automat. Electr. Syst. 17(4) (2012), 46.
- 4[4] E. Cheng, K. Qiu, Z. Shen, A general approach to deriving the g 𝑔 g -good-neighbor conditional diagnosability of interconnection networks, Theor. Comput. Sci. 757 (2019), 56–67.
- 5[5] A.T. Dahbura, G.M. Masson, An O ( n 2.5 ) 𝑂 superscript 𝑛 2.5 O(n^{2.5}) faulty identification algorithm for diagnosable systems, IEEE Trans. Comput. 33(6) (1984), 486–492.
- 6[6] A.H. Esfahanian, Generalized measures of fault tolerance with application to n 𝑛 n -cube networks, IEEE Trans. Comput. 38(11) (1989), 1586–1591.
- 7[7] A.H. Esfahanian and S.L. Hakimi, On computing a conditional edge-connectivity of a graph, Inform. Process. Lett. 27(4) (1988), 195–199.
- 8[8] J. Fàbrega and M.A. Fiol, Extra connectivity of graphs with large girth, Discrete Math. 127 (1994), 163–170.
