Next Article in Journal
Grey Wolf Optimizer-Based Optimal Controller Tuning Method for Unstable Cascade Processes with Time Delay
Previous Article in Journal
Certain Solutions of Abel’s Integral Equations on Distribution Spaces via Distributional Gα-Transform
Previous Article in Special Issue
Inequalities for Laplacian Eigenvalues of Signed Graphs with Given Frustration Number
 
 
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

On the Laplacian and Signless Laplacian Characteristic Polynomials of a Digraph

1
Department of School Education, Jammu and Kashmir Government, Kashmir 193404, India
2
Department of Computer and Information Sciences, Northumbria University, Newcastle NE1 8ST, UK
*
Author to whom correspondence should be addressed.
Symmetry 2023, 15(1), 52; https://doi.org/10.3390/sym15010052
Submission received: 3 October 2022 / Revised: 27 October 2022 / Accepted: 20 December 2022 / Published: 25 December 2022
(This article belongs to the Special Issue Symmetry, Graph Reconstruction and Molecular Conduction)

Abstract

:
Let D be a digraph with n vertices and a arcs. The Laplacian and the signless Laplacian matrices of D are, respectively, defined as L ( D ) = D e g + ( D ) A ( D ) and Q ( D ) = D e g + ( D ) + A ( D ) , where A ( D ) represents the adjacency matrix and D e g + ( D ) represents the diagonal matrix whose diagonal elements are the out-degrees of the vertices in D. We derive a combinatorial representation regarding the first few coefficients of the (signless) Laplacian characteristic polynomial of D. We provide concrete directed motifs to highlight some applications and implications of our results. The paper is concluded with digraph examples demonstrating detailed calculations.

1. Introduction

We consider a digraph D = ( V , Γ ) of order n, where the vertex set is given by V = { v 1 , v 2 , , v n } and | V | = n . The set of arcs is denoted by Γ , which contains ordered pairs of distinct vertices. Throughout this paper, digraphs mean directed graphs without loops or multiple arcs. Two vertices, u and v, of D are adjacent when they are linked via an arc ( u , v ) Γ , or ( v , u ) Γ , and they are called doubly adjacent when { ( u , v ) , ( v , u ) } Γ . For any vertex v i , the set
N i = { v j V : { ( v i , v j ) , ( v j , v i ) } Γ }
contains all vertices that are doubly adjacent of v i .
A sequence of vertices W : u = u 0 , u 1 , , u l = v is called a walk W, which has length l from vertex u to vertex v. Here, ( u k 1 , u k ) is an arc in D for 1 k l . W is a closed walk if u = v . If each pair of different vertices u , v in D admits a walk from u to v and a walk from v to u, then D is strongly connected. Let c 2 ( i ) be the number of closed walks having length 2 originated from v i . In the digraph D, the sequence ( c 2 ( 1 ) , c 2 ( 2 ) , , c 2 ( n ) ) is a closed walk sequence of length 2.
A digraph D is weakly connected or connected when the undirected version G of D is connected.
A digraph D is called symmetric if ( u , v ) and ( v , u ) Γ , for all u , v V . G G forms a one-to-one mapping between simple graphs and symmetric digraphs, where G and G share the same vertices and each edge u v of G is mapped to symmetric arcs ( u , v ) and ( v , u ) . Clearly, a graph corresponds to a symmetric digraph under this mapping.
We define the adjacency matrix A ( D ) as a n × n binary matrix with rows and columns indexed by the vertices. The element a i j takes a i j = 1 , if ( v i , v j ) Γ and a i j = 0 otherwise. We refer the readers to [1,2] for some recent works on the spectral properties of A ( D ) . Denote by D e g + ( D ) = d i a g ( d 1 + , d 2 + , , d n + ) the diagonal matrix of out-degrees. The matrices L ( D ) = D e g + ( D ) A ( D ) and Q ( D ) = D e g + ( D ) + A ( D ) are known as the Laplacian and the signless Laplacian matrix of the digraph D. We call the eigenvalues of the matrices L ( D ) and Q ( D ) the Laplacian and the signless Laplacian eigenvalues of D, respectively. Let P L ( D , x ) and P Q ( D , x ) be the characteristic polynomials of the matrices L ( D ) and Q ( D ) , respectively. They are the Laplacian and the signless Laplacian characteristic polynomial of D. Some recent results on the spectral properties of L ( D ) , Q ( D ) and related results can be found in, e.g., [3,4,5,6,7,8,9,10,11,12].
Two graphs that are non-isomorphic are co-spectral if they have the same spectrum relevant to a given graph matrix. Apparently, two isomorphic graphs have the same spectrum relevant to a given graph matrix. If two graphs share the same spectrum relevant to a given matrix, is it true that they are isomorphic? This is one of the most investigated and difficult problems in spectral theory of graphs and digraphs. It yields the following problem.
Problem 1.
With respect to a given graph (digraph) matrix, which graphs (digraphs) are determined by their spectra?
A graph G is determined by its spectrum relevant to a given graph matrix if there is no other graph sharing the same spectrum as G relevant to the graph matrix. When two graphs admit the same spectrum relevant to a given graph matrix, they share the same characteristic polynomial as well as the coefficients of the characteristic polynomial. This observation prompted many researchers to examine carefully the coefficients of characteristic polynomial relevant to a given graph matrix. Apparently, if two graphs differ in at least one coefficient of their characteristic polynomial relevant to a given graph matrix, they cannot be co-spectral regarding the graph matrix. Many results have been reported in the literature on this problem. Sachs [13] and Mowshowitz [14] established the coefficients of the adjacency characteristic polynomial of an arbitrary digraph. It is revealed that the coefficients of the adjacency characteristic polynomial of a tree count matchings. The formulas for the first four coefficients of the Laplacian characteristic polynomial of a graph were derived in [15]. Cvetković et al. [16] presented the formulas for the first three coefficients of the signless Laplacian characteristic polynomial of a graph. Guo et. al [17] provided combinatorial expression for the first few coefficient of the (normalized) Laplacian and the signless Laplacian characteristic polynomials of a graph. Based upon the fifth coefficient of the adjacency characteristic polynomial of trees, Lepović et al. [18] showed that no starlike trees are co-spectral. The coefficients of the Laplacian characteristic polynomial were employed [19] to reveal the Laplacian spectra of three-rose graphs. Some recent results can be found in, e.g., [20,21,22].
The issue of spectral determination in digraphs related to adjacency matrices has also been considered. Some families of digraphs are characterized through adjacency spectra [23]. Recently, some researchers [11] studied the spectral determination problem for oriented graphs relevant to a generalized skew matrix.
With motivation from the above works, we aim to investigate the coefficients of the characteristic polynomial of the (signless) Laplacian matrices of a digraph. We obtain algebraic expressions for the first few coefficients of the (signless) Laplacian characteristic polynomials of a digraph.
Given a graph G = ( V , E ) with the vertex set V ( G ) = { x 1 , x 2 , , x n } , let G i be the graph obtained from G by scrapping the ith vertex x i and its incident edges. The graph G is reconstructible if it can be determined (up to isomorphism) by all vertex-deleted graphs G i . Ulam’s reconstruction conjecture states that every graph with n > 3 vertices is reconstructible. Many variations of this problem have been considered, and one of them is the spectral reconstruction problem. It claims that “a graph G is spectrally reconstructible if it can be determined up to isomorphism by the adjacency characteristic polynomial of G and the adjacency characteristic polynomials of its vertex deleted subgraphs G i ”. Our study of coefficients of the characteristic polynomial of the (signless) Laplacian matrices of a digraph is also motivated by this line of research.

2. Laplacian Coefficients of D

In this section, we establish an algebraic representation for the first five Laplacian coefficients of a digraph.
To determine the coefficients of a Laplacian characteristic polynomial of a digraph D, we need the following Lemma, which can be found in [24].
Lemma 1.
Let B = ( b i j ) be a n × n matrix having the characteristic polynomial
Φ ( B , y ) = d e t ( y I B ) = y n + i = 1 n c i ( G ) y n i .
.
Let s k = t r ( B k ) be the trace of B k . Then, for the coefficients of Φ ( B , y ) , we have the following:
c 1 = s 1 a n d k c k = s k c 1 s k 1 c 2 s k 2 . . . a k 1 c 1 , k = 2 , 3 , , n .
The following interesting result about the trace of the product of matrices can be found in [24].
Lemma 2.
For n × n matrices A and B, we have t r ( A B ) = t r ( B A ) .
The following result is the main result of this section and shows how the first five coefficients of the Laplacian characteristic polynomial of a digraph D can be expressed in terms of the structure of the digraph.
Theorem 1.
Let D be a connected digraph of order n 3 with a arcs having vertex out-degrees d 1 + d 2 + d n + . Let P L ( D , x ) = x n + i = 1 n a i x n i be the Laplacian characteristic polynomial of D. Then,
(1) 
a 1 = a ,
(2) 
a 2 = a 2 2 1 2 i = 1 n ( d i + ) 2 + c 2 ( i ) ,
(3) 
a 3 = 1 3 i = 1 n n ( C 3 i ) i = 1 n c 2 ( i ) d i + 1 3 i = 1 n ( d i + ) 3 + a 2 i = 1 n ( d i + ) 2 + c 2 ( i ) a 3 6 ,
(4) 
and
a 4 = i = 1 n d i + n ( C 3 i ) 1 4 i = 1 n ( d i + ) 4 i = 1 n ( d i + ) 2 c 2 ( i ) 1 2 i = 1 n d i + M i ( 4 a + 3 ) 12 i = 1 n n ( C 3 i ) 1 4 i = 1 n n ( P 3 1 ( v i v j v k v j v i ) ) + n ( P 3 2 ( v i v j v i v k v i ) 1 4 i = 1 n c 2 ( i ) + a 3 i = 1 n ( d i + ) 3 + a i = 1 n d i + c 2 ( i ) a 2 4 i = 1 n ( d i + ) 2 + c 2 ( i ) + 1 8 i = 1 n ( ( d i + ) 2 + c 2 ( i ) ) 2 + a 4 24 .
Here, c 2 ( i ) is the number of closed walks of length 2 at the vertex v i , n ( C 3 i ) is the number of directed cycles of length 3 at the vertex v i , n ( C 4 i ) is the number of directed cycles of length 4 at the vertex v i , n ( P 3 1 ( v i v j v k v j v i ) ) is the number of paths on 3 vertices v i , v j , v k at the vertex v i with arcs ( v i , v j ) , ( v j , v k ) , ( v k , v j ) , ( v j , v i ) , n ( P 3 2 ( v i v j v i v k v i ) ) is the number of paths on 3 vertices v i , v j , v k at the vertex v i with arcs ( v i , v j ) , ( v j , v i ) , ( v i , v k ) , ( v k , v i ) and M i = ( v i , v j ) , ( v j , v i ) E ( D ) d i + is the sum of the out-degrees of the vertices which are both out-neighbor and the in-neighbor of the vertex v i .
Proof. 
Let s k = t r ( L ( D ) k ) be the trace of the matrix L ( D ) k , where L ( D ) = D e g + ( D ) A ( D ) is the Laplacian matrix of D and k N . We have, s 1 = t r ( L ( D ) ) = i = 1 n d i + = a , the number of arcs. This shows by using Lemma 1 that a 1 = s 1 = a . By a simple calculation, it can be seen that
L ( D ) 2 = D e g + ( D ) A ( D ) 2 = ( D e g + ( D ) ) 2 D e g + ( D ) A ( D ) A ( D ) D e g + ( D ) + A ( D ) 2 .
Clearly, t r ( D e g + ( D ) A ( D ) ) = t r ( A ( D ) D e g + ( D ) ) by Lemma 2. It is easy to establish that each of the diagonal entries of D e g + ( D ) A ( D ) are zero, therefore we obtain
t r ( D e g + ( D ) A ( D ) ) = t r ( A ( D ) D e g + ( D ) ) = 0 .
Additionally, t r ( ( D e g + ( D ) ) 2 ) = t = 1 n ( d t + ) 2 and t r ( A ( D ) 2 ) = i = 1 n j = 1 n a i j a j i = t = 1 n c 2 ( t ) , where c 2 ( t ) is the number of symmetric pair of arcs or the number of closed walks of length 2 at the vertex v t . Therefore, from (2), we obtain
s 2 = t r ( L ( D ) 2 ) = t r ( ( D e g + ( D ) ) 2 ) 2 t r ( D e g + ( D ) A ( D ) ) + t r ( A ( D ) 2 ) = t = 1 n ( d t + ) 2 + c 2 ( t ) .
Using this together with Lemma 1, we arrive at
a 2 = 1 2 ( s 2 a 1 s 1 ) = a 2 2 1 2 t = 1 n ( d t + ) 2 + c 2 ( t ) .
By binomial theorem with index 3, it can be seen that
L ( D ) 3 = ( D e g + ( D ) A ( D ) ) 3 = D e g + ( D ) 3 D e g + ( D ) 2 A ( D ) D e g + ( D ) A ( D ) D e g + ( D ) A ( D ) D e g + ( D ) 2 + D e g + ( D ) A ( D ) 2 + A ( D ) D e g + ( D ) A ( D ) + A ( D ) 2 D e g + ( D ) A ( D ) 3 .
Clearly, by Lemma 2, we have
t r ( ( D e g + ( D ) ) 2 A ( D ) = t r ( D e g + ( D ) A ( D ) D e g + ( D ) ) = t r ( A ( D ) D e g + ( D ) 2 )
and each of the diagonal entries of D e g + ( D ) 2 A ( D ) are zero, it follows that
t r ( ( D e g + ( D ) ) 2 A ( D ) ) = t r ( D e g + ( D ) A ( D ) D e g + ( D ) ) = t r ( A ( D ) ( D e g + ( D ) ) 2 ) = 0 .
Additionally, t r ( D e g + ( D ) A ( D ) 2 ) = t r ( A ( D ) D e g + ( D ) A ( D ) ) = t r ( A ( D ) 2 D e g + ( D ) ) , by Lemma 2 and the t-th diagonal entry of D e g + ( D ) A ( D ) 2 is c 2 ( t ) d t + , it follows that
t r ( D e g + ( D ) A ( D ) 2 ) = t r ( A ( D ) D e g + ( D ) A ( D ) ) = t r ( A ( D ) 2 D e g + ( D ) ) = t = 1 n c 2 ( t ) d t + .
Further, t r ( ( D e g + ( D ) ) 3 ) = t = 1 n ( d t + ) 3 and
t r ( A ( D ) 3 ) = i = 1 n j = 1 n k = 1 n a i j a j k a k i = ( number of directed cycles of length 3 in D ) = t = 1 n n ( C 3 t ) .
Therefore, from (3), we obtain
s 3 = t r ( L ( D ) 3 ) = t r ( ( D e g + ( D ) ) 3 ) 3 t r ( ( D e g + ( D ) ) 2 A ( D ) + 3 t r ( D e g + ( D ) A ( D ) 2 ) t r ( A ( D ) 3 ) = t = 1 n ( d t + ) 3 + 3 t = 1 n c 2 ( t ) d t + t = 1 n n ( C 3 t ) .
Using these observations together with Lemma 1, we arrive at
a 3 = 1 3 ( s 3 a 1 s 2 a 2 s 1 ) = 1 3 i = 1 n n ( C 3 i ) i = 1 n c 2 ( i ) d i + 1 3 i = 1 n ( d i + ) 3 + a 2 i = 1 n ( d i + ) 2 + c 2 ( i ) a 3 6 .
Lastly, by binomial theorem with index 4, we have
L ( D ) 4 = D e g + ( D ) A ( D ) 4 = ( D e g + ( D ) ) 4 ( D e g + ( D ) ) 3 A ( D ) ( D e g + ( D ) ) 2 A ( D ) D e g + ( D ) D e g + ( D ) A ( D ) ( D e g + ( D ) ) 2 + ( D e g + ( D ) ) 2 A ( D ) 2 + D e g + ( D ) A ( D ) 2 D e g + ( D ) + D e g + ( D ) A ( D ) D e g + ( D ) A ( D ) D e g + ( D ) A ( D ) 3 A ( D ) ( D e g + ( D ) ) 3 + A ( D ) ( D e g + ( D ) ) 2 A ( D ) + A ( D ) D e g + ( D ) A ( D ) D e g + ( D ) A ( D ) D e g + ( D ) A ( D ) 2 + A ( D ) 2 ( D e g + ( D ) ) 2 A ( D ) 2 D e g + ( D ) A ( D ) 2 A ( D ) 3 D e g + ( D ) + A ( D ) 4 .
Using Lemma 2, we have
t r ( ( D e g + ( D ) ) 3 A ( D ) ) = t r ( ( D e g + ( D ) ) 2 A ( D ) D e g + ( D ) ) = t r ( D e g + ( D ) A ( D ) ( D e g + ( D ) ) 2 ) = t r ( A ( D ) ( D e g + ( D ) ) 3 )
and the i-th diagonal entry of ( D e g + ( D ) ) 3 A ( D ) is zero, we get
t r ( ( D e g + ( D ) ) 3 A ( D ) ) = t r ( ( D e g + ( D ) ) 2 A ( D ) D e g + ( D ) ) = t r ( D e g + ( D ) A ( D ) ( D e g + ( D ) ) 2 ) = t r ( A ( D ) ( D e g + ( D ) ) 3 ) = 0 .
Additionally, by Lemma 2
t r ( ( D e g + ( D ) ) 2 A ( D ) 2 ) = t r ( A ( D ) ( D e g + ( D ) ) 2 A ( D ) ) = t r ( D e g + ( D ) A ( D ) 2 D e g + ( D ) ) = t r ( A ( D ) 2 ( D e g + ( D ) ) 2 )
and the i-th diagonal entry of ( D e g + ( D ) ) 2 A ( D ) 2 is ( d i + ) 2 c 2 ( i ) , we obtain
t r ( ( D e g + ( D ) ) 2 A ( D ) 2 ) = t r ( A ( D ) ( D e g + ( D ) ) 2 A ( D ) ) = t r ( D e g + ( D ) A ( D ) 2 D e g + ( D ) ) = t r ( A ( D ) 2 ( D e g + ( D ) ) 2 ) = i n ( d i + ) 2 c 2 ( i ) .
Again by Lemma 2, we have
t r ( D e g + ( D ) A ( D ) D e g + ( D ) A ( D ) ) = t r ( A ( D ) D e g + ( D ) A ( D ) D e g + ( D ) )
and the i-th diagonal entry of D e g + ( D ) A ( D ) D e g + ( D ) A ( D ) is
d i + k = 1 n d k + a i k a k i = d i + ( v i , v k ) , ( v k , v i ) E ( D ) d k + = d i + M i ,
where M i is the sum of the out-degrees of the vertices which are both out and in-neighbors of the vertex v i . It follows that
t r ( D e g + ( D ) A ( D ) D e g + ( D ) A ( D ) ) = t r ( A ( D ) D e g + ( D ) A ( D ) D e g + ( D ) ) = i = 1 n d i + M i .
Using Lemma 2, we have
t r ( D e g + ( D ) A ( D ) 3 ) = t r ( A ( D ) D e g + ( D ) A ( D ) 2 ) = t r ( A ( D ) 2 D e g + ( D ) A ( D ) ) = t r ( A ( D ) 3 D e g + ( D ) )
and the i-th diagonal entry of D e g + ( D ) A ( D ) 3 is
d i + j = 1 n k = 1 n a i j a j k a k i = d i + n ( C 3 i ) ,
where n ( C 3 i ) is the number of directed 3-cycles at the vertex v i . It follows that
t r ( D e g + ( D ) A ( D ) 3 ) = t r ( A ( D ) D e g + ( D ) A ( D ) 2 ) = t r ( A ( D ) 2 D e g + ( D ) A ( D ) ) = t r ( A ( D ) 3 D e g + ( D ) ) = i = 1 n d i + n ( C 3 i ) .
Further, t r ( ( D e g + ( D ) ) 4 ) = i = 1 n ( d i + ) 4 and
t r ( A ( D ) 4 ) = i = 1 n j = 1 n k = 1 n l = 1 n a i j a j k a k l a l i = ( number of directed four cycles in D ) + i = 1 n n ( P 3 1 ( v i v j v k v j v i ) ) + i = 1 n n ( P 3 2 ( v i v j v i v k v i ) ) + i = 1 n c 2 ( i ) = i = 1 n n ( C 4 i ) + i = 1 n n ( P 3 1 ( v i v j v k v j v i ) ) + n ( P 3 2 ( v i v j v i v k v i ) ) + i = 1 n c 2 ( i ) ,
Now, from (4), we have
s 4 = t r ( L ( D ) 4 ) = t r ( ( D e g + ( D ) ) 4 ) 4 t r ( ( D e g + ( D ) ) 3 A ( D ) ) + 4 t r ( ( D e g + ( D ) ) 2 A ( D ) 2 ) + 2 t r ( D e g + ( D ) A ( D ) D e g + ( D ) A ( D ) ) 4 t r ( D e g + ( D ) A ( D ) 3 ) + t r ( A ( D ) 4 ) = i = 1 n ( d i + ) 4 4 i = 1 n ( d i + ) 2 c 2 ( i ) + 2 i = 1 n d i + M i 4 i = 1 n d i + n ( C 3 i ) + i = 1 n n ( C 3 i ) + i = 1 n n ( P 3 1 ( v i v j v k v j v i ) ) + n ( P 3 2 ( v i v j v i v k v i ) + i = 1 n c 2 ( i ) .
Using these observations together with Lemma 1, we arrive at
a 4 = 1 4 ( s 4 a 1 s 3 a 2 s 2 a 3 s 1 ) = i = 1 n d i + n ( C 3 i ) 1 4 i = 1 n ( d i + ) 4 i = 1 n ( d i + ) 2 c 2 ( i ) 1 2 i = 1 n d i + M i ( 4 a + 3 ) 12 i = 1 n n ( C 3 i ) 1 4 i = 1 n n ( P 3 1 ( v i v j v k v j v i ) ) + n ( P 3 2 ( v i v j v i v k v i ) 1 4 i = 1 n c 2 ( i ) + a 3 i = 1 n ( d i + ) 3 + a i = 1 n d i + c 2 ( i ) a 2 4 i = 1 n ( d i + ) 2 + c 2 ( i ) + 1 8 i = 1 n ( ( d i + ) 2 + c 2 ( i ) ) 2 + a 4 24 .
This completes the proof of the theorem. □
The k-th spectral moment of a digraph D is defined as the sum of the k-th powers of the adjacency eigenvalues of D. Likewise, the Laplacian spectral moments of D are defined as
t r ( L ( D ) k ) = i = 1 n μ i ( D ) k , k = 1 , 2 , ,
where μ 1 ( D ) , μ 2 ( D ) , , μ n 1 ( D ) , μ n ( D ) are the Laplacian eigenvalues of D.
The next result presents the formula for the first four Laplacian spectral moments of D.
Corollary 1.
Let D be a connected digraph of order n 3 with a arcs having vertex out-degrees d 1 + d 2 + d n + . Then,
(1) 
i = 1 n μ i ( D ) = a ,
(2) 
i = 1 n μ i ( D ) 2 = i = 1 n ( d i + ) 2 + c 2 ( i ) ,
(3) 
i = 1 n μ i ( D ) 3 = i = 1 n ( d i + ) 3 + 3 i = 1 n c 2 ( i ) d i + i = 1 n n ( C 3 i ) , and
(4) 
i = 1 n μ i ( D ) 4 = i = 1 n ( d i + ) 4 4 i = 1 n ( d i + ) 2 c 2 ( i ) + 2 i = 1 n d i + M i 4 i = 1 n d i + n ( C 3 i ) + i = 1 n n ( C 3 i ) + i = 1 n n ( P 3 1 ( v i v j v k v j v i ) ) + n ( P 3 2 ( v i v j v i v k v i ) ) + i = 1 n c 2 ( i ) .
where c 2 ( i ) , n ( C 3 i ) , M i , n ( P 3 1 ( v i v j v k v j v i ) ) and n ( P 3 2 ( v i v j v i v k v i ) ) are defined in Theorem 1.

3. The Signless Laplacian Coefficients of D

In this section, we present algebraic representation for some of the signless Laplacian coefficients of D.
The next Theorem is the main result of this section and gives the first five coefficients of the signless Laplacian characteristic polynomial of a digraph D in terms of structure of the digraph.
Theorem 2.
Let D be a connected digraph of order n 3 with a arcs having vertex out-degrees d 1 + d 2 + d n + . Let P Q ( D , y ) = y n + i = 1 n b i y n i be the signless Laplacian characteristic polynomial of D. Then,
(1) 
b 1 = a ,
(2) 
b 2 = a 2 2 1 2 i = 1 n ( d i + ) 2 + c 2 ( i ) ,
(3) 
b 3 = 1 3 i = 1 n n ( C 3 i ) i = 1 n c 2 ( i ) d i + 1 3 i = 1 n ( d i + ) 3 + a 2 i = 1 n ( d i + ) 2 + c 2 ( i ) a 3 6 ,
(4) 
and
b 4 = i = 1 n d i + n ( C 3 i ) 1 4 i = 1 n ( d i + ) 4 i = 1 n ( d i + ) 2 c 2 ( i ) 1 2 i = 1 n d i + M i ( 4 a 3 ) 12 i = 1 n n ( C 3 i ) 1 4 i = 1 n n ( P 3 1 ( v i v j v k v j v i ) ) + n ( P 3 2 ( v i v j v i v k v i ) 1 4 i = 1 n c 2 ( i ) + a 3 i = 1 n ( d i + ) 3 + a i = 1 n d i + c 2 ( i ) a 2 4 i = 1 n ( d i + ) 2 + c 2 ( i ) + 1 8 i = 1 n ( ( d i + ) 2 + c 2 ( i ) ) 2 + a 4 24 .
Here, c 2 ( i ) is the number of closed walks of length 2 at the vertex v i , n ( C 3 i ) is the number of directed cycles of length 3 at the vertex v i , n ( C 4 i ) is the number of directed cycles of length 4 at the vertex v i , n ( P 3 1 ( v i v j v k v j v i ) ) is the number of paths on vertices v i , v j , v k at the vertex v i with arcs ( v i , v j ) , ( v j , v k ) , ( v k , v j ) , ( v j , v i ) , P 3 2 ( v i ) = v i v j v i v k v i is the number of paths on vertices v i , v j , v k at the vertex v i with arcs ( v i , v j ) , ( v j , v i ) , ( v i , v k ) , ( v k , v i ) and M i = ( v i , v j ) , ( v j , v i ) E ( D ) d i + is the sum of the out-degrees of the vertices which are both out-neighbor and the in-neighbor of the vertex v i .
Proof. 
Let s k = t r ( Q ( D ) k ) be the trace of the matrix Q ( D ) k , where Q ( D ) = D e g + ( D ) + A ( D ) is the signless Laplacian matrix of D and k N . The rest of the proof follows by proceeding similar to Theorem 1 and so is omitted. □
The signless Laplacian spectral moments of D are defined as
t r ( Q ( D ) k ) = i = 1 n q i ( D ) k , k = 1 , 2 , ,
where q 1 ( D ) , q 2 ( D ) , , q n 1 ( D ) , q n ( D ) are the signless Laplacian eigenvalues of D.
The following Corollary 2 follows from Theorem 2 and presents formula for the first four signless Laplacian spectral moments of D.
Corollary 2.
Let D be a connected digraph of order n 3 with a arcs having vertex out-degrees d 1 + d 2 + d n + . Then,
(1) 
i = 1 n q i ( D ) = a ,
(2) 
i = 1 n q i ( D ) 2 = i = 1 n ( d i + ) 2 + c 2 ( i ) ,
(3) 
i = 1 n q i ( D ) 3 = i = 1 n ( d i + ) 3 + 3 i = 1 n c 2 ( i ) d i + + i = 1 n n ( C 3 i ) , and
(4) 
i = 1 n q i ( D ) 4 = i = 1 n ( d i + ) 4 + 4 i = 1 n ( d i + ) 2 c 2 ( i ) + 2 i = 1 n d i + M i + 4 i = 1 n d i + n ( C 3 i ) + i = 1 n n ( C 3 i ) + i = 1 n n ( P 3 1 ( v i v j v k v j v i ) ) + n ( P 3 2 ( v i v j v i v k v i ) + i = 1 n c 2 ( i ) .
Here, c 2 ( i ) , n ( C 3 i ) , M i , n ( P 3 1 ( v i v j v k v j v i ) ) and n ( P 3 2 ( v i v j v i v k v i ) ) are defined in Theorem 1.
Symmetry 15 00052 i001

4. Examples

In this section, we consider some examples of the digraphs to highlight the applications of Theorems 1 and 2.
Example 1.
Let P n = v 1 v 2 v 3 v n 1 v n be the directed path on n 4 vertices with a = n 1 arcs of the form ( v i , v i + 1 ) , i = 1 , 2 , , n 1 . For the directed path P n , we have d i + = 1 , for i n 1 , d n + = 0 , c 2 ( i ) = 0 , for 1 i n , as there is no closed walk of length 2 at any vertex, M i = 0 , for 1 i n . Additionally, n ( C 3 i ) = 0 = n ( C 4 i ) as there is no triangle and 4-cycle at any vertex v i in P n . Further, it is clear that there are no paths of the form P 3 1 ( v i v j v k v j v i ) and P 3 2 ( v i v j v i v k v i ) in P n . With this information, it follows that
a 1 ( P n ) = ( n 1 ) , a 2 ( P n ) = ( n 1 ) 2 2 ( n 1 ) 2 , a 3 ( P n ) = ( n 1 ) 3 6 + ( n 1 ) 2 2 ( n 1 ) 3 , a 4 ( P n ) = ( n 1 ) 4 24 ( n 1 ) 3 4 + 11 ( n 1 ) 2 24 ( n 1 ) 4 .
Let us add arc ( v 2 , v 1 ) in P n and let H 1 be the resulting digraph. For H 1 , it is clear that a = n , d 1 + = 1 , d 2 + = 2 , d n + = 0 , d i + = 1 for i 3 , c 2 ( 1 ) = 1 = c 2 ( 2 ) , c 2 ( i ) = 0 , for i 3 , M 1 = 2 , M 2 = 1 and M i = 0 , for 2 i n . Additionally, n ( C 3 i ) = 0 = n ( C 4 i ) as there is no triangle and 4-cycle at any vertex v i in H 1 . Further, it is clear that there are no paths of the form P 3 1 ( v i v j v k v j v i ) and P 3 2 ( v i v j v i v k v i ) in H 1 . With this information, it follows that
a 1 ( H 1 ) = n , a 2 ( H 1 ) = n 2 2 ( n + 4 ) 2 , a 3 ( H 1 ) = n 3 6 + n ( n + 4 ) 2 n 3 5 , a 4 ( H 1 ) = n 4 24 n 2 ( n + 4 ) 4 + ( n + 4 ) 2 8 + n ( n + 6 ) 3 + 11 n 4 11 .
Let us add another arc ( v 3 , v 2 ) in H 1 and let H 2 be the resulting digraph. For H 2 , it is clear that a = n + 1 , d 1 + = 1 , d 2 + = d 3 + = 2 , d n + = 0 , d i + = 1 for i 4 , c 2 ( 1 ) = 1 = c 2 ( 3 ) , c 2 ( 2 ) = 2 c 2 ( i ) = 0 , for i 4 , M 1 = M 3 = 2 , M 2 = 3 and M i = 0 , for 4 i n . Also, n ( C 3 i ) = 0 = n ( C 4 i ) as there is no triangle and 4-cycle at any vertex v i in H 2 . Further, there are paths P 3 1 ( v 1 v 2 v 3 v 2 v 1 ) , P 3 2 ( v 2 v 3 v 2 v 1 v 2 ) and P 3 1 ( v 3 v 2 v 1 v 2 v 3 ) at the vertices v 1 , v 2 and v 3 in H 2 . With this information, it follows that
a 1 ( H 2 ) = ( n + 1 ) , a 2 ( H 2 ) = ( n + 1 ) 2 2 ( n + 9 ) 2 , a 3 ( H 2 ) = ( n + 1 ) 3 6 + ( n + 1 ) ( n + 9 ) 2 ( n + 13 ) 3 7 , a 4 ( H 2 ) = ( n + 1 ) 4 24 ( n + 1 ) 2 ( n + 9 ) 4 + ( n + 9 ) 2 8 + ( n + 1 ) ( n + 13 ) 3 + 27 n 4 21 .
Example 2.
Let C n = v 1 v 2 v 3 v n 1 v n v 1 be the directed cycle on n 5 vertices with a = n arcs of the form ( v i , v i + 1 ) , ( v n , v 1 ) , i = 1 , 2 , , n 1 . For the directed cycle C n , we have d i + = 1 , for 1 i n , c 2 ( i ) = 0 , M i = 0 for 1 i n , as there is no closed walk of length 2 at any vertex. Additionally, n ( C 3 i ) = 0 = n ( C 4 i ) as there is no triangle and 4-cycle at any vertex v i in P n . Further, it is clear that there are no paths of the form P 3 1 ( v i v j v k v j v i ) and P 3 2 ( v i v j v i v k v i ) in C n . With this information, it follows that
a 1 ( C n ) = n , a 2 ( C n ) = n 2 2 n 2 , a 3 ( C n ) = n 3 6 + n 2 2 n 3 , a 4 ( C n ) = n 4 24 n 3 4 + 11 n 2 24 n 4 .
Let us add arc ( v 2 , v 1 ) in C n and let K 1 be the resulting digraph. For K 1 , it is clear that a = n + 1 , d 1 + = 1 , d 2 + = 2 , d i + = 1 , for i 3 , c 2 ( 1 ) = c 2 ( 2 ) = 1 , c 2 ( i ) = 0 , M 1 = 2 , M 2 = 1 , M i = 0 for i 3 . Additionally, n ( C 3 i ) = 0 = n ( C 4 i ) as there is no triangle and 4-cycle at any vertex v i in K 1 . Further, it is clear that there are no paths of the form P 3 1 ( v i v j v k v j v i ) and P 3 2 ( v i v j v i v k v i ) in K 1 . With this information, it follows that
a 1 ( K 1 ) = ( n + 1 ) , a 2 ( K 1 ) = ( n + 1 ) 2 2 ( n + 5 ) 2 , a 3 ( K 1 ) = ( n + 1 ) 3 6 + ( n + 1 ) ( n + 5 ) 2 n 3 16 3 , a 4 ( K 1 ) = ( n + 1 ) 4 24 ( n + 1 ) 2 ( n + 5 ) 4 + ( n + 1 ) ( n + 7 ) 3 + ( n + 1 ) 2 8 + 11 n 13 4 .
Similar to Laplacian coefficients a 1 , a 2 , a 3 and a 4 we can obtain the signless Laplacian coefficients b 1 , b 2 , b 3 and b 4 of the digraphs P n , H 1 , C n and K 1 .

5. Conclusions

From Theorem 1, we arrive at the following conclusion about the Laplacian spectral determination of digraphs.
Theorem 3.
If digraphs D 1 and D 2 are Laplacian co-spectral, then (i) D 1 and D 2 have the same order;
(ii) D 1 and D 2 have the same number of arcs;
(iii) the quantity i = 1 n ( d i + ) 2 + c 2 ( i ) is same for D 1 and D 2 ;
(iv) the quantity i = 1 n ( d i + ) 3 + 3 i = 1 n c 2 ( i ) d i + i = 1 n n ( C 3 i ) is the same for D 1 and D 2 ;
(v) the quantity i = 1 n ( d i + ) 4 4 i = 1 n ( d i + ) 2 c 2 ( i ) + 2 i = 1 n d i + M i 4 i = 1 n d i + n ( C 3 i ) + i = 1 n n ( C 3 i ) + i = 1 n n ( P 3 1 ( v i v j v k v j v i ) ) + n ( P 3 2 ( v i v j v i v k v i ) ) + i = 1 n c 2 ( i ) is the same for D 1 and D 2 .
From Theorem 2, we arrive at the following conclusion about the signless Laplacian spectral determination of digraphs.
Theorem 4.
If digraphs D 1 and D 2 are signless Laplacian co-spectral, then (i) D 1 and D 2 have the same order;
(ii) D 1 and D 2 have the same number of arcs;
(iii) the quantity i = 1 n ( d i + ) 2 + c 2 ( i ) is the same for D 1 and D 2 ;
(iv) the quantity i = 1 n n ( C 3 i ) + 3 i = 1 n c 2 ( i ) d i + + i = 1 n ( d i + ) 3 is the same for D 1 and D 2 ;
(v) the quantity i = 1 n ( d i + ) 4 4 i = 1 n ( d i + ) 2 c 2 ( i ) + 2 i = 1 n d i + M i 4 i = 1 n d i + n ( C 3 i ) + i = 1 n n ( C 3 i ) + i = 1 n n ( P 3 1 ( v i v j v k v j v i ) ) + n ( P 3 2 ( v i v j v i v k v i ) ) + i = 1 n c 2 ( i ) is the same for D 1 and D 2 .
From (ii) of Theorems 3 and 4, it is clear that if two digraphs D 1 and D 2 have different numbers of arcs, then these digraphs have different (signless) Laplacian eigenvalues.

Author Contributions

Conceptualization, H.A.G. and Y.S.; investigation, H.A.G. and Y.S.; writing—original draft preparation, H.A.G. and Y.S.; writing—review and editing, H.A.G. and Y.S. All authors have read and agreed to the published version of the manuscript.

Funding

This research received no external funding.

Institutional Review Board Statement

Not applicable.

Informed Consent Statement

Not applicable.

Data Availability Statement

Not applicable.

Conflicts of Interest

The authors declare no conflict of interest.

References

  1. Bhat, M.A. Energy of weighted digraphs. Discret. Appl. Math. 2017, 223, 1–14. [Google Scholar] [CrossRef]
  2. Ganie, H.A.; Carmona, J.R. An (increasing) sequence of lower bounds for the spectral radius and energy of digraphs. Discret. Math. 2023, 346, 113118. [Google Scholar] [CrossRef]
  3. Alhevaz, A.; Baghipur, M.; Shang, Y. Merging the spectral theories of distance Estrada and distance signless Laplacian Estrada indices of graphs. Mathematics 2019, 7, 995. [Google Scholar] [CrossRef] [Green Version]
  4. Baghipur, M.; Ghorbani, M.; Ganie, H.A.; Shang, Y. On the second-largest reciprocal distance signless Laplacian eigenvalue. Mathematics 2021, 9, 512. [Google Scholar] [CrossRef]
  5. Chat, B.A.; Ganie, H.A.; Pirzada, S. Bounds for the skew Laplacian spectral radius of oriented graphs. Carpathian J. Math. 2019, 35, 31–40. [Google Scholar] [CrossRef]
  6. Ganie, H.A. Bounds for the skew Laplacian(skew adjacency) spectral radius of a digraph. Trans. Combin. 2019, 8, 1–12. [Google Scholar]
  7. Ganie, H.A.; Chat, B.A. Bounds for the energy of weighted graphs. Discret. Appl. Math. 2019, 268, 91–101. [Google Scholar] [CrossRef]
  8. Ganie, H.A.; Shang, Y. On the spectral radius and energy of signless Laplacian matrix of digraphs. Heliyon 2022, 8, e09186. [Google Scholar] [CrossRef]
  9. Lokesha, V.; Shanthakumari, Y.; Reddy, P.S.K. Skew-zagreb energy of directed graphs. Proc. Jangjeon Math. Soc. 2020, 23, 557–568. [Google Scholar]
  10. Pirzada, S.; Ganie, H.A.; Chat, B.A. On the real or integral spectrum of digraphs. Oper. Matrices 2020, 14, 795–813. [Google Scholar] [CrossRef]
  11. Qiu, L.; Wang, W.; Wang, W. Oriented graphs determined by their generalized skew spectrum. Linear Algebra Appl. 2021, 622, 316–332. [Google Scholar] [CrossRef]
  12. Shang, Y. More on the normalized Laplacian Estrada index. Appl. Anal. Discrete Math. 2014, 8, 346–357. [Google Scholar] [CrossRef]
  13. Sachs, H. Beziehungen zwischen den in einen Graphen enthalten Kreisenund seinem characterischen Polynom. Publ. Math. Debr. 1964, 11, 119–134. [Google Scholar] [CrossRef]
  14. Mowshowitz, A. The characteristic polynomial of a graph. J. Comb. Theory 1972, 12, 177–193. [Google Scholar] [CrossRef] [Green Version]
  15. Oliveira, C.S.; de Abreu, N.M.M.; Jurkiewicz, S. The characteristic polynomial of the Laplacian of graphs in (a, b)-linear classes. Linear Algebra Appl. 2002, 356, 113–121. [Google Scholar] [CrossRef] [Green Version]
  16. Cvetković, D.; Rowlinson, P.; Simić, S.K. Signless Laplacians of finite graphs. Linear Algebra Appl. 2007, 423, 155–171. [Google Scholar] [CrossRef] [Green Version]
  17. Guo, J.M.; Li, J.; Huang, P.; Shiu, W.C. Coefficients of the characteristic polynomial of the (signless, normalized) Laplacian of a graph. Graphs Comb. 2017, 33, 1155–1164. [Google Scholar] [CrossRef]
  18. Lepović, M.; Gutman, I. No starlike trees are cospectral. Discret. Math. 2002, 242, 291–295. [Google Scholar] [CrossRef] [Green Version]
  19. Wang, J.F.; Huang, Q.X.; Belardo, F. On the spectral characterizations of 3-rose graphs. Util. Math. 2013, 91, 33–46. [Google Scholar]
  20. Ganie, H.A.; Ingole, A.; Deshmukh, U. On the coefficients of skew Laplacian characteristic polynomial of digraphs. Discret. Math. Algorithms Appl. 2022, 14, 2250131. [Google Scholar] [CrossRef]
  21. Liu, X.; Liu, S. On the Aα-characteristic polynomial of a graph. Linear Algebra Appl. 2018, 546, 274–288. [Google Scholar] [CrossRef] [Green Version]
  22. Oboudi, M.R.; Abdian, A.Z. Peacock graphs are determined by their Laplacian spectra. Iran. J. Sci. Technol. Trans. Sci. 2020, 44, 787–790. [Google Scholar] [CrossRef]
  23. Lin, H.; Shu, J. A note on the spectral characterization of strongly connected bicyclic digraphs. Linear Algebra Appl. 2012, 436, 2524–2530. [Google Scholar] [CrossRef]
  24. Prasolov, V.V. Problems and Theorems in Linear Algebra (Translations of Mathematical Monographs); American Mathematical Society: Cambridge, UK, 1994. [Google Scholar]
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Ganie, H.A.; Shang, Y. On the Laplacian and Signless Laplacian Characteristic Polynomials of a Digraph. Symmetry 2023, 15, 52. https://doi.org/10.3390/sym15010052

AMA Style

Ganie HA, Shang Y. On the Laplacian and Signless Laplacian Characteristic Polynomials of a Digraph. Symmetry. 2023; 15(1):52. https://doi.org/10.3390/sym15010052

Chicago/Turabian Style

Ganie, Hilal A., and Yilun Shang. 2023. "On the Laplacian and Signless Laplacian Characteristic Polynomials of a Digraph" Symmetry 15, no. 1: 52. https://doi.org/10.3390/sym15010052

APA Style

Ganie, H. A., & Shang, Y. (2023). On the Laplacian and Signless Laplacian Characteristic Polynomials of a Digraph. Symmetry, 15(1), 52. https://doi.org/10.3390/sym15010052

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