Next Article in Journal
Evolutionary Optimization for the Belief-Rule-Based System: Method and Applications
Next Article in Special Issue
Fault Tolerant Addressing Scheme for Oxide Interconnection Networks
Previous Article in Journal
Einstein–Yang–Mills-Aether Theory with Nonlinear Axion Field: Decay of Color Aether and the Axionic Dark Matter Production
Previous Article in Special Issue
The Generalized Competition Indices of Doubly Symmetric Primitive Digraphs with d Loops
 
 
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

The kth Local Exponent of Doubly Symmetric Primitive Digraphs with d Loops

College of Sciences, Shanghai Institute of Technology, Shanghai 201418, China
Submission received: 1 July 2022 / Revised: 3 August 2022 / Accepted: 5 August 2022 / Published: 7 August 2022
(This article belongs to the Special Issue Graph Theory and Its Applications)

Abstract

:
Let D be a primitive digraph of order n . The exponent of a vertex x in V ( D ) is denoted γ D ( x ) , which is the smallest integer q such that for any vertex y , there is a walk of length q from x to y. Let V ( D ) = { v 1 , v 2 , , v n } . We order the vertices of V ( D ) so that γ D ( v 1 ) γ D ( v 2 ) γ D ( v n ) is satisfied. Then, for any integer k satisfying 1 k n , γ D ( v k ) is called the kth local exponent of D and is denoted by e x p D ( k ) . Let D S n ( d ) represent the set of all doubly symmetric primitive digraphs with n vertices and d loops, where d is an integer such that 1 d n . In this paper, we determine the upper bound for the kth local exponent of D S n ( d ) , where 1 k n . In addition, we find that the upper bound for the kth local exponent of D S n ( d ) can be reached, where 1 k n .

1. Introduction

Let D = ( V , E ) denote a digraph (directed graph) with n vertices, where the vertex set V = V ( D ) and the arc set E = E ( D ) . Loops are permitted, but multiple arcs are not. A walk from x to y in D , we mean a sequence of vertices x , v 1 , , v t , y where each vertex in the sequence of vertices belongs to V , and a sequence of arcs ( x , v 1 ) , ( v 1 , v 2 ) , , ( v t , y ) where each arc in the sequence of arcs belongs to E , and the vertices and arcs are not necessarily distinct. The number of arcs in W is the length of the walk W . The notation x k y means that there exists a walk of length k from x to y . The distance from vertex x to vertex y in D is written as d D ( x , y ) (for short, d ( x , y ) ), which refers to the length of the shortest walk from x to y . If x = y , then a walk from x to y is a closed walk. A cycle is a closed walk from x to y with distinct vertices except for x = y .
Let x , y be any pair of vertices in a digraph D . The digraph D is called primitive, if there exists a positive integer k such that there is a walk of length k from x to y . This smallest such k is denoted by e x p ( D ) , which is called the exponent of D . The greatest common divisor of the lengths of all the cycles in D is recorded as l ( D ) . It is well known (see [1]) that D is primitive if and only if D is strongly connected and l ( D ) = 1 .
Brualdi and Liu [2] generalized the concept of exponent for a primitive digraph (primitive matrix). Let D be a primitive digraph with n vertices. The exponent of D can be broken down into more local exponents [3]. For any pair of vertices x , z V ( D ) , let γ D ( x , z ) denote the smallest integer p such that there is a walk of length t from x to z, for each integer t p . Since D is a primitive digraph, then γ D ( x , z ) is a finite number. For any vertex x V ( D ) , the exponent of vertex x is written as γ D ( x ) , which is the smallest integer q so that for any vertex y V ( D ) , there exists a walk of length q from x to y . Moreover, for any vertex z V ( D ) and any integer t γ D ( x , z ) , there is a walk of length t from x to z . So, we have q = max { γ D ( x , z ) : z V ( D ) } . Then, for any vertex y V ( D ) , there is a walk of length t from x to y for each integer t q . Therefore, we have
γ D ( x ) = max { γ D ( x , z ) : z V ( D ) } .
Let the vertices of D be ordered as v 1 , v 2 , , v n such that
γ D ( v 1 ) γ D ( v 2 ) γ D ( v n ) .
γ D ( v k ) is called the kth local exponent (generalized exponent) of D , and it is denoted by e x p D ( k ) , where 1 k n . Then,
e x p D ( 1 ) e x p D ( 2 ) e x p D ( n ) .
Furthermore, we have γ ( D ) = max { γ D ( x ) : x V ( D ) } = max { γ D ( x , y ) : x , y V ( D ) } . Obviously, the exponent of D equals e x p D ( n ) . That is, γ ( D ) = e x p ( D ) = e x p D ( n ) . So, for a primitive digraph D , the local exponents of D are generalizations of the exponent of D .
Brualdi and Liu [2] proposed a memoryless communication system. In the memoryless communication system represented by a primitive digraph D of order n , the kth local exponent is the smallest time for each vertex to simultaneously hold all k bits of the information. For more details, please refer to [2,3].
For any vertices x and y of a digraph D , ( x , y ) E ( D ) is an arc if and only if ( y , x ) E ( D ) is an arc, which is represented by x y , then such a digraph D is called a symmetric digraph. An undirected graph (possibly with loops) can be viewed as a symmetric digraph. For some research on undirected graphs, please see [4,5,6]. When D is symmetric, the notation x k y indicates that there is a walk of length k from x to y .
Let D = ( V , E ) be a symmetric digraph, we can regard D as an undirected graph. For convenience, undirected graph terms such as edges, edge set, etc., are used directly to describe a symmetric digraph. Then, let E ( D ) denote the set of undirected edges (edges) in D . Moreover, we assume that the notation x , y E ( D ) represents that there is an edge in D with x , y as end vertices.
Let D = ( V , E ) be a symmetric digraph, where V = { v 1 , v 2 , , v n } . If for any vertices v i and v j , [ v i , v j ] E ( D ) if and only if [ v n + 1 i , v n + 1 j ] E ( D ) , then such a symmetric digraph D is called a doubly symmetric digraph. Moreover, [ v i , v j ] and [ v n + 1 i , v n + 1 j ] are called a pair of symmetrical edges, or [ v i , v j ] is a symmetrical edge of [ v n + 1 i , v n + 1 j ] , where 1 i n and 1 j n . The vertices v n + 1 i , v i are called a pair of symmetric vertices, or v i is a symmetric vertex of v n + 1 i , where 1 i n . According to this definition, when n is odd, v n + 1 2 is symmetric to itself. If v i is a loop vertex, then [ v i , v i ] E ( D ) and [ v n + 1 i , v n + 1 i ] E ( D ) . Therefore, for i n + 1 i , if [ v i , v i ] is a loop, then [ v n + 1 i , v n + 1 i ] is also a loop, the loops appear in pairs. A doubly symmetric digraph D is called a doubly symmetric primitive digraph provided D is primitive.
If a doubly symmetric primitive digraph D contains exactly d loops, then we call D a doubly symmetric primitive digraph with d loops. Let D S n denote the set of all doubly symmetric primitive digraphs of order n. Let D S n ( d ) denote the set of all doubly symmetric primitive digraphs of order n with d loops, where d is an integer such that 1 d n . Obviously, we have D S n ( d ) D S n .
Let D D S n ( d ) . After deleting any pair of symmetrical edges [ v i , v j ] and [ v n + 1 i , v n + 1 j ] of D , the obtained digraph D is not a doubly symmetric primitive digraph (that is, D is not connected), then we call D D S n ( d ) , where 1 i < j n . Obviously, we have D S n ( d ) D S n ( d ) .
For example, we consider the kth local exponent of the graph G . Let V ( G ) = { v 1 , v 2 , , v 7 } . Let E ( G ) = { [ v i , v i + 1 ] | 1 i 6 } { [ v 4 , v 4 ] } . G is shown in Figure 1.
We easily get γ G ( v 4 ) = 3 , γ G ( v 3 ) = γ G ( v 5 ) = 4 , γ G ( v 2 ) = γ G ( v 6 ) = 5 , γ G ( v 1 ) = γ G ( v 7 ) = 6 . Then, we have e x p G ( 1 ) = 3 , e x p G ( 2 ) = e x p G ( 3 ) = 4 , e x p G ( 4 ) = e x p G ( 5 ) = 5 , e x p G ( 6 ) = e x p G ( 7 ) = 6 . Moreover, we have γ ( G ) = e x p ( G ) = e x p G ( 7 ) = 6 .
Some studies [7,8,9,10,11,12] have investigated exponents and their generalization. Chen and Liu [11] studied the kth local exponent of doubly symmetric primitive matrices (primitive digraphs). Chen and Liu [12] characterized the doubly symmetric primitive digraphs with the kth local exponent reaching the maximum value. A doubly symmetric primitive digraph with d loops is a special doubly symmetric primitive digraph. It is important to mention that the kth local exponent of such a class of digraphs has not been studied before. Using graph theory methods, we obtain the upper bound of the kth local exponent of digraphs in D S n ( d ) , where 1 k n . Some studies have investigated the scrambling index [13,14,15,16] and generalized competition index [17,18,19,20,21,22,23]. Several studies explored the generalized μ -scrambling indices, please refer to [24,25,26].
Let D D S n ( d ) . Let V ( L ( D ) ) represent the set of d loop vertices in D . Let E ( L ( D ) ) denote the set of d loops in D . Let v i , v j be any pair of vertices of the digraph D . If the walk from v i to v j in D is denoted as W D ( v i , v j ) (for short, W ( v i , v j ) ), then | W ( v i , v j ) | is used to denote the length of the walk W ( v i , v j ) , and V ( W ( v i , v j ) ) is used to denote the set of all vertices in this walk W ( v i , v j ) . If there is a unique path from v i to v j in D , then let P D ( v i , v j ) (for short, P ( v i , v j ) ) denote the unique path, and let V ( P D ( v i , v j ) ) (for short, V ( P ( v i , v j ) ) ) denote the set of all vertices on the path. If v i = v j , then V ( P ( v i , v j ) ) = { v i } = { v j } . If a walk W ( v i , v j ) from v i to v j in D does not pass through a loop vertex, then let V ( W ( v i , v j ) ) V ( L ( D ) ) = , otherwise V ( W ( v i , v j ) ) V ( L ( D ) ) . Similarly, if the unique path from v i to v j passes through a loop vertex, that is V ( P ( v i , v j ) ) V ( L ( D ) ) , otherwise V ( P ( v i , v j ) ) V ( L ( D ) ) = .
For a vertex v V ( D ) and a set X V ( D ) , let d ( v , X ) = min { d ( v , v i ) : v i X } . If v X , let d ( v , X ) = 0 . For any vertex u V ( D ) and v V ( D ) , if u = v , let d ( u , v ) = 0 . If T is a set, the notation | T | is used to denote the number of all elements in T . The notation a is used to denote the largest integer not greater than a , and the notation b is used to denote the smallest integer not less than b .
In this paper, let n , d and k be integers with n 5 , 1 k n , 1 d n . We give the upper bound of the kth local exponent of digraphs in D S n ( d ) , where 1 k n .

2. The Upper Bound for the k th Local Exponent of DS n ( d )

In this section, let D = ( V , E ) , where V = { v 1 , v 2 , , v n } .
In the case of D D S n ( d ) , we observe the exponent of any vertex in D , it is easy to get the following Proposition 1, let us omit the proof.
Proposition 1.
Let D D S n ( d ) and let v i be any vertex of D , then γ D ( v i ) = γ D ( v n + 1 i ) , where 1 i n .
Lemma 1
(Lemma 3.3 [2]). Let D be a primitive digraph with n vertices. Then, e x p D ( k + 1 ) e x p D ( k ) + 1 , where 1 k n 1 .
Remark 1.
Lemma 1 is actually very useful. Next, we repeat the proof of Brualdi and Liu (see [2]). Since D is strongly connected, for any integer k such that 1 k n 1 , there is a vertex x that is joined by an arc to one of the vertices with the k smallest exponents. Therefore, e x p D ( k + 1 ) e x p D ( k ) + 1 , where 1 k n 1 .
Lemma 2.
Let D D S n ( d ) and let v i , v j be any pair of vertices of D . Then, γ D ( v j ) γ D ( v i ) + d ( v i , v j ) .
Proof. 
For any vertex x V ( D ) , there is a walk of length t from v i to x, which is v i t x , for each integer t γ D ( v i ) . So, there is a walk of length s from v j to x, which is v j d ( v j , v i ) v i t x , for each integer s γ D ( v i ) + d ( v i , v j ) . Therefore, γ D ( v j ) γ D ( v i ) + d ( v i , v j ) .
Lemma 3.
Let D D S n ( d ) and let x , y be any pair of vertices of D . If there exists a walk W ( x , y ) from x to y such that V ( W ( x , y ) ) V ( L ( D ) ) , then γ D ( x , y ) | W ( x , y ) | .
Proof. 
Let h = | W ( x , y ) | . We consider the following.
Case 1 x V ( D ) \ V ( L ( D ) ) and y V ( D ) \ V ( L ( D ) ) .
Suppose a walk from x to y through a loop vertex is denoted by x a v i h a y , where v i is a loop vertex and a is an integer such that 1 a h 1 . Then, the length of the walk x a v i 1 v i h a y is h + 1 . The length of the walk x a v i 1 v i 1 v i h a y is h + 2 . Similarly, we can easily conclude that there is a walk of length s from x to y, for each integer s h . So, γ D ( x , y ) | W ( x , y ) | .
Case 2 x V ( L ( D ) ) or y V ( L ( D ) ) .
Similar to Case 1, it is easy to get that there is a walk of length s from x to y, for each integer s h . So, γ D ( x , y ) | W ( x , y ) | .
Therefore, the lemma holds. □
Lemma 4
(Lemma 1 [23]). Let D D S n . If n is odd and x is any vertex of D, then d ( x , v n + 1 2 ) n 1 2 .
Theorem 1.
Let D D S n ( d ) . If n is odd and d is odd, then e x p D ( k ) n 1 2 + k 2 , where 1 k n .
Proof. 
If d is odd, then v n + 1 2 is a loop vertex. Let x be any vertex. Then, a shortest path from x to v n + 1 2 goes through the loop vertex v n + 1 2 . According to Lemma 4, we have d ( x , v n + 1 2 ) n 1 2 . Furthermore, according to Lemma 3, we have γ D ( v n + 1 2 , x ) d ( x , v n + 1 2 ) . Further, we have γ D ( v n + 1 2 , x ) = d ( x , v n + 1 2 ) . So, γ D ( v n + 1 2 ) = max { d ( v n + 1 2 , x ) : x V ( D ) } n 1 2 . Then, we have e x p D ( 1 ) γ D ( v n + 1 2 ) n 1 2 . Further, according to Proposition 1, we have γ D ( v 1 ) = γ D ( v n ) , γ D ( v 2 ) = γ D ( v n 1 ) , , γ D ( v n 1 2 ) = γ D ( v n + 3 2 ) . So, according to Lemma 1, we conclude e x p D ( 2 ) = e x p D ( 3 ) e x p D ( 1 ) + 1 n 1 2 + 1 , e x p D ( 4 ) = e x p D ( 5 ) e x p D ( 1 ) + 2 n 1 2 + 2 , , e x p D ( n 1 ) = e x p D ( n ) e x p D ( 1 ) + n 1 2 n 1 2 + n 1 2 . Therefore, we have e x p D ( k ) e x p D ( 1 ) + k 2 n 1 2 + k 2 , where 1 k n .
Let D D S n ( d ) and D D S n ( d ) . If D is a subgraph of D such that V ( D ) = V ( D ) and E ( D ) E ( D ) , then e x p D ( k ) e x p D ( k ) , where 1 k n . So, if we investigate the upper bound of the kth local exponent of digraphs in D S n ( d ) , we only need to investigate the digraphs in D S n ( d ) .
Referring to Definition 3 in [23], we give the following Definition 1.
Definition 1.
Let D D S n ( d ) , where n is odd, d is even such that d 2 . There exist two connected subgraphs D * = ( V * , E * ) and D * * = ( V * * , E * * ) of D , and D * , D * * satisfy V ( D ) = V * V * * and E ( D ) = E * E * * E ( L ( D ) ) . Where V * = { v n + 1 i : v i V * * } and E * = { [ v n + 1 i , v n + 1 j ] : [ v i , v j ] E * * } . Moreover, | V * | = | V * * | = n + 1 2 , | E * | = | E * * | = n 1 2 .
Remark 2.
Suppose n is odd and d is even that satisfies d 2 . If D D S n ( d ) , then there are two connected subgraphs D * and D * * of D . In addition, there is a unique path for any two different vertices in D * and D * * , respectively. Moreover, there is a unique path for any two different vertices in D . Let x , y be any pair of vertices of D such that x V * and y V * * , then V ( P ( x , v n + 1 2 ) ) V ( P ( v n + 1 2 , y ) ) = { v n + 1 2 } (see [23]). After removing d loops from D , the obtained graph is a tree. Therefore, D is a special tree with loops that satisfies [ v i , v j ] E ( D ) if and only if [ v n + 1 i , v n + 1 j ] E ( D ) , where 1 i < j n .
Lemma 5
(Lemma 3 [23]). Let D D S n ( d ) , where n is odd, d is even such that d 2 . Let x , y be any pair of vertices of D such that V ( P ( x , v n + 1 2 ) ) V ( L ( D ) ) = and V ( P ( y , v n + 1 2 ) ) V ( L ( D ) ) = . Then, there is a walk W ( x , y ) from x to y such that V ( W ( x , y ) ) V ( L ( D ) ) , and | W ( x , y ) | n d + 1 .
Corollary 1.
Let D D S n ( d ) , where n is odd, d is even and d 2 . Let x , y be any pair of vertices of D satisfying V ( P ( x , v n + 1 2 ) ) V ( L ( D ) ) = and V ( P ( y , v n + 1 2 ) ) V ( L ( D ) ) = . Then, γ D ( x , y ) n d + 1 .
Proof. 
According to Lemma 5, there is a walk W ( x , y ) from x to y passing through a loop vertex, and | W ( x , y ) | n d + 1 . Moreover, according to Lemma 3, we have γ D ( x , y ) | W ( x , y ) | n d + 1 .
In Corollary 1, if x = v n + 1 2 and x isn’t a loop vertex, then V ( P ( x , v n + 1 2 ) ) = { v n + 1 2 } , we have V ( P ( x , v n + 1 2 ) ) V ( L ( D ) ) = .
Theorem 2.
Let D D S n ( d ) . If n is odd, d is even and d 2 , then
(1)
If n 2 d 3 , then e x p D ( k ) n 1 2 + k 2 , where 1 k n .
(2)
If n 2 d 1 , then
e x p D ( k ) n d + 1 , where 1 k n 2 d + 4 , n d + 1 + k ( n 2 d + 4 ) 2 , where n 2 d + 4 k n .
Proof. 
We only need to consider D D S n ( d ) . Since d is even, then v n + 1 2 isn’t a loop vertex. Let x be any vertex. By Lemma 4, we have d ( x , v n + 1 2 ) n 1 2 . If the path from v n + 1 2 to x passes through a loop vertex, that is V ( P ( v n + 1 2 , x ) ) V ( L ( D ) ) , then γ D ( v n + 1 2 , x ) = d ( x , v n + 1 2 ) n 1 2 . If vertex x satisfies V ( P ( v n + 1 2 , x ) ) V ( L ( D ) ) = , according to Corollary 1, we have γ D ( v n + 1 2 , x ) n d + 1 . Therefore, we have γ D ( v n + 1 2 ) max { n 1 2 , n d + 1 } .
(1)
If n 2 d 3 , then n d + 1 n 1 2 . Then, we have e x p D ( 1 ) γ D ( v n + 1 2 ) n 1 2 . According to Proposition 1, we have γ D ( v 1 ) = γ D ( v n ) , γ D ( v 2 ) = γ D ( v n 1 ) , , γ D ( v n 1 2 ) = γ D ( v n + 3 2 ) . Then, according to Lemma 1, we can conclude that e x p D ( 2 ) = e x p D ( 3 ) e x p D ( 1 ) + 1 n 1 2 + 1 ,   e x p D ( 4 ) = e x p D ( 5 ) e x p D ( 1 ) + 2 n 1 2 + 2 , . Therefore, we have e x p D ( k ) e x p D ( 1 ) + k 2 n 1 2 + k 2 , where 1 k n .
(2)
If n 2 d 1 , then n d + 1 n + 1 2 . We have e x p D ( 1 ) γ D ( v n + 1 2 ) n d + 1 .
Next, we construct a set V ( M ) such that | V ( M ) | n 2 d + 4 , and for any vertex v i V ( M ) , γ D ( v i ) n d + 1 holds. Suppose V ( M ) = { v : d ( v , v n + 1 2 ) n 2 d + 3 2 } . Then, v n + 1 2 V ( M ) .
Suppose v h V ( M ) V * and v h v n + 1 2 . If the vertex sequence of the unique path in D * from v h to v n + 1 2 is v h , , v t , , v n + 1 2 , then the vertex sequence of the unique path in D * * from v n + 1 h to v n + 1 2 is v n + 1 h , , v n + 1 t , , v n + 1 2 . Moreover, V ( P ( v h , v n + 1 2 ) ) V ( P ( v n + 1 h , v n + 1 2 ) ) = { v n + 1 2 } . So, d ( v h , v n + 1 2 ) = d ( v n + 1 h , v n + 1 2 ) n 2 d + 3 2 . Therefore, we have v n + 1 h V ( M ) . Next, we prove that | V ( M ) | n 2 d + 4 . For any vertex v s V ( D ) , if v s V ( M ) , then | V ( M ) | = n . If there is a vertex v l satisfying v l V ( D ) \ V ( M ) , then d ( v l , v n + 1 2 ) n 2 d + 3 2 + 1 . So, V ( P ( v l , v n + 1 2 ) ) V ( M ) = n 2 d + 3 2 + 1 . Moreover, V ( P ( v n + 1 l , v n + 1 2 ) ) V ( M ) = n 2 d + 3 2 + 1 and V ( P ( v l , v n + 1 2 ) ) V ( P ( v n + 1 l , v n + 1 2 ) ) = { v n + 1 2 } . Then, we have | V ( P ( v l , v n + 1 2 ) ) V ( P ( v n + 1 l , v n + 1 2 ) ) V ( M ) | = n 2 d + 4 . Therefore, we have | V ( M ) | n 2 d + 4 . For any vertex v i V ( M ) , next we consider γ D ( v i ) .
For any vertex v i V ( M ) , let the walk V ( W ( v i , x ) ) from v i to x be v i d ( v i , v n + 1 2 ) v n + 1 2 d ( v n + 1 2 , x ) x . If the walk V ( W ( v i , x ) ) passes through a loop vertex, we have γ D ( v i , x ) d ( v i , v n + 1 2 ) + d ( v n + 1 2 , x ) n 2 d + 3 2 + n 1 2 = n d + 1 . If the walk V ( W ( v i , x ) ) doesn’t pass through a loop vertex, that is, V ( W ( v i , x ) ) V ( L ( D ) ) = . Then, we have V ( P ( v i , v n + 1 2 ) ) V ( L ( D ) ) = and V ( P ( x , v n + 1 2 ) ) V ( L ( D ) ) = . So, if V ( W ( v i , x ) ) V ( L ( D ) ) = , according to Corollary 1, we have γ D ( v i , x ) n d + 1 . Therefore, we have γ D ( v i ) n d + 1 . According to Proposition 1, we have γ D ( v n + 1 i ) = γ D ( v i ) n d + 1 . Therefore, we have e x p D ( k ) n d + 1 , where 1 k | V ( M ) | .
For any vertex v j such that v j V ( D ) \ V ( M ) , then v n + 1 j V ( D ) \ V ( M ) . There is a unique path for any pair of vertices in D . So for any vertex v i V ( M ) , we have d ( v j , v i ) = d ( v n + 1 j , v n + 1 i ) . Suppose d ( v j , V ( M ) ) = d ( v j , v l ) , where v l V ( M ) . We have d ( v n + 1 j , v n + 1 l ) = d ( v j , v l ) d ( v j , v i ) = d ( v n + 1 j , v n + 1 i ) . So, we have d ( v j , V ( M ) ) = d ( v n + 1 j , V ( M ) ) . Furthermore, according to Lemma 2, we can easily conclude that γ D ( v j ) = γ D ( v n + 1 j ) n d + 1 + d ( v j , V ( M ) ) . Since | V ( M ) | n 2 d + 4 , the conclusion is clearly established.
Therefore, the theorem holds. □
Referring to Definition 4 in [23], we give the following Definition 2.
Definition 2.
Let D D S n ( d ) , where n is even, d is even such that d 2 .
(1)
There exist two connected subgraphs D 1 = ( V 1 , E 1 ) and D 2 = ( V 2 , E 2 ) of D , and D 1 , D 2 satisfy V ( D ) = V 1 V 2 and E ( D ) = E 1 E 2 { [ v f , v n + 1 g ] } { [ v n + 1 f , v g ] } E ( L ( D ) ) . Where V 1 = { v n + 1 i : v i V 2 } and E 1 = { [ v n + 1 i , v n + 1 j ] : [ v i , v j ] E 2 } , v f V 1 and v g V 1 . Moreover, | V 1 | = | V 2 | = n 2 , | E 1 | = | E 2 | = n 2 1 .
(2)
If { v f , v n + 1 f } V ( L ( D ) ) = . Let V ( H 1 ) = { v : V ( P D 1 ( v f , v ) ) V ( L ( D ) ) = , where v V 1 } and V ( H 2 ) = { v : V ( P D 2 ( v n + 1 f , v ) ) V ( L ( D ) ) = , where v V 2 } . Suppose V ( H ) = V ( H 1 ) V ( H 2 ) .
Definition 3.
Let D D S n ( d ) , where n is even, d is even such that d 2 . In Definition 2(1), we give the following definition:
(1)
Let W ( v f , v f ) be v f d D 1 ( v f , v g ) v g d D ( v g , v n + 1 f ) v n + 1 f d D 2 ( v n + 1 f , v n + 1 g ) v n + 1 g
d D ( v n + 1 g , v f ) v f , then W ( v f , v f ) is a closed walk from v f to v f . Let us write V ( W ( v f , v f ) ) = V ( R ) .
(2)
If f g , let D D S n , 1 ( d ) . If f = g , let D D S n , 2 ( d ) .
Remark 3.
Suppose n is even and d is even that satisfies d 2 . If D D S n ( d ) , then there are two connected subgraphs D 1 and D 2 of D . Moreover, there is a unique path for any two different vertices in D 1 and D 2 , respectively. Since D is connected, then there are edges [ v f , v n + 1 g ] E ( D ) and [ v n + 1 f , v g ] E ( D ) . Then, d D ( v g , v n + 1 f ) = d D ( v n + 1 g , v f ) = 1 . If v f and v n + 1 f are not loop vertices, then v f V ( H ) and v n + 1 f V ( H ) . If D D S n , 1 ( d ) , then f g , and | V ( R ) | is even such that | V ( R ) | 4 . Furthermore, if D D S n , 1 ( d ) , after removing d loops from D , the obtained graph D * is not a tree. According to Definitions 2 and 3, it is not difficult to see that | V ( D * ) | = n and | E ( D * ) | = n . If D D S n , 2 ( d ) , then f = g , V ( R ) = { v f , v n + 1 f } and | V ( R ) | = 2 . If D D S n , 2 ( d ) , then D is a special tree with loops that satisfies [ v i , v j ] E ( D ) if and only if [ v n + 1 i , v n + 1 j ] E ( D ) , where 1 i < j n . In fact, D D S n , 2 ( d ) can be regarded as a special case of f = g in D D S n , 1 ( d ) .
In Lemma 2 in [23], let D D S n ( d ) , where n is even and d be even such that d 2 . We can directly get the following Lemma 6.
Lemma 6.
Let D D S n ( d ) . Let n be even and d be even such that d 2 . Let x be any vertex of D . Then, for any vertex v s V ( R ) , we have d ( x , v s ) n 2 .
Lemma 7
(Lemma 5 [23]). Let D D S n ( d ) , where n is even, d is even and d 2 . Let x , y be any pair of vertices of D such that x , y V ( H ) . If V ( R ) V ( L ( D ) ) = , then there exists a walk W ( x , y ) from x to y such that V ( W ( x , y ) ) V ( L ( D ) ) , and | W ( x , y ) | n d + 1 .
Corollary 2.
Let D D S n ( d ) , where n is even, d is even such that d 2 . Let x , y be any pair of vertices of D satisfying x , y V ( H ) . If V ( R ) V ( L ( D ) ) = , then γ D ( x , y ) n d + 1 .
Proof. 
According to Lemma 7, there is a walk W ( x , y ) from x to y passing through a loop vertex, and | W ( x , y ) | n d + 1 . Furthermore, according to Lemma 3, we have γ D ( x , y ) | W ( x , y ) | n d + 1 .
Theorem 3.
Let D D S n ( d ) . If n is even, d is even and d 2 , then
(1)
If n 2 d 2 , then e x p D ( k ) n 2 1 + k 2 , where 1 k n .
(2)
If n 2 d , then
e x p D ( k ) n d + 1 , where 1 k n 2 d + 4 , n d + 1 + k ( n 2 d + 4 ) 2 , where n 2 d + 4 k n .
Proof. 
We only need to consider D D S n ( d ) . Let x be any vertex. Let us consider the following two cases.
Case 1 D D S n , 1 ( d ) .
Since v f V 1 and v g V 1 , then v n + 1 f V 2 and v n + 1 g V 2 .
Case 1.1 V ( R ) V ( L ( D ) ) .
Suppose { v m , v n + 1 m } V ( R ) V ( L ( D ) ) . Then, v m and v n + 1 m are loop vertices. According to Lemma 6, we have d ( x , v m ) n 2 . Further, γ D ( v m ) = γ D ( v n + 1 m ) n 2 . Then, e x p D ( 1 ) = e x p D ( 2 ) γ D ( v m ) n 2 . Therefore, according to Proposition 1 and Lemma 1, we have e x p D ( k ) n 2 1 + k 2 , where 1 k n .
(1)
If n 2 d 2 , then the conclusion is clearly established.
(2)
If n 2 d , for 1 k n 2 d + 4 , then e x p D ( k ) n 2 1 + k 2 n 2 1 + n 2 d + 2 = n d + 1 . For n 2 d + 4 k n , we have e x p D ( k ) n 2 1 + k 2 = n 2 1 + k ( n 2 d + 4 ) 2 + n 2 d + 2 = n d + 1 + k ( n 2 d + 4 ) 2 .
Case 1.2 V ( R ) V ( L ( D ) ) = .
Then, v f V ( H ) V ( R ) , v n + 1 f V ( H ) V ( R ) .
For x V ( H ) , according to Corollary 2, we have γ D ( v f , x ) n d + 1 .
For any vertex x V 1 \ V ( H ) , then V ( P D 1 ( v f , x ) ) V ( L ( D ) ) . We have γ D ( v f , x ) d D 1 ( v f , x ) n 2 .
For any vertex x V 2 \ V ( H ) , then V ( P D 2 ( v n + 1 f , x ) ) V ( L ( D ) ) . Let the walk W D ( v f , x ) be v f d D ( v f , v n + 1 g ) v n + 1 g d D 2 ( v n + 1 g , x ) x . Since V ( R ) V ( L ( D ) ) = , then V ( P D 2 ( v n + 1 f , v n + 1 g ) ) V ( L ( D ) ) = . In addition, V ( P D 2 ( v n + 1 f , x ) ) V ( L ( D ) ) . We have V ( P D 2 ( v n + 1 g , x ) ) V ( L ( D ) ) . So V ( W D ( v f , x ) ) V ( L ( D ) ) . Moreover, | W D ( v f , x ) | = 1 + d D 2 ( v n + 1 g , x ) n 2 . We have γ D ( v f , x ) | W D ( v f , x ) | n 2 .
Therefore, we have γ D ( v f ) = γ D ( v n + 1 f ) max { n 2 , n d + 1 } .
(1)
If n 2 d 2 , then n d + 1 n 2 . Then, we have e x p D ( 1 ) = e x p D ( 2 ) γ D ( v f ) n 2 . Therefore, according to Proposition 1 and Lemma 1, e x p D ( k ) n 2 1 + k 2 , where 1 k n .
(2)
If n 2 d , then n d + 1 n 2 + 1 . We have e x p D ( 1 ) = e x p D ( 2 ) γ D ( v f ) n d + 1 . Next, we construct a set V ( L 1 ) such that | V ( L 1 ) | n 2 d + 4 , and for any vertex v i V ( L 1 ) , γ D ( v i ) n d + 1 holds. Let V ( L 11 ) = { v : d D 1 ( v , v f ) n 2 d + 2 2 , where v V 1 } . Let V ( L 12 ) = { v : d D 2 ( v , v n + 1 f ) n 2 d + 2 2 , where v V 2 } . Suppose V ( L 1 ) = V ( L 11 ) V ( L 12 ) . Then, v f V ( L 1 ) and v n + 1 f V ( L 1 ) . Suppose v h V ( L 11 ) and v h v f . If the vertex sequence of the unique path in D 1 from v h to v f is v h , , v t , , v f , then the vertex sequence of the unique path in D 2 from v n + 1 h to v n + 1 f is v n + 1 h , , v n + 1 t , , v n + 1 f . So, d D 1 ( v h , v f ) = d D 2 ( v n + 1 h , v n + 1 f ) n 2 d + 2 2 . Therefore, we have v n + 1 h V ( L 1 ) . Next, we prove that | V ( L 1 ) | n 2 d + 4 . For any vertex v s V ( D ) , if v s V ( L 1 ) , then | V ( L 1 ) | = n . If there is a vertex v l satisfying v l V ( D ) \ V ( L 1 ) , we might as well assume v l V 1 . Then d D 1 ( v l , v f ) n 2 d + 2 2 + 1 . So, V ( P D 1 ( v l , v f ) ) V ( L 11 ) = n 2 d + 2 2 + 1 . Moreover, V ( P D 2 ( v n + 1 l , v n + 1 f ) ) V ( L 12 ) = n 2 d + 2 2 + 1 . Then, | V ( P D 1 ( v l , v f ) ) V ( P D 2 ( v n + 1 l , v n + 1 f ) ) V ( L 1 ) | = n 2 d + 4 . Therefore, we have | V ( L 1 ) | n 2 d + 4 . Let us assume v i V ( L 11 ) . Next, we consider γ D ( v i ) .
Case 1.2.1 x V 1 .
Let W D 1 ( v i , x ) be the walk from v i to x , which is v i d D 1 ( v i , v f ) v f d D 1 ( v f , x ) x . If the walk W D 1 ( v i , x ) passes through a loop vertex, we have | W D 1 ( v i , x ) | = d D 1 ( v i , v f ) + d D 1 ( v f , x ) n 2 d + 2 2 + n 2 = n d + 1 . If the walk W D 1 ( v i , x ) does not pass through a loop vertex, we have x V ( H ) and v i V ( H ) . According to Corollary 2, then γ D ( v i , x ) n d + 1 .
Case 1.2.2 x V 2 .
Let the walk W D ( v i , x ) be v i d D 1 ( v i , v f ) v f d D ( v f , v n + 1 g ) v n + 1 g d D 2 ( v n + 1 g , x ) x . If the walk W D ( v i , x ) passes through a loop vertex, we have | W D ( v i , x ) | = d D 1 ( v i , v f ) + 1 + d D 2 ( v n + 1 g , x ) n 2 d + 2 2 + n 2 = n d + 1 . If the walk W D ( v i , x ) does not pass through a loop vertex. Then, v i V ( H ) . Since V ( R ) V ( L ( D ) ) = , then V ( P D 2 ( v n + 1 f , v n + 1 g ) ) V ( L ( D ) ) = . Moreover, V ( P D 2 ( v n + 1 g , x ) ) V ( L ( D ) ) = , we have V ( P D 2 ( v n + 1 f , x ) ) V ( L ( D ) ) = . So, we have x V ( H ) . Since x V ( H ) and v i V ( H ) , according to Corollary 2, then γ D ( v i , x ) n d + 1 .
Therefore, whether x V 1 or x V 2 , we have γ D ( v i , x ) n d + 1 . So γ D ( v i ) n d + 1 . According to Proposition 1, we have γ D ( v n + 1 i ) = γ D ( v i ) n d + 1 . Therefore, we have e x p D ( k ) n d + 1 , where 1 k | V ( L 1 ) | . For any vertex v j such that v j V 1 \ V ( L 11 ) , then v n + 1 j V 2 \ V ( L 12 ) . We have d D 1 ( v j , V ( L 11 ) ) = d D 2 ( v n + 1 j , V ( L 12 ) ) . Furthermore, according to Lemma 2, we can conclude that γ D ( v j ) = γ D ( v n + 1 j ) n d + 1 + d D 1 ( v j , V ( L 11 ) ) . Since | V ( L 1 ) | n 2 d + 4 , the conclusion is clearly established.
Case 2 D D S n , 2 ( d ) .
For D D S n , 2 ( d ) , it is equivalent to f = g in D D S n , 1 ( d ) , let us omit it. □

3. The k th Local Exponents of H * and H * *

In this section, we study the kth local exponents of the special graphs H * and H * * .
Definition 4.
Suppose 1 d n . Let V ( H * ) = { v 1 , v 2 , , v n } and E ( H * ) = { [ v i , v i + 1 ] | 1 i n 1 } E ( L ( H * ) ) , where E ( L ( H * ) ) are d loops arranged arbitrarily such that H * D S n ( d ) .
Definition 5.
Suppose d is even and d 2 . Let V ( H * * ) = { v 1 , v 2 , , v n } and let E ( H * * ) = { [ v i , v i + 1 ] | 1 i n 1 } { [ v i , v i ] : where 1 i d 2 and n d 2 + 1 i n } .
From the definition of H * and H * * , we know that H * D S n ( d ) and H * * D S n ( d ) . Furthermore, there is a unique path for any two different vertices in H * and H * * , respectively.
Remark 4.
In Figure 2, n can be either odd or even. In Figure 3, since n is odd and n 2 d 1 , then v n + 1 2 is not a loop vertex. In Figure 4, since n is even and n 2 d , then v n 2 and v n 2 + 1 are not loop vertices.
Theorem 4.
If n is odd and d is odd, then e x p H * ( k ) = n 1 2 + k 2 , where 1 k n .
Proof. 
Since d is odd, then v n + 1 2 is a loop vertex. We have γ H * ( v n + 1 2 ) = max { d ( v n + 1 2 , x ) : x V ( H * ) } = d ( v n + 1 2 , v n ) = n 1 2 . Moreover, according to Lemma 2, we have γ H * ( v n + 1 2 a ) γ H * ( v n + 1 2 ) + d ( v n + 1 2 a , v n + 1 2 ) = n 1 2 + a , where 1 a n 1 2 . Further, γ H * ( v n + 1 2 a ) = γ H * ( v n + 1 2 + a ) = γ H * ( v n + 1 2 a , v n ) = d ( v n + 1 2 a , v n ) = n 1 2 + a , where 1 a n 1 2 . Therefore, the theorem now holds. □
Theorem 5.
If n is odd, d is even and d 2 , then
(1)
If n 2 d 3 , then e x p H * ( k ) = n 1 2 + k 2 , where 1 k n .
(2)
n 2 d 1 , then
e x p H * * ( k ) = n d + 1 , where 1 k n 2 d + 4 , n d + 1 + k ( n 2 d + 4 ) 2 , where n 2 d + 4 k n .
Proof. 
Since d is even, then v n + 1 2 is not a loop vertex.
(1)
H * is shown in Figure 2. Let x be any vertex of H * . If V ( P ( v n + 1 2 , x ) ) V ( L ( H * ) ) , then γ H * ( v n + 1 2 , x ) = d ( v n + 1 2 , x ) n 1 2 . If V ( P ( v n + 1 2 , x ) ) V ( L ( H * ) ) = , according to Corollary 1, we have γ H * ( v n + 1 2 , x ) n d + 1 . Then, γ H * ( v n + 1 2 ) max { n 1 2 , n d + 1 } . If n 2 d 3 , then n d + 1 n 1 2 . Then, we have γ H * ( v n + 1 2 ) n 1 2 . Further, γ H * ( v n + 1 2 , v n ) = d ( v n + 1 2 , v n ) = n 1 2 . Therefore, γ H * ( v n + 1 2 ) = n 1 2 . Let a be an integer satisfying 1 a n 1 2 . According to Lemma 2, we have γ H * ( v n + 1 2 a ) d ( v n + 1 2 a , v n + 1 2 ) + γ H * ( v n + 1 2 ) = a + n 1 2 . Further, we have γ H * ( v n + 1 2 a , v n ) = d ( v n + 1 2 a ,
v n ) = n 1 2 + a . Then, γ H * ( v n + 1 2 a ) = γ H * ( v n + 1 2 + a ) = n 1 2 + a by Proposition 1, where 1 a n 1 2 . Therefore, we have e x p H * ( k ) = n 1 2 + k 2 , where 1 k n .
(2)
H * * is shown in Figure 3. If n 2 d 1 , then n d + 1 n + 1 2 . Since n 5 and n 2 d 1 , then d 2 n + 1 4 n 1 2 and n d 2 + 1 n n + 1 4 + 1 n + 3 2 . Then, v n + 1 2 is not a loop vertex.
Suppose any integer a satisfies d 1 a n + 1 2 . Since d 2 , then a d 1 d 2 . Let x be any vertex of H * * . Let the walk W ( v a , x ) be v a d ( v a , v n + 1 2 ) v n + 1 2 d ( v n + 1 2 , x ) x . If V ( W ( v a , x ) ) V ( L ( H * * ) ) , then γ H * * ( v a , x ) ( n + 1 2 a ) + n 1 2 = n a . If V ( W ( v a , x ) ) V ( L ( H * * ) ) = , we have V ( P ( x , v n + 1 2 ) ) V ( L ( D ) ) = and V ( P ( v a , v n + 1 2 ) )
V ( L ( D ) ) = . According to Corollary 1, we have γ H * * ( v a , x ) n d + 1 . Therefore, we have γ H * * ( v a ) max { n a , n d + 1 } = n d + 1 , where d 1 a n + 1 2 . Since d 2 and d 1 a n + 1 2 , then n + 1 2 n a + 1 n d + 2 n d 2 + 1 . Further, γ H * * ( v a , v n a + 1 ) = d ( v a , v d 2 ) + d ( v d 2 , v n a + 1 ) = d ( v a , v n d 2 + 1 ) + d ( v n d 2 + 1 , v n a + 1 ) = n d + 1 . Therefore, γ H * * ( v a ) = n d + 1 , where d 1 a n + 1 2 . According to Proposition 1, if n + 1 2 a n d + 2 , we have γ H * * ( v a ) = γ H * * ( v n a + 1 ) = n d + 1 . So, for d 1 a n d + 2 , we have γ H * * ( v a ) = n d + 1 . Therefore, we have e x p H * * ( k ) = n d + 1 , where 1 k n 2 d + 4 .
Suppose a satisfies 1 a d 1 . We have γ H * * ( v a ) d ( v a , v d 1 ) + γ H * * ( v d 1 ) = n d + 1 + ( d 1 a ) = n a . Moreover, γ H * * ( v a ) = γ H * * ( v n a + 1 ) = d ( v a , v n ) = n a . So, we can get e x p H * * ( k ) = n d + 1 + k ( n 2 d + 4 ) 2 , where n 2 d + 4 k n .
Therefore, the theorem now holds. □
Theorem 6.
If n is even, d is even and d 2 , then
(1)
If n 2 d 2 , then e x p H * ( k ) = n 2 1 + k 2 , where 1 k n .
(2)
n 2 d , then
e x p H * * ( k ) = n d + 1 , where 1 k n 2 d + 4 , n d + 1 + k ( n 2 d + 4 ) 2 , where n 2 d + 4 k n .
Proof. 
(1) H * is shown in Figure 2. We have H * D S n , 2 ( d ) , f = g = n 2 or f = g = n 2 + 1 . Let us assume f = g = n 2 , then n + 1 f = n + 1 g = n 2 + 1 . If n 2 d 2 , let us consider the following two cases.
Case 1 { v n 2 , v n 2 + 1 } V ( L ( H * ) ) .
It is easy to see that γ H * ( v n 2 ) = γ H * ( v n 2 + 1 ) = max { d ( v n 2 , x ) : x V ( H * ) } = d ( v n 2 , v n ) = n 2 . Suppose a satisfies 1 a n 2 1 . According to Lemma 2, we can get γ H * ( v n 2 a ) d ( v n 2 a , v n 2 ) + γ H * ( v n 2 ) = n 2 + a . Further, γ H * ( v n 2 a ) = γ H * ( v n 2 a , v n ) = d ( v n 2 a , v n ) = n 2 + a . According to Proposition 1, we have γ H * ( v n 2 a ) = γ H * ( v n 2 + a + 1 ) = n 2 + a , where 1 a n 2 1 . Therefore, we have e x p H * ( k ) = n 2 1 + k 2 , where 1 k n .
Case 2 { v n 2 , v n 2 + 1 } V ( L ( H * ) ) = .
Let x be any vertex of H * . If V ( P ( v n 2 , x ) ) V ( L ( H * ) ) , then γ H * ( v n 2 , x ) = d ( v n 2 , x ) n 2 . If V ( P ( v n 2 , x ) ) V ( L ( H * ) ) = , according to Corollary 2, we have γ H * ( v n 2 , x )
n d + 1 . Then, γ H * ( v n 2 ) max { n 2 , n d + 1 } = n 2 . Further, we have γ H * ( v n 2 ) = γ H * ( v n 2 , v n ) = d ( v n 2 , v n ) = n 2 . Let a satisfy that 1 a n 2 1 . Then, γ H * ( v n 2 a ) d ( v n 2 a , v n 2 ) + γ H * ( v n 2 ) = n 2 + a . Further, we have γ H * ( v n 2 a ) = γ H * ( v n 2 a , v n ) = d ( v n 2 a , v n )
= n 2 + a . According to Proposition 1, γ H * ( v n 2 + 1 + a ) = γ H * ( v n 2 a ) = n 2 + a , where 0 a n 2 1 . Therefore, we have e x p H * ( k ) = n 2 1 + k 2 , where 1 k n .
(2) H * * is shown in Figure 4. We have H * * D S n , 2 ( d ) , f = g = n 2 or f = g = n 2 + 1 . Let us assume f = g = n 2 , then n + 1 f = n + 1 g = n 2 + 1 . If n 2 d , then n d + 1 n 2 + 1 . Since n 6 and n 2 d , then d 2 n 4 n 2 1 and n d 2 + 1 n n 4 + 1 n 2 + 2 . Then v n 2 and v n 2 + 1 are not loop vertices.
Suppose any integer a satisfies d 1 a n 2 . Since d 2 , then a d 1 d 2 . Let x be any vertex of H * * . Let the walk W ( v a , x ) be v a d ( v a , v n 2 ) v n 2 d ( v n 2 , x ) x . If V ( W ( v a , x ) ) V ( L ( H * * ) ) , then γ H * * ( v a , x ) ( n 2 a ) + n 2 = n a . If V ( W ( v a , x ) ) V ( L ( H * * ) ) = , we have V ( P ( v a , v n 2 ) ) V ( L ( D ) ) = and V ( P ( x , v n 2 ) ) V ( L ( D ) ) = . According to Corollary 2, we have γ H * * ( v a , x ) n d + 1 . So γ H * * ( v a ) max { n a , n d + 1 } = n d + 1 , where d 1 a n 2 . Since d 2 and d 1 a n 2 , then n 2 + 1 n a + 1 n d + 2 n d 2 + 1 . Further, we have γ H * * ( v a ) = γ H * * ( v a , v n a + 1 ) = d ( v a , v d 2 ) + d ( v d 2 , v n a + 1 ) = d ( v a , v n d 2 + 1 ) + d ( v n d 2 + 1 , v n a + 1 ) = n d + 1 , where d 1 a n 2 . According to Proposition 1, if n 2 + 1 a n d + 2 , we have γ H * * ( v a ) = γ H * * ( v n a + 1 ) = n d + 1 . So, for d 1 a n d + 2 , we have γ H * * ( v a ) = n d + 1 . Therefore, we have e x p H * * ( k ) = n d + 1 , where 1 k n 2 d + 4 .
Suppose a satisfies 1 a d 1 . We have γ H * * ( v a ) d ( v a , v d 1 ) + γ H * * ( v d 1 ) = n d + 1 + ( d 1 a ) = n a . Moreover, γ H * * ( v a ) = γ H * * ( v n a + 1 ) = d ( v a , v n ) = n a . So, we can get e x p H * * ( k ) = n d + 1 + k ( n 2 d + 4 ) 2 , where n 2 d + 4 k n .
Therefore, the theorem now holds. □
It can be seen from Theorems 4–6, the upper bound for the kth local exponent of doubly symmetric primitive digraphs of order n with d loops can be reached, where 1 k n .

4. Conclusions

In this paper, we study the upper bound for the kth local exponent of doubly symmetric primitive digraphs of order n with d loops, where d is an integer such that 1 d n . For the class of doubly symmetric primitive digraphs, we get the upper bound for the kth local exponent, where 1 k n . Furthermore, for the class of doubly symmetric primitive digraphs, we find that the upper bound for the kth local exponent can be reached, where 1 k n . The upper bounds of the generalized μ -scrambling indices for a doubly symmetric primitive digraph are not given. It would be meaningful and interesting to solve the problems in future research.

Funding

This paper is supported by Shanghai Institute of Technology (YJ2021-55).

Institutional Review Board Statement

Not applicable.

Informed Consent Statement

Not applicable.

Data Availability Statement

Not applicable.

Acknowledgments

We would like to thank the editor and reviewers for their valuable suggestions and comments which greatly improved the article.

Conflicts of Interest

The author declares no conflict of interest.

References

  1. Brualdi, R.A.; Ryser, H.J. Combinatorial Matrix Theory; Cambridge University Press: Cambridge, UK, 1991. [Google Scholar]
  2. Brualdi, R.A.; Liu, B. Generalized exponents of primitive directed graphs. J. Graph Theory 1990, 14, 483–499. [Google Scholar] [CrossRef]
  3. Brualdi, R.A.; Shao, J.Y. Generalized exponents of primitive symmetric digraphs. Discret. Appl. Math. 1997, 74, 275–293. [Google Scholar] [CrossRef]
  4. Cruz, R.; Gutman, I.; Rada, J. Sombor index of chemical graphs. Appl. Math. Comput. 2021, 399, 126018. [Google Scholar] [CrossRef]
  5. Rather, B.A.; Pirzada, S.; Naikoo, T.A.; Shang, Y. On Laplacian Eigenvalues of the Zero-Divisor Graph Associated to the Ring of Integers Modulo n. Mathematics 2021, 9, 482. [Google Scholar] [CrossRef]
  6. Nath, R.K.; Fasfous, W.N.T.; Das, K.C.; Shang, Y. Common Neighborhood Energy of Commuting Graphs of Finite Groups. Symmetry 2021, 13, 1651. [Google Scholar] [CrossRef]
  7. Shao, J. The exponent set of symmetric primitive matrices. Sci. Sin. Ser. A 1987, 30, 348–358. [Google Scholar]
  8. Zhou, B.; Shen, J. On generalized exponents of tournaments. Taiwan. J. Math. 2002, 6, 565–572. [Google Scholar] [CrossRef]
  9. Gao, Y.; Shao, Y. Generalized exponents of primitive two-colored digraphs. Linear Algebra Appl. 2009, 430, 1550–1565. [Google Scholar] [CrossRef]
  10. Miao, Z.; Zhang, K. The local exponent sets of primitive digraphs. Linear Algebra Appl. 2000, 307, 15–33. [Google Scholar] [CrossRef]
  11. Chen, S.; Liu, B. The kth local exponent of doubly symmetric primitive matrices. Appl. Math. Lett. 2006, 19, 392–397. [Google Scholar] [CrossRef]
  12. Chen, S.; Liu, B. Matrices with maximum kth local exponent in the class of doubly symmetric primitive matrices. Discret. Math. 2008, 308, 3386–3392. [Google Scholar] [CrossRef]
  13. Akelbek, M.; Kirkland, S. Coefficients of ergodicity and the scrambling index. Linear Algebra Appl. 2009, 430, 1111–1130. [Google Scholar] [CrossRef]
  14. Akelbek, M.; Kirkland, S. Primitive digraphs with the largest scrambling index. Linear Algebra Appl. 2009, 430, 1099–1110. [Google Scholar] [CrossRef]
  15. Chen, S.; Liu, B. The scrambling index of symmetric primitive matrices. Linear Algebra Appl. 2010, 433, 1110–1126. [Google Scholar] [CrossRef]
  16. Kim, H.K. Scrambling index set of primitive digraphs. Linear Algebra Appl. 2013, 439, 1886–1893. [Google Scholar] [CrossRef]
  17. Kim, H.K. Generalized competition index of a primitive digraph. Linear Algebra Appl. 2010, 433, 72–79. [Google Scholar] [CrossRef]
  18. Sim, M.S.; Kim, H.K. On generalized competition index of a primitive tournament. Discret. Math. 2011, 311, 2657–2662. [Google Scholar] [CrossRef]
  19. Kim, H.K.; Park, S.G. A bound of generalized competition index of a primitive digraph. Linear Algebra Appl. 2012, 436, 86–98. [Google Scholar] [CrossRef]
  20. Kim, H.K.; Lee, S.H. Generalized competition indices of symmetric primitive digraphs. Discret. Appl. Math. 2012, 160, 1583–1590. [Google Scholar] [CrossRef]
  21. Shao, Y.; Gao, Y. The m-competition indices of symmetric primitive digraphs with loop. Ars Combin. 2013, 108, 217–223. [Google Scholar]
  22. Fang, W.; Gao, Y.; Shao, Y.; Gao, W.; Jing, G.; Li, Z. The generalized competition indices of primitive minimally strong digraphs. Linear Algebra Appl. 2016, 493, 206–226. [Google Scholar] [CrossRef]
  23. Chen, D.; Li, X. The Generalized Competition Indices of Doubly Symmetric Primitive Digraphs with d Loops. Symmetry 2022, 14, 1192. [Google Scholar] [CrossRef]
  24. Huang, Y.; Liu, B. Generalized scrambling indices of a primitive digraph. Linear Algebra Appl. 2010, 433, 1798–1808. [Google Scholar] [CrossRef]
  25. Zhang, L.; Huang, T.Z. Bounds on the generalized μ-scrambling indices of primitive digraphs. Int. J. Comput. Math. 2012, 89, 17–29. [Google Scholar] [CrossRef]
  26. Zhang, L.; Mou, G.F.; Liu, F.; Li, Z.S. Some bounds of the generalized μ-scrambling indices of primitive digraphs with d loops. J. Inequal. Appl. 2021, 2021, 128. [Google Scholar] [CrossRef]
Figure 1. G.
Figure 1. G.
Symmetry 14 01623 g001
Figure 2. The H * for d is even and d 2 .
Figure 2. The H * for d is even and d 2 .
Symmetry 14 01623 g002
Figure 3. The H * * for n is odd, d is even such that d 2 , and n 2 d 1 .
Figure 3. The H * * for n is odd, d is even such that d 2 , and n 2 d 1 .
Symmetry 14 01623 g003
Figure 4. The H * * for n is even, d is even such that d 2 , and n 2 d .
Figure 4. The H * * for n is even, d is even such that d 2 , and n 2 d .
Symmetry 14 01623 g004
Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Share and Cite

MDPI and ACS Style

Chen, D. The kth Local Exponent of Doubly Symmetric Primitive Digraphs with d Loops. Symmetry 2022, 14, 1623. https://0-doi-org.brum.beds.ac.uk/10.3390/sym14081623

AMA Style

Chen D. The kth Local Exponent of Doubly Symmetric Primitive Digraphs with d Loops. Symmetry. 2022; 14(8):1623. https://0-doi-org.brum.beds.ac.uk/10.3390/sym14081623

Chicago/Turabian Style

Chen, Danmei. 2022. "The kth Local Exponent of Doubly Symmetric Primitive Digraphs with d Loops" Symmetry 14, no. 8: 1623. https://0-doi-org.brum.beds.ac.uk/10.3390/sym14081623

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Metrics

Back to TopTop