Next Article in Journal
Extended Half-Power Exponential Distribution with Applications to COVID-19 Data
Previous Article in Journal
Non-Parametric Semi-Supervised Learning in Many-Body Hilbert Space with Rescaled Logarithmic Fidelity
 
 
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

The Extremal Structures of r-Uniform Unicyclic Hypergraphs on the Signless Laplacian Estrada Index

1
School of Science, Xi Jing University, Xi’an 710123, China
2
College of Mathematics and Statistics, South Central University for Nationalities, Wuhan 430074, China
*
Author to whom correspondence should be addressed.
Mathematics 2022, 10(6), 941; https://doi.org/10.3390/math10060941
Submission received: 10 February 2022 / Revised: 7 March 2022 / Accepted: 10 March 2022 / Published: 15 March 2022

Abstract

:
S L E E has various applications in a large variety of problems. The signless Laplacian Estrada index of a hypergraph H is defined as S L E E ( H ) = i = 1 n e λ i ( Q ) , where λ 1 ( Q ) , λ 2 ( Q ) , , λ n ( Q ) are the eigenvalues of the signless Laplacian matrix of H. In this paper, we characterize the unique r-uniform unicyclic hypergraphs with maximum and minimum S L E E .

1. Introduction

1.1. Basic Concepts

Let H = ( V ( H ) , E ( H ) ) be a hypergraph, V ( H ) = { v 1 , v 2 , , v n } and E ( H ) = { e 1 , e 2 , , e m } P ( V ) be the vertex set and edge set, and P ( V ) be the power set of V ( H ) . If each edge e i E ( H ) has precisely r vertices ( r 2 ) , H is a r-uniform hypergraph. If r = 2 , H is an ordinary graph. Let d H ( v ) be the degree of a vertex v, which is the number of edges containing v. Let Δ ( H ) , δ ( H ) and d ( H ) = v V ( H ) d H ( v ) n be the maximum, minimum and average degrees.
The cyclomatic number of a graph is defined as E + P N , where E is the number of edges, P is the number of parts and N is the number of nodes. For a r-uniform connected hypergraph H, the cyclomatic number c ( H ) of H is m ( r 1 ) n + 1 . If c ( H ) = 0 (resp. 1), the corresponding r-uniform connected hypergraph H is exactly a hypertree (resp. unicyclic hypergraph). Denote S n , r as the r-uniform hyperstar with one common vertex for all edges. T n , r (resp. U n , r ) denotes the set of r-uniform hypertrees (resp. unicyclic hypergraph) with n vertices. For v V ( H ) , H v denotes the hypergraph obtained from H by deleting v and all the edges incident with v. For a subset A V ( H ) with A , H [ A ] denotes the subgraph of H induced by A, where H [ A ] has the vertex set A and the edge set { V ( e i ) A | e i E ( H ) , 1 i m } . For a vertex v e E ( H ) with | e | 3 , we apply a v-shrinking on e when we just remove v from the edge e.
For u , v V ( H ) , a ( u , v ) -semi-walk [1] is a vertex–edge alternative sequence u ( = u 1 ) , e 1 , u 2 , e 2 , , u s ,   e s , u s + 1 ( = v ) , where u i , u i + 1 e i ( u i , u i + 1 is not necessarily distinct) for i [ s ] = { 1 , 2 , , s } . If u i u i + 1 for i [ s ] , the ( u , v ) -semi-walk is called a ( u , v ) -walk. Further, we name the ( u , v ) -walk as a ( u , v ) -path, if u 1 , u 2 , , u s + 1 are distinct vertices and e 1 , e 2 , , e s are distinct edges. If v e i { u i , u i + 1 } for i [ s ] , the degree of v is 1, we call the ( u , v ) -path a loose ( u , v ) -path. A loose ( u , u ) -path is also called a loose cycle. We call the smallest length of all ( u , v ) -paths the distance d H ( u , v ) between u and v. For a hypergraph H = ( V ( H ) , E ( H ) ) , its incidence matrix is I ( H ) = ( I i j ) n × m , where
I i j = 1 i f v i e j , 0 o t h e r w i s e ,
and its adjacent matrix is A ( H ) = ( a i j ) n × n , where a i i = 0 and a i j = | { e E ( H ) , v i , v j e } | for i j . The signless Laplacian matrix Q ( H ) = ( ( Q ( H ) ) i j ) of H is I ( H ) I ( H ) T . It is obvious that ( Q ( H ) ) i i = d H ( v i ) for i [ n ] and ( Q ( H ) ) i j = a i j for i j . The signless Laplacian Estrada index S L E E ( H ) [1] of a hypergraph H is defined as
S L E E ( H ) = i = 1 n e λ i ( Q ) ,
where λ 1 ( Q ) , λ 2 ( Q ) , , λ n ( Q ) are the eigenvalues of the signless Laplacian matrix Q ( H ) . We denote by T k ( H ) the k-th signless Laplacian spectral moment of a hypergraph H, that is, T k ( H ) = i = 1 n λ i k ( Q ) , then the signless Laplacian Estrada index can be rewritten as S L E E ( H ) = k = 0 T k ( H ) k ! . From [1], we know that T k ( H ) is exactly equal to the number of closed semi-walks of length k.

1.2. Background and Related Works

Let G = ( V , E ) be a molecular graph (2-uniform hypergraph) represented by a molecular structure, where vertex set V is corresponding to the set of its atoms, and edge set E to its chemical bonds. It is widely used in theoretical chemical research and computational chemistry. A topological index is a single number, which is used to characterize some properties of the graph of a molecule structure. Topological indices are mainly used for the nonempirical quantitative structure–property/activity relationship (QSPR/QSAR) [2,3,4].
Based on the adjacency matrix A of 2-uniform hypergraph G, Estrada [5] put forward the Estrada index E E ( H ) as follows:
E E ( G ) = i = 1 n e μ i ( A ) ,
where μ 1 ( A ) , μ 2 ( A ) , , μ n ( A ) are the eigenvalues of A. Many applications in a large variety of problems on this index have been proposed. For examples, it was successfully employed to quantify the degree of folding of long-chain molecules, especially proteins [6,7], and to measure the centrality of complex (reaction, metabolic, communication, social, etc.) networks [8,9,10], network robustness against failures/attacks [11,12]. There is also a connection between the Estrada index and the extended atomic branching of molecules [13].
The signless Laplacian matrix Q ( G ) has many positive properties, such as being symmetric, irreducible, positive semi-definite, the multiplicity of 0 as a eigenvalue of Q ( G ) is equal to the number of bipartite connected components of G, and so on. The study on Q ( G ) has received extensive attention by many researchers. For example, in [14], Ayyaswamy et al. generalized the definition of the Estrada index to the signless Laplacian Estrada index, many extremal and mathematics properties on this index were obtained [15,16,17,18,19,20].
Graph models are widely used to represent pairwise relationships between entities; however, real-world phenomena can be rich in multi-way relationships involving interactions among more than two entities. Then graph models do not adequately describe these types of applications [21]. Hypergraphs are generalizations of graphs in which edges may connect any number of vertices, and as such, hypergraphs are the natural model of multi-way relationships. It is natural to generalize the signless Laplacian Estrada index S L E E of graphs to hypergraphs. However, we obtained a few results on S L E E of hypergraphs. This paper is a new attempt to study on this topic and is expected to attract more and more attention in the near future. Recently, H.Y. Lu et al. [1] provided lower and upper bounds for S L E E on r-uniform hypergraphs, and characterized the hypertrees with the largest and the smallest S L E E among all r-uniform hypertrees, respectively. Along this line, in this paper, we continue to study S L E E for r-uniform unicyclic hypergraphs.

2. Preliminaries

In this section, we will introduce some lemmas to prove our main results.
In a hypergraph H, let S W k ( H ; u , v ) be the set of all ( u , v ) -semi walks of length k, and S W k ( H ; u ) be the set of all ( u , u ) -semi walks of length k. Denote S W k ( H ) = u V ( H ) S W k ( H ; u ) , then | S W k ( H ) | = T k ( H ) , that is, the number of closed semi-walks of length k in H.
For two hypergraphs G and H with x , y V ( G ) and u , v V ( H ) , if | S W k ( G ; x , y ) | | S W k ( H ; u , v ) | for any k 0 , we say ( G ; x , y ) s ( H ; u , v ) . Moreover, if ( G ; x , y ) s ( H ; u , v ) and there exists some k 0 such that | S W k 0 ( G ; x , y ) | < | S W k 0 ( H ; u , v ) | , then we write ( G ; x , y ) s ( H ; u , v ) . Similarly, if | S W k ( G ; x ) | | S W k ( G ; y ) | for any k 0 , we call ( G ; x ) s ( G ; y ) , further if ( G ; x ) s ( G ; y ) , and there is some k 0 such that | S W k 0 ( G ; x ) | < | S W k 0 ( G ; y ) | , we write ( G ; x ) s ( G ; y ) .
For a hypergraph H, let E = { e i E ( H ) , u i e i } . By applying the u i -shrinking on each edge e i in E , denote the resulted edge by e i , let E 0 = { e i : e i E } . For u , v V ( H ) { V ( e i ) : e i E 0 } , let E u = { e i { u } , e i E 0 } and E v = { e i { v } , e i E 0 } , respectively. Further, H u = H E 0 + E u denotes the hypergraph obtained from H by deleting edges in E 0 and adding edges in E u . Similarly, we can attain hypergraph H v = H E 0 + E v .
Lemma 1
([1]). For a hypergraph H, let E 0 , H u , H v be defined as above, respectively. For u , v V ( H ) { V ( e i ) : e i E 0 } and w { V ( e i ) : e i E 0 } , if ( H ; u ) s ( H ; v ) and ( H ; w , u ) s ( H ; w , v ) , then S L E E ( H u ) S L E E ( H v ) .
H 1 and H 2 denote two disjoint connected hypergraphs and v V ( H 1 ) , w V ( H 2 ) , respectively. H 1 ( v ) ( w ) H 2 denotes the hypergraph obtained from H 1 and H 2 by identifying v of H 1 with w of H 2 .
Lemma 2
([1]). Let H 1 and H 2 be two disjoint connected hypergraphs with u , v V ( H 1 ) and w V ( H 2 ) , respectively. If ( H 1 ; u ) s ( H 1 ; v ) , then S L E E ( H 1 ( u ) ( w ) H 2 ) < S L E E ( H 1 ( v ) ( w ) H 2 ) .
Lemma 3
([1]). Let e be a cut edge of hypergraph H with u , v e and | e | 3 . When d H ( u ) 2 and d H ( v ) = 1 , ( H ; v ) s ( H ; u ) .
H is a connected r-uniform hypergraph. Let u V ( H ) and
P p ( r 1 ) + 1 , r = u 0 , e 1 , u 1 , e 2 , , e p 1 , u p 1 , e p , u p + 1
P q ( r 1 ) + 1 , r = v 0 , f 1 , v 1 , f 2 , , f q 1 , v q 1 , f q , v q + 1
be two disjoint r-uniform loose paths, where p , q 1 . Denote H u ( p , q ) as the hypergraph obtained from H, P p ( r 1 ) + 1 , r and P q ( r 1 ) + 1 , r by coalescing u , u 0 , v 0 as a new vertex, denoted by u. Obviously, if G = P p ( r 1 ) + 1 , r ( u 0 ) ( v 0 ) P q ( r 1 ) + 1 , r , then H u ( p , q ) = G ( u 0 ) ( u ) H (as shown in Figure 1).
Lemma 4.
Let H u ( p , q ) be a hypergraph defined as above (as shown in Figure 1). If p q 1 , S L E E ( H u ( p + 1 , q 1 ) ) < S L E E ( H u ( p , q ) ) .
Proof. 
Let P p ( r 1 ) + 1 , r , P q ( r 1 ) + 1 , r be the loose paths defined in (1) and (2), respectively. In G = P p ( r 1 ) + 1 , r ( u 0 ) ( v 0 ) P q ( r 1 ) + 1 , r , if q = 1 , d G ( v 1 ) = 1 , and if q 2 , d G ( v 1 ) = 2 . Let f 1 { v 0 , v 1 } = { w 1 , w 2 , , w r 2 } . After applying the v 0 -shrinking on f 1 of P q ( r 1 ) + 1 , r , the resulting hypergraph is denoted by Q 1 . Let Q be the hypergraph obtained from P p ( r 1 ) + 1 , r by adding f 1 at u 0 . After applying the v 1 -shrinking on f 1 of Q, the resulting hypergraph is denoted by Q 2 . In G, we have
S W k ( G ; v 1 ) = S W k ( Q 1 ; v 1 ) S W k ( G ; v 1 , [ v 0 ] ) S W k ( G ; v 0 ) = S W k ( Q 2 ; v 0 ) S W k ( G ; v 0 , [ v 1 ] )
Suppose that k 1 , k 2 1 and 1 i r 2 . For any W S W k ( G ; v 1 , [ v 0 ] ) , W can be decomposed into W 1 W 2 , where W 1 is either a walk which consists of a ( v 1 , v 1 ) -section of Q 1 with length ( k 1 1 ) and a ( v 1 , f 1 , v 0 ) -section of length 1, or a walk which consists of a ( v 1 , w i ) -section of Q 1 with length ( k 1 1 ) and a ( w i , f 1 , v 0 ) -section of length 1, and W 2 is a ( v 0 , v 1 ) -section of G with length k 2 . Then, we have
| S W k ( G ; v 1 ) | = | S W k ( Q 1 ; v 1 ) | + | S W k ( G ; v 1 , [ v 0 ] ) | = | S W k ( Q 1 ; v 1 ) | + k 1 + k 2 = k | S W k 1 1 ( Q 1 ; v 1 ) | | S W k 2 ( G ; v 0 , v 1 ) | + i = 1 r 2 k 1 + k 2 = k | S W k 1 1 ( Q 1 ; v 1 , w i ) | | S W k 2 ( G ; v 0 , v 1 ) | | S W k ( G ; v 0 ) | = | S W k ( Q 2 ; v 0 ) | + | S W k ( G ; v 0 , [ v 1 ] ) | = | S W k ( Q 2 ; v 0 ) | + k 1 + k 2 = k | S W k 1 1 ( Q 2 ; v 0 ) | | S W k 2 ( G ; v 1 , v 0 ) | + i = 1 r 2 k 1 + k 2 = k | S W k 1 1 ( Q 2 ; v 0 , w i ) | | S W k 2 ( G ; v 1 , v 0 ) |
For p q 1 , Q 1 is isomorphic to a proper subgraph of Q 2 , then we have | S W k ( Q 1 ; v 1 ) | | S W k ( Q 2 ; v 0 ) | for k 0 and | S W k 1 1 ( Q 1 ; v 1 , w i ) | | S W k 1 1 ( Q 2 ; v 0 , w i ) | for k 1 1 and 1 i r 2 . Furthermore, we find | S W 2 ( q + 1 ) ( Q 1 ; v 1 ) | < | S W 2 ( q + 1 ) ( Q 2 ; v 0 ) | . Therefore, we obtain | S W k ( G ; v 1 ) | | S W k ( G ; v 0 ) | for k 0 and there exists an integer k 0 = 2 ( q + 1 ) such that | S W k ( G ; v 1 ) | < | S W k ( G ; v 0 ) | . Thus, we obtain ( G ; v 1 ) s ( G ; v 0 ) . By Lemma 2, we obtain S L E E ( G ( v 1 ) ( u ) H ) < S L E E ( G ( v 0 ) ( u ) H ) . Since H u ( p + 1 , q 1 ) G ( v 1 ) ( u ) H and H u ( p , q ) G ( v 0 ) ( u ) H , thus we have S L E E ( H u ( p + 1 , q 1 ) ) < S L E E ( H u ( p , q ) ) . □
Let P p ( r 1 ) + 1 , r , P q ( r 1 ) + 1 , r be the two loose paths defined in (1) and (2) respectively. For convenience, let c = p ( r 1 ) + 1 , d = q ( r 1 ) + 1 , V ( P p ( r 1 ) + 1 , r ) = { x 1 , x 2 , , x c } and V ( P q ( r 1 ) + 1 , r ) = { y 1 , y 2 , , y d } , respectively. Let e = { w 1 , w 2 , , w r } be an edge of a connected r-uniform hypergraph H. Denote H w r , w 1 ( p , q ) as the r-uniform hypergraph obtained from H, P c , r and P d , r by identifying w r of H with x c of P c , r and identifying w 1 of H with y d of P d , r , as shown in Figure 2.
Lemma 5.
Let H w r , w 1 ( p , q ) be the r-uniform hypergraph as shown in Figure 2. If H P r , r and p > q 0 , S L E E ( H w r , w 1 ( p + 1 , q ) ) < S L E E ( H w r , w 1 ( p , q + 1 ) ) .
Proof. 
If q = 0 in H w r , w 1 ( p , q ) , let w 1 = y 1 . For simplicity, let H ˜ = H w r , w 1 ( p , q ) . We obtain R 1 by applying the x c -shrinking of P c , r . Denote V 2 = V ( H ˜ ) { x 1 , x 2 , , x d + r 1 } and R 2 = H ˜ [ V 2 ] = ( V 2 , E 2 ) , where E 2 = { e V 2 , e E ( H ˜ ) } . Since H P r , r and p > q , R 1 is isomorphic to a proper subgraph R 2 . For any W S W k ( H ˜ ; x 1 ) , W can be decomposed into two sections according to whether it passes x c (namely w r ) or not. Then
S W k ( H ˜ ; x 1 ) = S W k ( R 1 ; x 1 ) S W k ( H ˜ ; x 1 , [ x c ] ) .
Suppose that W passes through x i ( 1 i c ) , then we construct a mapping f as follows:
(i)
If 1 i d , we replace x i in W by y i ;
(ii)
If d + 1 i d + r 2 , we replace x i in W by w i ( d 1 ) ;
(iii)
If d + r 1 i c , we replace x i in W by x ( c + d + r 1 ) i .
If W S W k ( R 1 ; x 1 ) , we construct a mapping f 1 as follows: f 1 ( W ) = f ( W ) . Obviously, f 1 ( W ) S W k ( R 2 ; y 1 ) S W k ( H ˜ ; y 1 ) . Since R 1 is isomorphic to a proper subgraph of R 2 , f 1 is an injection but not a bijection from S W k ( R 1 ; x 1 ) to S W k ( R 2 ; y 1 ) .
If W S W k ( H ˜ ; x 1 , [ x c ] ) , we can decompose W into W 1 W 2 W 3 W 4 W 5 , where
(i)
W 1 is the longest ( x 1 , x d + r 1 ) -section;
(ii)
W 2 is the longest ( x d + r 1 , x c ) -section;
(iii)
W 3 is the longest ( x c , x c ) -section;
(iv)
W 4 is the longest ( x c , x d + r 1 ) -section;
(v)
W 5 is the longest ( x d + r 1 , x 1 ) -section.
Now, we construct a mapping f 2 ( W ) = f ( W 1 ) W 3 f ( W 2 ) f ( W 4 ) f ( W 5 ) , where
(i)
f ( W 1 ) is the longest ( y 1 , w r ) -section;
(ii)
f ( W 2 ) is the longest ( w r , x d + r 1 ) -section;
(iii)
f ( W 4 ) is the longest ( x d + r 1 , w r ) -section;
(iv)
f ( W 5 ) is the longest ( w r , y 1 ) -section.
Then f 2 ( W ) S W k ( H ˜ ; y 1 , [ x d + r 1 ] ) S W k ( H ˜ ; y 1 ) . Further, we construct a mapping f ,
f ( W ) = f 1 ( W ) W S W k ( R 1 ; x 1 ) f 2 ( W ) W S W k ( H ˜ ; x 1 , [ x c ] )
Obviously, f ( W ) S W k ( H ˜ ; y 1 ) and f is injective but not bijective from S W k ( H ˜ ; x 1 ) to S W k ( H ˜ ; y 1 ) . Then ( H ˜ ; x 1 ) s ( H ˜ ; y 1 ) , that is, ( H w r , w 1 ( p , q ) ; x 1 ) s ( H w r , w 1 ( p , q ) ; y 1 ) . By Lemma 2, we have S L E E ( H w r , w 1 ( p + 1 , q ) ) < S L E E ( H w r , w 1 ( p , q + 1 ) ) . □
Lemma 6.
Let p , q , l be three integers with 0 p < q , p + q l 1 and a = [ p + q 2 ] (that is, the largest integer is no more than p + q 2 ). In a connected r-uniform hypergraph H, let P = x 1 , e 1 , x r , e 2 , , e l , x l ( r 1 ) + 1 be a path with length l and d H ( x 1 ) = 1 , and v = x p ( r 1 ) + 1 and u = x q ( r 1 ) + 1 . Further if p + q is even, Q = x 1 e 1 x r e 2 e a x a ( r 1 ) + 1 is a loose path; if p + q is odd, Q = x 1 e 1 x r e 2 e a + 1 x a ( r 1 ) + r is a loose path (as shown in Figure 3). Then ( H ; v ) s ( H ; u ) and ( H ; v , w ) s ( H ; u , w ) for any w { x t ( r 1 ) + 1 , t > a } .
Proof. 
If p + q is even, by applying the x a ( r 1 ) + 1 -shrinking of P, let R 1 be the component containing x 1 and R 2 be the component containing x c . Obviously, R 1 is a loose path and R 1 is isomorphic to a proper subgraph of R 2 .
In H, we can classify ( v , v ) -closed walks of length k 1 into two types according to whether they pass x a ( r 1 ) + 1 or not. Namely, we obtain S W k ( H ; v ) = S W k ( R 1 ; v ) S W k ( H ; v , [ x a ( r 1 ) + 1 ] ) . Let W be a walk passing x t ( r 1 ) + 1 , where 1 t < a . We construct a mapping φ as follows: we replace x t ( r 1 ) + 1 for 1 t < a by x t ( r 1 ) + 1 for t = p + q t in W.
For W S W k ( R 1 ; v ) , we construct a mapping φ 1 as follows: let φ 1 ( W ) = φ ( W ) . We can check φ 1 ( W ) S W k ( R 2 ; u ) S W k ( H ; u ) . Since R 1 is isomorphic to a proper subgraph of R 2 , φ 1 is an injection but not a bijection from S W k ( R 1 ; v ) to S W k ( R 2 ; u ) .
For W S W k ( H ; v ; [ x a ( r 1 ) + 1 ] ) , we decompose W into W 1 W 2 W 3 , where W 1 and W 3 are semi-edge walks in P, and W 2 S W k ( H ; x a ( r 1 ) + 1 ) is as long as possible. Let φ 2 ( W ) = W 1 W 2 W 3 , and we can check φ 2 ( W ) S W k ( H ; u ; [ x a ] ) S W k ( H ; u ) .
Finally, for W S W k ( H ; x p ) , we can construct a mapping φ as follows:
φ ( W ) = φ 1 ( W ) W S W k ( R 1 ; v ) φ 2 ( W ) W S W k ( H ˜ ; v , [ x a ( r 1 ) + 1 ] )
Obviously, φ ( W ) S W k ( H ; u ) . Since φ 1 is an injection but not bijective,
S W k ( R 2 ; u ) S W k ( H ; u ; [ x a ( r 1 ) + 1 ] ) = .
φ is an injection but not bijective from S W k ( H ; v ) to S W k ( H ; u ) . Then we obtain ( H ; v ) s ( H ; u ) .
By the same procedure as that for p + q being even, we can prove that Lemma 6 holds for p + q being odd. □
Corollary 1.
For an integer l 3 , let C = x 1 e 1 x r e 2 x ( l 1 ) ( r 1 ) + 1 e l x 1 be the unique cycle in a unicyclic r-uniform hypergraph H. Suppose that H is the hypergraph obtained from H by moving edges in E ( x 1 ) e 1 from x 1 to x r . Then S L E E ( H ) < S L E E ( H ) .
Proof. 
Let P = x 1 e 1 x r e 2 x 2 ( r 1 ) + 1 x ( l 1 ) ( r 1 ) + 1 . Applying Lemma 6 for p = 0 and q = 1 implies that ( H ; v ) s ( H ; u ) . Now the result follows from Lemma 2. □
Note that the result of Corollary 1 holds for any v = v i ( r 1 ) + 1 and u = v ( i + 1 ) ( r 1 ) + 1 because we can rewrite the cycle C as x i ( r 1 ) + 1 e i + 1 x ( i + 1 ) ( r 1 ) + 1 e l x 1 e 1 x ( i 1 ) ( r 1 ) + 1 e i + 1 x i ( r 1 ) + 1 .

3. Extremal r -Uniform Unicyclic Hypergraph on Minimum SLEE

In this section, we determine the extremal r-uniform unicyclic hypergraph with minimum S L E E .
Let U n , r g be the class of the connected r-uniform unicyclic hypergraph with order n and unique cycle C g ( r 1 ) , r , where r 3 and g 2 . For convenience, for any H U n , r g , let E ( C g ( r 1 ) , r ) = { e 1 , e 2 , , e g } and e i = { v ( i 1 ) ( r 1 ) + 1 , , v ( i 1 ) ( r 1 ) + r } for i [ g ] and v ( g 1 ) ( r 1 ) + r = v 1 . If n = g ( r 1 ) , it is easy to see that H C n , r . Let H 1 , , H g ( r 1 ) be the components of H E ( C g ( r 1 ) , r ) , where v i V ( H i ) for i [ g ( r 1 ) ] . So we can denote H by C g ( r 1 ) r ( H 1 , , H g ( r 1 ) ) .
Lemma 7.
Let H = C g ( r 1 ) r ( H 1 , , H g ( r 1 ) ) U n , r g . Then for i [ g ( r 1 ) ]
S L E E ( C g ( r 1 ) r ( H 1 , , H i 1 , P i , H i + 1 , , H g ( r 1 ) ) ) S L E E ( C g ( r 1 ) r ( H 1 , , H i 1 , H i , H i + 1 , , H g ( r 1 ) ) .
The equality holds if, and only if, P i H i , where P i is a loose path of length | E ( H i ) | .
Proof. 
It is trivial if | E ( H i ) | = 1 or 2. Suppose that | E ( H i ) | 3 .
Case 1. There is at least one vertex in V ( H i ) { v i } with a degree of at least 3. Denote u as a vertex of a degree of at least 3. The distance d H i ( u , v i ) is as large as possible in H i . Denote f 1 , f 2 , , f d H ( u ) as all the edges containing u in H. By u-shrinking on f i for i [ d H ( u ) ] , denote the components by T 1 , , T d H ( u ) . Without loss of generality, let v i V ( T 1 ) .
Subcase 1.1 If r = 2 , 2 j d H ( u ) , H [ V ( T j ) { u } ] is a pendant path at u. Repeatedly by Lemma 4, we obtain a hypergraph H with smaller S L E E by replacing j = 2 d H ( u ) H [ V ( T j ) { u } ] by a pendant path P of length | E ( j = 2 d H ( u ) H [ V ( T j ) { u } ] ) | at u.
If H C g ( r 1 ) r ( H 1 , , H i 1 , P i , H i + 1 , , H g ( r 1 ) ) , we obtain the desirable result.
Subcase 1.2 If r 3 , j { 2 , , d H ( u ) } , H [ V ( T j ) { u } ] is not a pendant path at u, then there is at least one edge in H [ V ( T j ) { u } ] with at least three vertices of degree 2. We choose such an edge e = { w 1 , , w r } by requiring that d H ( u , w 1 ) is as large as possible, where d H ( u , w 1 ) = d H ( u , w l ) 1 for 2 l r . Then, there are two pendant paths at different vertices of e; let P and Q be pendant paths at vertices w s and w t of e ( 2 s < t r ) , respectively. Denote p and q ( p q 1 ) as the lengths of P and Q. Then H G w s , w t ( p , q ) , and G = H [ V ( H ) ( V ( P Q ) { w s , w t } ) ] . Let H = G w s , w t ( p + 1 , q 1 ) , by Lemma 5, S L E E ( H ) < S L E E ( H ) , we can obtain a hypergraph H 1 with smaller S L E E by reusing Lemma 5, where H [ V ( T j ) { u } ] is a pendant path at u for any j { 2 , , d H ( u ) } .
Denote a j as the lengths of the pendant path P j at u in H 1 , where 2 j d H ( u ) and a j 1 . Suppose without loss of generality that a 2 a 3 . Then H 1 G u ( a 2 , a 3 ) , where G = H 1 [ V ( H ) ( V ( T 2 ) V ( T 3 ) ) ] . Let H 2 = G u ( a 2 + 1 , a 3 1 ) , by Lemma 5, S L E E ( H 1 ) < S L E E ( H 2 ) .
We can obtain a hypergraph H 3 with smaller S L E E by attaching a pendant path P of length | E ( j = 2 d H ( u ) H [ V ( T j ) { u } ] ) | at u by reusing Lemma 4.
If H 3 ( C g ( r 1 ) r ( H 1 , , H i 1 , P i , H i + 1 , , H g ( r 1 ) ) , we obtain the desired result. Otherwise, repeating the above process on H 3 , we will obtain our desired result.
Case 2. There is no vertex with degree at least 3 in V ( H i ) { v i } .
It is trivial for r = 2 . When r 3 , similar to the proof in Subcase 1.2, we obtain the desirable result. □
Applying Lemma 7, we have the following results.
Corollary 2.
Let H 1 = C g ( r 1 ) r ( P 1 , P 2 , , P g ( r 1 ) ) be a hypergraph obtained from H = C g ( r 1 ) r ( H 1 , , H g ( r 1 ) ) by replacing each H i by a loose path P i with | E ( H i ) | = | E ( P i ) | for i [ g ( r 1 ) ] . Then S L E E ( H 1 ) S L E E ( H ) , and the equality holds if, and only if, H H 1 .
Theorem 1.
Let H 1 = C g ( r 1 ) r ( P 1 , P 2 , , P r 1 , P r , , P g ( r 1 ) ) be a connected r-uniform unicyclic hypergraph with r 3 and g 2 , where P i is a path attached at v i in V ( C g ( r 1 ) r ) for 1 i g ( r 1 ) . Then
S L E E ( C n , r ) < S L E E ( H 1 )
Proof. 
Without loss of generality, let P 1 = v 1 e 1 u 2 e 2 u n 1 1 e n 1 1 u n 1 be the pendant path at v 1 . H 2 is the unicyclic hypergraph obtained from H 1 by moving e 1 from v 1 to u n 1 . By Lemma 2 and Corollary 1, S L E E ( H 2 ) < S L E E ( H 1 ) . By repeating the process on every pendant path, we conclude that S L E E ( C n , r ) < S L E E ( H 1 ) . □
The following theorem shows that the extremal r-uniform unicyclic hypergraph with minimum S L E E is C n , r .
Theorem 2.
For n r 1 3 , let H be a r-uniform unicyclic hypergraph with n vertices, then
S L E E ( C n , r ) S L E E ( H ) ,
and the equality holds if, and only if, H C n , r .

4. Extremal r -Uniform Unicyclic Hypergraph on Maximum SLEE

In this section, we obtain the extremal r-uniform unicyclic hypergraphs with the largest S L E E .
Lemma 8.
Let H = C g ( r 1 ) r ( H 1 , , H g ( r 1 ) ) U n , r g . Then
S L E E ( C g ( r 1 ) r ( H 1 , , H i 1 , H i , H i + 1 , , H g ( r 1 ) ) ) S L E E ( C g ( r 1 ) r ( S 1 , , S i 1 , S i , S i + 1 , , S g ( r 1 ) ) ) ,
and the equality holds if, and only if, S i H i , where S i is a hyperstar with center v i and | E ( S i ) | = | E ( H i ) | .
Proof. 
It is trivial if | E ( H i ) | 1 . Now we assume | E ( H i ) | 2 .
Let H ˇ be the hypergraph with largest S L E E in U n , r g and the diameter of H i be d i for i [ g ( r 1 ) ) ] . Next, we prove d i = 2 for i [ g ( r 1 ) ] .
Without loss of generality, let P = u 0 , h 1 , u 1 , , u d 1 1 , h d 1 , u d 1 be a diametral path of H 1 , where u i 1 , u i h i E ( H 1 ) ( i [ d 1 ] ) and d 1 3 .
We denote by R the component of H ˇ h d 1 1 which contains u d 1 1 . Let G = H ˇ [ V ( H ˇ ) ( V ( R ) { u d 1 1 } ) ] . Then, we have d G ( v d 1 ) = 1 and d G ( v d 2 ) 2 . By Lemma 4, we obtain ( G ; v d 1 ) ( G ; v d 2 ) . Obviously, G ( v d 1 ) ( v d 1 ) R H ˇ . By Lemma 3, we have S L E E ( G ( v d 2 ) ( v d 1 ) R ) > S L E E ( H ˇ ) . This contradicts the maximality of H ˇ . So we obtain d 1 = 2 . By the same procedure, we have d i = 2 , for i = 2 , , g ( r 1 ) . Thus, we have H ˇ C g ( r 1 ) r S 1 , , S i 1 , S i , S i + 1 , , S g ( r 1 ) ) . □
Lemma 9.
Let H ˇ = C g ( r 1 ) r ( S 1 , S 2 , S r 1 , S r , , S g ( r 1 ) ) ) be a connected r-uniform unicyclic hypergraph with r 3 and g 2 , where S i is a r-uniform hyperstar with center v i for 1 i g ( r 1 ) . Then
S L E E ( H ˇ ) < S L E E ( C g ( r 1 ) r ( S 1 , v 2 , v 3 , , v r 1 , S r , , S g ( r 1 ) ) )
where S 1 is a hyperstar with center v 1 and | E ( S 1 ) | = | E ( S 1 ) | + | E ( S 2 ) | + + | E ( S r 1 ) | .
Proof. 
We denote by R the component of H ˇ e 1 which contains v 2 . Let G = H ˇ [ V ( H ˇ ) ( V ( R ) { v 2 } ) ] . Then, we have d G ( v 2 ) = 1 and d G ( v 1 ) 2 . By Lemma 4, we obtain ( G ; v 2 ) ( G ; v 1 ) . Obviously, G ( v 2 ) ( v 2 ) R H ˇ . By Lemma 3, we have S L E E ( H ˇ ) < S L E E ( G ( v 1 ) ( v 2 ) R ) . That is to say, S L E E ( H ˇ ) < S L E E ( C g ( r 1 ) r ( S 1 , v 2 , S 3 , , S r 1 , S r , , S g ( r 1 ) ) ) , where S 1 is a hyperstar with center v 1 and | E ( S 1 ) | = | E ( S 1 ) | + | E ( S 2 ) | .
Repeatedly by the above process on S 3 , S 4 , , S r 1 , we will obtain the desirable result. □
Lemma 10.
For r 3 and g 3 , H = C g ( r 1 ) r ( S 1 , v 2 , , v r 1 , S r , v r + 1 , , S ( i 1 ) ( r 1 ) + r , , v g ( r 1 ) ) , where S ( i 1 ) ( r 1 ) + r are hyperstars with centers v ( i 1 ) ( r 1 ) + r for i = 1 , 2 , , m . Let e be the edge containing v 1 , v r . Let H = C 2 ( r 1 ) r ( S 1 , v 2 , , v r 1 , S r , v r + 1 , , v 2 ( r 1 ) ) be the r-uniform unicyclic hypergraph obtained from H by moving each edge of E ( C g ( r 1 ) r ) e from v ( i 1 ) ( r 1 ) + r to v 1 . Then, S L E E ( H ) < S L E E ( H ) .
Proof. 
If g = 2 , the equality holds. Let g 3 , and C g ( r 1 ) r be the unique cycle of H. Applying Lemma 9, S L E E ( H ) < C ( g 1 ) ( r 1 ) r ( S 1 , v r + 1 , , , S ( i 1 ) ( r 1 ) + r , , v g ( r 1 ) ) , where | S 1 | = | S 1 | + | S r | + 1 . Repeatedly by the above process, we have S L E E ( H ) < S L E E ( H ) . □
Lemma 11.
For r 3 , H = C 2 ( r 1 ) r ( S 1 , v 2 , , v r 1 , S r , v r + 1 , , v 2 ( r 1 ) ) , where S 1 and S r are two hyperstars with centers v 1 and v r and | E ( S 1 ) | | E ( S r ) | . Let H = C 2 ( r 1 ) r ( S 1 , v 2 , v 3 , , v 2 ( r 1 ) ) be the r-uniform unicyclic hypergraph obtained from H by moving all edges in E ( S r ) from v r to v 1 . Then, S L E E ( H ) < S L E E ( H ) .
Proof. 
Let V 1 = V ( S 1 ) { v 2 , v 3 , , v r 1 , v r + 1 , , v 2 r 2 } , V 2 = V ( S r ) { v 2 , v 3 , , v r 1 , v r + 1 , , v 2 r 2 } . Let Q 1 = H [ V 1 ] and Q 2 = H [ V 2 ] . We have S W k ( H ; v 1 ) = S W k ( Q 1 ; v 1 ) S W k ( H ; v 1 , [ v r ] ) and S W k ( H ; v r ) = S W k ( Q 2 ; v r ) S W k ( H ; v r , [ v 1 ] ) . Then we obtain
| S W k ( H ; v 1 ) | = | S W k ( Q 1 ; v 1 ) | + | S W k ( H ; v 1 , [ v r ] ) | = | S W k ( Q 1 ; v 1 ) | + k 1 + k 2 = k | S W k 1 1 ( Q 1 ; v 1 ) | | S W k 2 ( H ; v r , v 1 | ) + i = 1 r 2 k 1 + k 2 = k | S W k 1 1 ( Q 1 ; v 1 , w i ) | | S W k 2 ( H ; v r , v 1 | ) | S W k ( H ; v r ) | = | S W k ( Q 2 ; v r ) | + | S W k ( H ; v r , [ v 1 ] ) | = | S W k ( Q 2 ; v r ) | + k 1 + k 2 = k | S W k 1 1 ( Q 2 ; v r ) | | S W k 2 ( H ; v 1 , v r | ) + i = 1 r 2 k 1 + k 2 = k | S W k 1 1 ( Q 2 ; v r , w i ) | | S W k 2 ( H ; v 1 , v r | )
Since | E ( S 1 ) | | E ( S r ) | 2 , Q 2 is isomorphic to a proper subgraph of Q 1 . Then we have | S W k 1 1 ( Q 1 ; v 1 ) | | S W k 1 1 ( Q 2 ; v r ) | for k 1 1 , | S W k 1 1 ( Q 1 ; v 1 , w i ) | | S W k 1 1 ( Q 2 ; v r , w i ) | for k 1 1 and 1 i r 2 and | S W k 2 ( H ; v r , v 1 | = | S W k 2 ( H ; v 1 , v r | for k 2 1 . Furthermore, we have | S W k ( Q 1 ; v 1 ) | | S W k ( Q 2 ; v 2 ) | . Thus, we obtain ( H ; v r ) s ( H ; v 1 ) . By Lemma 3, we have S L E E ( H ( v r ) ( u 1 ) e ) < S L E E ( H ( v 1 ) ( u 1 ) e ) , where e = { u 1 , u 2 , , u r } is an edge.
Then we show that the hypergraph obtained from H has a larger S L E E by moving an edge in E ( S r ) from v r to v 1 . Repeat the above process to obtain the desired result. □
Theorem 3.
For n r 1 3 , let H be a r-uniform unicyclic hypergraph with n vertices. Then
S L E E ( H ) S L E E ( C 2 ( r 1 ) r ( S 1 , v 2 , v 3 , , v 2 ( r 1 ) ) )
with equality if, and only if, H C 2 ( r 1 ) r ( S 1 , v 2 , v 3 , , v 2 ( r 1 ) ) , where S 1 is a hyperstar with center v 1 .
Proof. 
Denote H as a r-uniform unicyclic hypergraph with n vertices; we denote it by C g ( r 1 ) r ( H 1 , , H g ( r 1 ) ) with r 3 and g 2 , where H i is a r-uniform hypertree for 1 i g ( r 1 ) .
Case 1. g = 2 .
By repeating the use of Lemmas 8, 9 and 11, it is easy to obtain our desirable result.
Case 2. g 3 .
By repeating our use of Lemmas 8 and 9, we can obtain a r-uniform unicyclic hypergraph H satisfying S L E E ( H ) S L E E ( H ) , where H C g ( r 1 ) r ( S 1 , v 2 , , v r 1 , S r , v r + 1 , , S g ( r 1 ) ) . we obtain a r-uniform unicyclic hypergraph H with S L E E ( H ) S L E E ( H ) by Lemma 10, where H C 2 ( r 1 ) r ( S 1 , v 2 , , v r 1 , S r , v r + 1 , v 2 ( r 1 ) ) and S 1 is a hyperstar with center v 1 , and S r is a hyperstar with center v r .
By Lemma 11, we obtain a r-uniform unicyclic hypergraph H satisfying S L E E ( H ) S L E E ( H ) , where H C 2 ( r 1 ) r ( S 1 , v 2 , v 3 , , v 2 ( r 1 ) ) . We can obtain our desirable result. □

5. Conclusions

S L E E is one of the important chemical topological indices and has various applications in a large variety of problems. In addition, molecular structures that cannot be represented in ordinary graphs need to be represented by hypergraphs. It is necessary to study S L E E of hypergraphs. On the basis of the study of r-uniform hypergraphs with respect to S L E E , we study the extremal structures of r-uniform unicyclic hypergraphs. In this paper, we characterize the extremal hypergraphs with maximum and minimum S L E E among r-uniform unicyclic hypergraphs, respectively. Few results on the S L E E of hypergraphs have been obtained. S L E E can be studied by the structure and method of hypergraphs.

Author Contributions

Conceptualization, H.L.; methodology, Z.Z.; writing—original draft, H.L. All authors contributed equally to writing this article. All authors have read and agreed to the published version of the manuscript.

Funding

This work is supported by the Xijing University Research Fund (Program No.XJ200205) and the Special Fund for Basic Scientific Research of Central Colleges, South-Central University for Nationalities (CZZ21014).

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. Lu, H.Y.; Xue, N.; Zhu, Z. The Signless Laplacian Estrada Index of uniform hypergraphs. Int. J. Quantum Chem. 2021, 121, e26579. [Google Scholar] [CrossRef]
  2. Prado-Prado, F.J.; García-Mera, X.; González-Díaz, H. Multi-target spectral moment QSAR versus ANN for antiparasitic drugs against different parasite species. Bioorg. Med. Chem. 2010, 18, 2225–2231. [Google Scholar] [CrossRef] [PubMed]
  3. Prado-Prado, F.J.; Borges, F.; Uriarte, E.; Peréz-Montoto, L.G.; González-Díaz, H. Multi-target spectral moment: QSAR for antiviral drugs vs. different viral species. Anal. Chim. Acta 2009, 651, 159–164. [Google Scholar] [CrossRef] [PubMed]
  4. Prado-Prado, F.J.; Borges, F.; Peréz-Montoto, L.G.; González-Díaz, H. Multi-target spectral moment: QSAR for antifungal drugs vs. different fungi species. Eur. J. Med. Chem. 2009, 44, 4051–4056. [Google Scholar] [CrossRef] [PubMed]
  5. Estrada, E. Characterization of 3D molecular structure. Chem. Phys. Lett. 2000, 319, 713–718. [Google Scholar] [CrossRef]
  6. Estrada, E. Characterization of the folding degree of proteins. Bioinformatics 2002, 18, 697–704. [Google Scholar] [CrossRef]
  7. Estrada, E. Characterization of the amino acid contribution to the folding degree of proteins. Prot.-Struct. Funct. Bioinform. 2004, 54, 727–737. [Google Scholar] [CrossRef]
  8. Estrada, E.; Hatano, N. Statistical-mechanical approach to subgraph centrality in complex networks. Chem. Phys. Lett. 2007, 439, 247–251. [Google Scholar] [CrossRef]
  9. Estrada, E.; Rodríguez-Valázquez, J.A. Spectral measures of bipartivity in complex networks. Phys. Rev. E 2005, 72, 046105. [Google Scholar] [CrossRef] [Green Version]
  10. Estrada, E.; Rodríguez-Valázquez, J.A. Subgraph centrality in complex networks. Phys. Rev. E 2005, 71, 056103. [Google Scholar] [CrossRef] [Green Version]
  11. Shang, Y. Local natural connectivity in complex networks. Chem. Phys. Lett. 2011, 28, 068903. [Google Scholar] [CrossRef]
  12. Shang, Y. Biased edge failure in scale-free networks based on natural connectivity. Indian J. Phys. 2012, 86, 485–488. [Google Scholar] [CrossRef]
  13. Estrada, E.; Rodríguez-Valázquez, J.A.; Randić, M. Atomic branching in molecules. Int. J. Quant. Chem. 2006, 106, 823–832. [Google Scholar] [CrossRef]
  14. Ayyaswamy, S.K.; Balachandran, S.; Venkatakrishnan, Y.B.; Gutman, I. Signless laplacian estrada index. MATCH Commun. Math. Comput. Chem. 2011, 66, 785–794. [Google Scholar]
  15. Ellahi, H.R.; Fath-Tabar, G.H.; Gholami, A.; Nasiri, R. On maximum signless laplacian estrada index of graphs with given parameters. Ars Math. Contemp. 2016, 11, 381–389. [Google Scholar] [CrossRef] [Green Version]
  16. Ellahi, H.R.; Nasiri, R.; Fath-Tabar, G.H.; Gholami, A. The signless laplacian estrada index of unicyclic graphs. Math. Interdiscip. Res. 2017, 2, 155–167. [Google Scholar]
  17. Gao, S.; Liu, H.Q. Sharp bounds on the signless laplacian estrada index of graphs. Filomat 2014, 28, 1983–1988. [Google Scholar] [CrossRef] [Green Version]
  18. Nasiri, R.; Ellahi, H.R.; Fath-Tabar, G.H.; Gholami, A. On maximum signless laplacian estrada index of graphs with given parameters II. arXiv 2014, arXiv:1410.0229. [Google Scholar] [CrossRef] [Green Version]
  19. Wang, K.; Ning, W.; Lu, M. On the signless laplacian Estrada index of bicyclic graphs. Discret. Appl. Math. 2018, 235, 169–174. [Google Scholar] [CrossRef]
  20. Wang, K.; Pan, X.; Ning, W. On the signless laplacian Estrada index of bicyclic cacti. Discret. Appl. Math. 2019, 254, 189–195. [Google Scholar] [CrossRef]
  21. Konstantinova, E.V.; Skorobogatov, V.A. Application of hypergraph theory in chemistry. Discret. Math. 2001, 235, 365–383. [Google Scholar] [CrossRef] [Green Version]
Figure 1. H u ( p , q ) .
Figure 1. H u ( p , q ) .
Mathematics 10 00941 g001
Figure 2. H w r , w 1 ( p , q ) .
Figure 2. H w r , w 1 ( p , q ) .
Mathematics 10 00941 g002
Figure 3. An illustration of graph H in Lemma 6.
Figure 3. An illustration of graph H in Lemma 6.
Mathematics 10 00941 g003
Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Share and Cite

MDPI and ACS Style

Lu, H.; Zhu, Z. The Extremal Structures of r-Uniform Unicyclic Hypergraphs on the Signless Laplacian Estrada Index. Mathematics 2022, 10, 941. https://doi.org/10.3390/math10060941

AMA Style

Lu H, Zhu Z. The Extremal Structures of r-Uniform Unicyclic Hypergraphs on the Signless Laplacian Estrada Index. Mathematics. 2022; 10(6):941. https://doi.org/10.3390/math10060941

Chicago/Turabian Style

Lu, Hongyan, and Zhongxun Zhu. 2022. "The Extremal Structures of r-Uniform Unicyclic Hypergraphs on the Signless Laplacian Estrada Index" Mathematics 10, no. 6: 941. https://doi.org/10.3390/math10060941

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