Minimum edge cuts of distance-regular and strongly regular digraphs
S. Ashkboos, G.R. Omidi, F. Shafiei, K. Tajbakhsh

TL;DR
This paper proves that in certain highly symmetric directed graphs, the minimum edge cuts are simple and correspond to edges incident to a single vertex, extending known undirected graph results.
Contribution
It establishes that the edge connectivity equals the valency and characterizes minimum edge cuts in distance-regular and strongly regular digraphs, with new proofs.
Findings
Edge connectivity equals valency k in these digraphs
Minimum edge cuts are edges incident to a single vertex for k>2
Results extend undirected graph properties to directed graphs
Abstract
In this paper, we show that the edge connectivity of a distance-regular digraph with valency is and for , any minimum edge cut of is the set of all edges going into (or coming out of) a single vertex. Moreover we show that the same result holds for strongly regular digraphs. These results extend the same known results for undirected case with quite different proofs.
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
TopicsGraph theory and applications · Interconnection Networks and Systems · Finite Group Theory Research
Minimum edge cuts of distance-regular and strongly regular digraphs
S. Ashkboos, G.R. Omidi, F. Shafiei, K. Tajbakhsh
Department of Electrical and Computer Engineering, Isfahan University of Technology,
Isfahan, 84156-83111, Iran
Department of Mathematical Sciences, Isfahan University of Technology,
Isfahan, 84156-83111, Iran
School of Mathematics, Institute for Research in Fundamental Sciences (IPM),
P.O.Box: 19395-5746, Tehran, Iran
Department of Mathematics, Faculty of Mathematical Sciences, Tarbiat Modares University,
Tehran, 14115-134, Iran
E-mails: [email protected], [email protected],
[email protected], [email protected]
Abstract
In this paper, we show that the edge connectivity of a distance-regular digraph with valency is and for , any minimum edge cut of is the set of all edges going into (or coming out of) a single vertex. Moreover we show that the same result holds for strongly regular digraphs. These results extend the same known results for undirected case with quite different proofs.
Keywords: Distance-regular digraphs, strongly regular digraphs, minimum edge cut, edge connectivity.
AMS subject classification: 05C20, 05E30
11footnotetext: This research is partially carried out in the IPM-Isfahan Branch and in part supported by a grant from IPM (No. 94050217).
1 Introduction
A digraph (or a directed graph) is an ordered pair , where is a set whose elements are called vertices or nodes, and is a set of ordered pairs of vertices, called arcs or edges. In contrast, a graph in which the edges are bidirectional is called an undirected graph. A digraph with no multiple edges or loops (corresponding to a binary adjacency matrix with 0’s on the diagonal) is called simple. Here we only consider finite simple graphs and digraphs. A digraph is called regular with degree (valency) , if the in-degree and the out-degree at each vertex of are equal to . We will denote by (or briefly ) the distance from a vertex to a vertex in a digraph . For every vertex , we define the directed shell (resp. ) to be the set of vertices at distance from (resp. the set of vertices from which is at distance ). The maximum (directed) distance between distinct pairs of vertices is called the diameter of and is denoted by . The girth is the smallest length of a cycle in . In this paper, by a walk, path or cycle, we mean a directed walk, path or cycle. A digraph is (strongly) connected if there is a path between every pair of vertices. For a connected digraph , a set of edges (resp. a set of vertices ) is called an edge-cut (resp. a vertex-cut) if is disconnected. The sizes of the minimum edge-cut and the minimum vertex-cut of a connected digraph are called the edge connectivity and the vertex connectivity of , respectively. For distinct vertices and of , we say that is adjacent to if there is an edge (directed edge) from to . For every and , by and we mean the number of out-neighbors and in-neighbors of in , respectively. For more information on digraphs, we refer the reader to [1]. Throughout this paper, let be a connected simple digraph of order and diameter .
A distance-regular graph is a regular graph such that for any two vertices and at distance , the number of vertices adjacent to and at distance from only depends on and . Distance-regular graphs with diameter two are precisely the strongly regular graphs, which have been studied by several mathematicians [2]. For more background on different concepts of distance-regularity in graphs see [4, 5, 10, 14]. The concept of “distance-regular digraphs” was introduced by Damerell [11]. A digraph with diameter is distance-regular if for every two vertices and with for , the numbers for each with , are independent of the choices of and . Trivial examples of distance-regular digraphs are the directed cycles (the distance-regular digraphs of degree 1). Moreover distance-regular digraphs with girth are precisely the distance-regular graphs. We refer the reader to [13, 17] and the references therein, for more information on the distance-regular digraphs.
If we replace by , in the definition of distance-regularity, we get a new family of digraphs called “weakly distance-regular digraphs”. This concept was introduced by F. Comellas et al. [9], as a generalization of distance-regular digraphs. In fact, distance-regular digraphs are precisely weakly distance-regular digraphs with normal adjacency matrices (a matrix A is normal if , where is the transpose of A). Also, in [9] it has been shown that a digraph of diameter is weakly distance-regular if, for each nonnegative integer , the number of walks of length from a vertex to a vertex only depends on and their distance . Note that in [18], Suzuki and Wang suggested that a “weakly distance-regular digraph” is a digraph with the following property: for all vertices and with , the number of vertices satisfying and depends only on the values . In this paper, we do not assume the Suzuki and Wang’s definition of weakly distance-regular digraphs and we stress that we only consider the mentioned definition that was introduced by F. Comellas et al. in [9].
The weakly distance-regular digraphs with diameter two are the same as the strongly regular digraphs introduced by Duval in [12] as an extension of strongly regular graphs to the directed case. A -regular digraph on vertices is called a strongly regular digraph with parameters if the number of walks of length two between two vertices is , or when these vertices are the same, adjacent, or not adjacent, respectively. The case is the undirected case. On the other extreme, the case , we have tournaments. For more details, we refer to Brouwer’s website [3].
In [8] Brouwer and Mesner showed that the minimum vertex cuts of a given strongly regular graph are the sets of all neighbors of a vertex . In 2005 Brouwer and Haemers proved that a distance-regular graph of degree can not be disconnected by removing fewer than edges and for the only disconnecting sets of edges are the sets of edges on a single vertex (see [6]). The same result for the minimum vertex cuts of distance-regular graphs is obtained by Brouwer and Koolen in [7]. In fact they showed that the vertex-connectivity of a non-complete distance-regular graph of degree equals and when , the only disconnecting sets of vertices of size not more than are the point neighborhoods. The eigenvalue methods are the main tools to obtain the most of the above results. In this paper, we investigate to the minimum edge cuts in distance-regular and strongly regular digraphs and we only use the combinatorial techniques to extend the mentioned results on the minimum edge cuts for directed case.
The paper is organized as follows. In the next section, we show that the edge connectivity of a distance-regular digraph with valency is and if is not an undirected cycle, then any minimum edge cut of is the set of all edges going into (or coming out of) a single vertex. In Section 3, we show that the same result holds for strongly regular digraphs. In fact we prove a strongly regular digraph with valency can not be disconnected by removing fewer than edges and the only disconnecting sets of edges are the sets of edges going into (or coming out of) a single vertex, unless is either an undirected cycle with four or five vertices or the strongly regular digraph with parameters . Note that distance-regular digraphs are precisely weakly distance-regular digraphs with normal adjacency matrices. Also, weakly distance-regular digraphs with diameter 2 are precisely strongly regular digraphs. Based on the above results, in the last section we conjecture that the same result holds for all weakly distance-regular digraphs. Finally, we give an example that shows that the same result does not hold for the minimum vertex cuts and vertex connectivities of strongly regular digraphs (and so for weakly distance-regular digraphs).
2 Distance-regular digraphs
In this section, we investigate to the minimum edge cuts of distance-regular digraphs and we show that the edge connectivity of a distance-regular digraph with valency is and if is not an undirected cycle, then any minimum edge cut of is the set of all edges going into (or coming out of) a single vertex. Note that distance-regular digraphs with girth are precisely distance-regular graphs and due to a result of Brouwer and Haemers in [6], the edge connectivity of a distance-regular graph of degree equals and if the minimum edge cuts of are the sets of all edges crossing a single vertex. On the other hand, distance-regular graphs with degree 2 are precisely the (undirected) cycles. Hence here we only focus on distance-regular digraphs with girth . First we give a useful known result that will be used later on.
Lemma 2.1**.**
([15]) In any edge cut of a regular digraph, the number of edges from to equals the number of edges from to .
We remind that for every two vertices and with of a distance-regular digraph with diameter , the numbers for each with , are independent of the choices of and . Since the adjacency matrix A of a distance-regular digraph is normal, that is, the matrix satisfying , we have for two vertices and with . Here we denote by .
Now we introduce a family of distance-regular digraphs as an extension of trivial examples (the directed cycles). Assume that for two sets and . For , we denote by a digraph with vertex set and edge set . If and is constant, then is a distance-regular digraph with . We denote this family of distance-regular digraphs with by . In the following we will see that the family are exactly all distance-regular digraphs with .
In [11] Damerell showed that every distance-regular digraph with girth is stable; that is, for every two vertices and at distance . Consequently, every distance-regular digraph with girth has diameter (long type) or (short type). Also, he showed that every distance-regular digraph of long type is obtained from a distance-regular digraph of short type by a known construction as follows:
Let be a distance-regular digraph of short type and be an integer. Now let be a digraph, where
[TABLE]
and
[TABLE]
It is easy to see that is a distance-regular digraph of long type with the same girth as . As you see in the following theorem, Damerell showed that the converse is true.
Theorem 2.2**.**
([11]) Every distance-regular digraph of long type is obtained from a distance-regular digraph of short type, of the same girth, by the construction described above. Starting from a distance-regular digraph of long type, a distance-regular digraph of short type is obtained by identifying all antipodal vertices of .
In [16] it is shown that for every non-trivial distance-regular digraph of short type we have . Therefore, the only distance-regular digraphs of short type with are the directed cycles. Now let be the distance-regular digraph of long type that is obtained from a distance-regular digraph of short type by the Damerell’s construction described above. Clearly the parameter for is times of the same parameter for . Therefore, every distance-regular digraph of long type with is obtained from a directed cycle by the Damerell’s construction and thus it is a member of . Hence are precisely all distance-regular digraphs with .
The statement of the following lemma about was shown in [16] for non-trivial distance-regular digraphs of short type. The proof is not correct as stated, although the statement remains valid as we demonstrate. Here we give an alternative way to prove this result for all distance-regular digraphs with .
Lemma 2.3**.**
Let be a distance-regular digraph with diameter , girth and . Then for every , we have .
Proof.
Note that , so we have . First let (this case can happen only when is a distance-regular digraph of long type). Then consider a path of minimum length between two vertices and at distance . Clearly . Using the fact that is stable (which means that for every two vertices and at distance ) we have and so . This fact implies that . Now let . Assume that , and is the digraph induced by . Clearly and for each we have , where and are the distances from to in and , respectively. Now let and be a minimum path in from to . We have , since . First assume that for each . Then and for every . Since , we have , which implies that , this follows from the fact that the adjacency matrix of is normal (this means that , where is the transpose of ). Now let be the minimum number with . Let for every and . Our goal is to show that . Therefore, for every , we have . On the other hand, for each we have and so , which implies that . To show that , consider the integers with maximum such that for each , an integer is the minimum number with . Clearly .
Claim 2.4**.**
For each , we have .
Proof.
We give a proof for Claim 2.4 by induction on . For each , we have and so . Hence our claim holds for . Now assume that the statement of Claim 2.4 holds for an integer , we are going to show that the statement of this claim holds for . That follows from the equality and the fact that for every . ∎
Now using Claim 2.4 for , we have . On the other hand . Therefore and we are done.
∎
The following theorem is the main result of this section.
Theorem 2.5**.**
The edge connectivity of a distance-regular digraph is equal to its valency. Moreover if is not an undirected cycle, then any minimum edge cut of is the set of all edges going into (or coming out of) a single vertex.
Proof.
Clearly each undirected cycle is a distance-regular graph with edge connectivity 2 (equals to its valency). Assume that is a distance-regular digraph with valency and it is not an undirected cycle. If , then is a directed cycle and clearly any minimum edge cut is an edge. Now, assume that . Suppose that is a minimum edge cut of . Since the set of all edges going into (or coming out of) a single vertex is an edge cut, we have .
First let and , where and for each . We are going to show that the minimum edge cut is the set of all edges going into (or coming out of) a single vertex. For simplicity we denote by the digraph induced by on the vertices of . With no loss of generality suppose that there is a vertex with for every vertex . Now let . If , then is the set of all edges coming out of and there is no thing to prove. So assume that and is the set of all vertices such that . Since is an edge cut there is no path from to each in . Since , there are vertices such that and for each . Now let . Clearly and since . On the other hand, , since there is no path from to each in . Hence
[TABLE]
Therefore , which implies that . If , then for each (because of ), a contradiction to the fact that for every vertex . Hence . Since , we have . Therefore , where and . Note that implies that is an undirected graph and so there is no thing to prove due to a result of Brouwer and Haemers in [6]. So and hence for in , the path is a path from to in , a contradiction.
Now let . Note that implies that is an undirected graph and so we are done. Hence we may assume that . Set , where . Clearly every vertex has at least out-neighbors in . It follows that there are at least edges from to and so, we have . Therefore .
If , then and is a set of all edges coming out of a single vertex in and we are done. Now suppose that . So and since , every vertex has exactly one out-neighbor in and out-neighbors in . This implies that and so is a distance-regular graph. As we mentioned for this case the assertion holds due to a result of Brouwer and Haemers in [6]. Hence we may assume that and .
Since we do not have an undirected edge (note that ), we have
[TABLE]
Therefore or . If , then is a set of all edges coming out of a single vertex in and there is no thing to prove. Hence we assume that . Let be the set of all vertices , where and . Clearly and so . Similarly assume that is the set of all vertices , where and . With the same argument, we have and .
Now choose two vertices and . Set . Clearly . Since and , using Lemma 2.3 we have and so , a contradiction to the fact that and . ∎
3 Strongly regular digraphs
As an immediate consequent of a result of Brouwer and Haemers in [6], any minimum edge cut of a given strongly regular graph with valency is a set of all edges crossing a single vertex. Here we show that the same result is correct for the directed case.
Theorem 3.1**.**
The edge connectivity of a strongly regular digraph equals to its valency. Moreover any minimum edge cut of is the set of all edges going into (or coming out of) a single vertex, unless is either an undirected cycle with four or five vertices or the strongly regular digraph with parameters .
Proof.
Assume that is a strongly regular digraph with parameters and is a minimum edge cut of . Clearly each of the mentioned digraphs in Theorem 3.1 is a strongly regular digraph with edge connectivity 2 (equals to its valency). Now assume that is not an undirected cycle with four or five vertices or the strongly regular digraph with parameters . If , then is a complete graph and there is no thing to prove. Therefore we may assume that . Since the set of all edges going into (or coming out of) a single vertex is an edge cut, we have . By Lemma 2.1, we may assume that . Let . Each vertex has at least out-neighbors in , this implies that there are at least edges from to and so . If , then and there is no thing to prove. Now suppose that . Therefore and .
First let . Consider a vertex with . Since for each , we have paths of length 2 from to and so , a contradiction to the fact that .
Now let . Hence every vertex has at least one out-neighbor in . Using , we have and so every vertex in has exactly one out-neighbor in and out-neighbors in . This implies that and . If , then one can easily see that is a complete graph, a contradiction to our assumption that . Hence we may assume that . If , then is a strongly regular graph and for each edge with and , exactly vertices of are adjacent to . Therefore there are at least paths of length two from to for each and so , a contradiction to the fact that . Now let . Assume that is an edge such that and . Since , exactly vertices of are adjacent to . On the other hand since , we have and so . If , then there is at least one vertex such that . Since there are at least paths of length 2 from to , we have , a contradiction. Hence we may assume that .
If , then is a directed triangle and so there is no thing to prove. For , the digraph induced by is a directed cycle of length two and there is no undirected edge between and . It is easy to see that is the strongly regular digraph with parameters , a contradiction to our assumptions. Now let and , where and . Since , exactly vertices of are adjacent to and so there is exactly one vertex such that . On the other hand, there is exactly one vertex such that . Again vertices of are adjacent to and so for only one vertex we have . Therefore , and so , a contradiction. ∎
4 Concluding remarks and open problems
The concept of weakly distance-regular digraphs is an extension of two concepts distance-regular digraphs and strongly regular digraphs. Based on the above results, the investigation to the minimum edge cuts in weakly distance-regular digraphs is an interesting problem. In general undirected cycles are the weakly distance-regular digraphs with a minimum edge cut that is not a vertex out-neighborhoods (or in-neighborhoods). As we mentioned in Sections 3 besides two small undirected cycles, the strongly regular digraph with parameters is a nice exception in strongly regular digraphs with a minimum edge cut that is not a vertex (out-in) neighborhoods. Therefore, besides undirected cycles it is natural to think about the family of infinite weakly distance-regular digraphs, each has a minimum edge cut that is not a vertex out-neighborhoods (or in-neighborhoods). Here we show that such a family exists. In fact for every positive integer , we construct a -regular weakly distance-regular digraph with vertices, diameter such that the statement of Theorem 2.5 does not hold for . To do this, add the edges and for to two disjoint directed cycles and , to get a 2-regular weakly distance-regular digraph with the desired properties. Now based on the previous results and the above discussion we pose the following conjecture about weakly distance-regular digraphs:
Conjecture 4.1**.**
For every weakly distance-regular digraph with valency , the edge connectivity equals to . Moreover if , any minimum edge cut is the set of all edges going into (or coming out of) a single vertex.
As we mentioned in the first section, Brouwer and Koolen in [7] showed that the vertex-connectivity of a non-complete distance-regular graph of degree equals , and the only disconnecting sets of vertices of size not more than are the point neighborhoods. The digraph shown in Figure 1 shows that the same result is not correct for strongly regular digraphs (and so for weakly distance-regular digraphs). As you see in Figure 1, this digraph is a strongly regular digraph with parameters and vertex cut of size 2 (less than its valency). We could not find such an example for distance-regular digraphs. An interesting research problem in this direction is to deduce whether the statement of Brouwer and Koolen’s result is correct for distance-regular digraphs.
The reference list from the paper itself. Each links out to its DOI / PubMed record.
- 1[1] J. Bang-Jensen and G. Gutin. Digraphs. Springer Monographs in Mathematics. Springer-Verlag London Ltd., London, second edition, 2009. Theory, algorithms and applications.
- 2[2] R.C. Bose, Strongly regular graphs, partial geometries and partially balanced designs, Pacific J. Math. 13 (1963) 389–419.
- 3[3] A.E. Brouwer, Parameters of directed strongly regular graphs, http://homepages.cwi.nl/~aeb/math/dsrg/dsrg.html (January 2015).
- 4[4] A.E. Brouwer, A.M. Cohen and A. Neumaier, Distance-Regular Graphs, Springer- Verlag, Berlin-New York, 1989.
- 5[5] A.E. Brouwer and W.H. Haemers, Spectra of Graphs, Springer, 2012; available online at http://homepages.cwi.nl/ ∼ similar-to \sim aeb/math/ipm/.
- 6[6] A.E. Brouwer and W.H. Haemers, Eigenvalues and perfect matchings, Linear Algebra Appl. 395 (2005) 155–162.
- 7[7] A.E. Brouwer and J.H. Koolen, The vertex-connectivity of a distance-regular graph, European J. Combin. 30 (2009) 668–673.
- 8[8] A.E. Brouwer and D.M. Mesner, The connectivity of strongly regular graphs, European J. Combin. 6 (1985) 215–216.
