In this paper, we study the -spectral radius of graphs in terms of given size and minimum degree , and characterize the corresponding extremal graphs completely. Furthermore, we characterize extremal graphs having maximum -spectral radius among (minimally) -edge-connected graphs with given size .
All graphs considered here are simple and undirected. For a graph
, denotes its adjacency matrix and
denotes the diagonal matrix of
its degrees. The matrix is called the signless
Laplacian matrix of . The largest
eigenvalue of is called the
spectral radius of , and the
largest eigenvalue of is
called the signless Laplacian spectral radius of . For any real number , Nikiforov [1] defined the -matrix of as ,
which can be regarded as a common generalization of and . The largest eigenvalue of is called the -spectral radius of , denoted by . For a connected graph
, by the Perron-Frobenius theory
of non-negative matrices [1],
has multiplicity one
and there exists a unique positive unit eigenvector corresponding to
. We shall refor to
such an eigenvector as the Perron vector of .
The investigation on the extremal problems of the spectral radius and
the signless Laplacian spectral radius of graphs is an important topic
in the theory of graph spectra. For related results, one may refer to
[2-5] and the
references therein. Specially, the problem of characterizing the graph
with maximal spectral radius for given size is initiated by Brualdi and
Hoffman [6], and completely
solved by Rowlinson [7]. For
further investigation, one may refer to [8-13] and the references therein.
Just recently, one of the hot topics in the study of the -spectrum is to characterize the
spectral extreme under the conditions of given size and graph
parameters. Zhai et al. [14]
determined the graph with maximal spectral radius among all graphs with
given size, and characterized the graph with maximal spectral radius among all graphs with
given size and clique number (resp., chromatic number). Lou et al. [15] determined the maximal
signless Laplacian spectral radius (Laplacian spectral radius) of
connected graphs with fixed size and diameter. For more results, one may
refer to [16-19].
The -spectral radius
of a graph has been widely concerned. However, the results on the -spectral radius under
edge-condition are still relatively little known. Li and Qin [20] generalized the conclusion in
[14] to -spectral radius for . Feng et al. [21] and Huang et al. [22] determined independently the
graph having the maximum -spectral radius for among all connected
graphs of size and diameter (at
least) .
A friendship graph is one in which every pair of vertices has exactly
one common neighbour, denoted by for given . The join of graphs
and , denoted by , is the graph obtained from
by joining each vertex of
with every vertex of . In this paper, we completely
characterize the graphs attaining the maximal -index among all graphs with
given size and minimum degree
for .
Figure 1
Theorem 1. Let and be a graph with edges and minimum degree , and be the graphs shown in Figure 1.
If and , then , with equality if and only if
.
If and , then , with
equality if and only if ,
where .
If and , then , with
equality if and only if ,
where .
A graph is -edge-connected if
removing fewer than edges always
leaves the remaining graph connected, and is minimally -edge-connected if it is -edge-connected and deleting any
arbitrary chosen edge always leaves a graph which is not -edge-connected. For graphs of order
, Chen and Guo [23] showed that attained the maximal spectral
radius among all the minimally -(edge)-connected graphs. Fan et al.
[24] proved that has the largest spectral radius
over all minimally -connected
graphs. For graphs of size , Guo
and Zhang [16,19] gave sharp
upper bounds on the -index of
(minimally) -connected graphs with
given size and characterized the corresponding extremal graphs
completely. Noting that a connected graph having no cut edges is -edge-connected, we have following
corollary.
Corollary 1. Let and be a -edge-connected graph with edges.
If and , then , with equality if and only if
.
If and , then , with
equality if and only if ,
where .
If and , then , with
equality if and only if ,
where .
Figure 2
In this paper, we further study the problem of characterizing graphs
among minimally -edge-connected
graph with maximal –
spectral radius. For , let (shown in Figure 2) be the
graph obtained from the friendship graph by subdividing an edge
once. For , let (shown in Figure 2) be the
graph obtained from the friendship graph by subdividing an edge
twice. Employing Theorem 1, we can prove the following theorem.
Theorem 2. Let and be a minimally -edge-connected graph with edges.
If and , then , with equality if and only if
.
If and , then , with
equality if and only if .
If and , then , with
equality if and only if .
The remainder of the paper is organized as follows. In Section 2, we
recall some useful notions and lemmas that will be used later. In
Section 3, we give proofs of Theorems 1 and 2 respectively.
2. Preliminaries
For a graph , and denote the vertex set and edge set
of respectively, and denotes the number of edges
in . For or denotes the degree of , or denotes the set of all neighbors of
in , and . For a subset of , denotes the subgraph of induced by , denotes the number of edges in , and denotes the set of all neighbors
of in . For two disjoint subsets and of , denotes the number of edges with
one endpoint in and the other in
. Let denote the graph obtained from by deleting the edge Similarly, is the graph obtained from by adding an edge where The average degree of the
neighbors of a vertex of is . The degree sequence of is the non-increasing sequence of its
vertex degrees. Whenever necessary, the vertices of can be renumbered so that for . In that case, we say that
has degree sequence , denoted by .
Let be a connected graph on
vertices and . Then
can be considered as a function
defined on , that is, each
vertex is mapped to . One can find in [1] that
and for arbitrary unit vector , with the equality if and only if is the Perron vector of .
In order to prove our main results, we need the following lemmas.
Lemma 1. ([1])If
is a graph with no isolated vertices, then
Lemma 2. ([1])Let
be a graph with vertices and
. If , then The equality holds if and
only if and
is the star .
Lemma 3. ([1])Let
be a connected graph with and be a proper
subgraph of , then .
Lemma 4. ([25])Let
be a connected graph, and be two vertices of . Suppose that and is the
Perron vector of , where corresponds to the vertex . Let be the
graph obtained from by deleting
the edges and adding the edges
. If , then .
An internal path in some graph is a path (, or whenever ) such that , , and for . Li, Chen and Meng [26] proved the following
subdivision theorem.
Lemma 5. ([26])Let
be a connected graph with and be some edge
on an internal path of . Let denote the graph obtained from
by subdivision of edge into edges and . Then .
A cycle of a graph is said to have a chord if there is an
edge of that joins a pair of
non-adjacent vertices of .
Lemma 6. ([11])If
is a minimally -edge-connected
graph, then no cycle of has a
chord.
For a connected graph, Yu, Wu and Shu [27] gave a sharp upper bound on -index in terms of its degree sequence.
The authors of the current paper [28] generalized their result to -index of a connected graph.
The following Lemma is a corollary of our result.
Lemma 7. ([28])Let
be a simple connected graph with
vertices and degree sequence . If , then , where
Figure 3
Lemma 8. Let ,
and be the graph shown in
Figure 3. Then .
Proof. Label the vertices is as shown in Figure 1. Let
be a unit eigenvector corresponding to where corresponds to and corresponds to and corresponds to . By the eigenvalue equation , we have and . It
follows that .
Define
such that for , and . Clearly,
Noting that and , by Lemma 2, we
have . By (1), we have Therefore .
This completes the proof.
3. Proofs of Theorems 1 and 2
Proof of Theorem 1. We may assume that is connected. Otherwise, suppose that
are connected components of , where . Since , then . For , let be a vertex of , and be the graph obtained from by identifying vertices . Clearly, , and
is a connected graph with edges and minimum degree . So, in order to complete
the proof of Theorem 1, we may assume that is connected.
Furthermore, we may assume that is -edge-connected. Otherwise, suppose that
is a cut edge of
. Since , then there exist a path
, where , such that and
belongs to a cycle . Suppose that . Let be a unit eigenvector corresponding to where corresponds to , corresponds to . If , let ; Otherwise,
let . In
the both cases, is a connected
graph with edges and minimum
degree and is not a cut edge of any more. By Lemma 4, we have . So we
may assume that is -edge-connected.
Let denote the
set of all -edge-connected graphs
with edges. For and , it is easy to see that and has no isolated vertices. Noting that
, we have It
follows that
with equality if and only if .
For , let
be a vertex of such that Noting that
and , we have By Lemma 1, we have
Let and . It is easy to see that
. By Lemma 2, we have .
If , noting that , we have By (2), we have for and .
If , noting that , we have By (2), we have for and .
If ,
let . It is easy to see that the function
is convex for and its maximum in any closed
interval is attained at one of the ends of this interval. Combining this
and (4), we have for
and .
If , then
. Let
, then
. It follows that . This implies that . By Lemma 7, we
have for and .
If , then , completing the proof
of (i).
Let and . It is easy to see that
. By Lemma 2, we have .
For , by similar reasoning as in the proof of (i), we
can derived that for and .
If , then
. Let
, then
. It follows that
. This implies that
. By Lemma 7, we
have for and .
If , then
. Let
, then
. It follows that
.
Figure 4
Case 1., it follows that . Noting that
, it is well known that
.
Since , then we known
that might be
If , then .
If , then
or , shown in Figure 4. Let
be a unit eigenvector corresponding to
where corresponds to and corresponds to . If , let ; Otherwise, let
. In the both
cases, . By Lemma 4, we
have .
Applying Lemma 4 to the vertices and of , we can similarly derive that .
Case 2., then . Noting that
, then has degree sequence . It follows that . Noting that is an internal path of , by Lemma 5, we have . Furthermore, noting that is a proper subgraph of
, by Lemma 3, we have .
Therefore, we have .
Let and . It is easy to see that
. By Lemma
2,
we have .
For , by similar reasoning as in the proof of (i), we
can similarly derived that
for and .
If , let
, then
It follows that
. This implies that . By Lemma 7, we
have
for and .
If , let
, then
It follows that
. Noting that , we known that has degree sequence . It follows that , completing the proof of
(iii).
Proof of Theorem 2. Let denote the set of all
minimally -edge-connected graphs
with edges.
Let and . It is easy to see that
. By Theorem 1(i), we have for and the equality holds if and only if
.
Let and . It is easy to see that
. By Lemma 2, we have .
From the proof of Theorem 1(ii), we know that for . This implies that for
, and the
equality holds if and only if .
Let and . It is easy to see that
. By Lemma 2, we have .
For , let
be a vertex of such that where .
For , by similar reasoning as in the proof of
Theorem 1(i), we can prove that
for and .
If , let
, then
It follows that
. This implies that
. By Lemma 7, we
have
for and .
If . let
, then
It follows that
. We consider the
following three cases.
Case 1., then and . Since is minimally -edge-connected, it follows that , where nd are nonnegative integers with . This implies that
, a
contradiction.
Case 2.. Let . Then , and or . If and there exist a vertex such that , then there exist at least two
vertices such that
. Obviously,
we obtain a cycle with
a chord , a contradiction to
Lemma 6.
If and , then , shown in Figure 3. By
Lemma 8, we have .
If , then exists a vertex such that . By Lemma 6, we know that
. It follows
that . Suppose . Noting that , we deduce that there exists
another vertex such
that . If , we obtain a cycle with a chord ; if , we obtain a cycle with a chord . This contracts Lemma 6.
Figure 5
Case 3.. Let . Then and the degree
sequence of must be . Noting that
and , it follows
that or , shown in Figure <5.
Applying Lemma 4 to vertices and of , we can derive . By
Lemma 8, we have .
Therefore .
Combining the above arguments, we complete the proof.
Acknowledgments
The authors are grateful to the anonymous referees for valuable
suggestions and corrections which result in an improvement of the
original manuscript.
Conflict of
Interest
The author declares no conflict of interest.
Funding Information
This research is supported by the National Natural Science Foundation
of China (Nos. 12071411, 12171222).
References:
Nikiforov, V., 2017. Merging the A-and Q-spectral theories. Applicable
Analysis and Discrete Mathematics, 11(1), pp.81-107.
Cvetković, D., Rowlinson, P. and Simić, S., 2010. An Introduction
to the Theory of Graph Spectra, Cambridge University Press,
Cambridge.
Nikiforov, V., 2011. Some new results in extremal graph theory.
arXiv preprint arXiv:1107.1121.
Stanić, Z., 2015. Inequalities for Graph Eigenvalues (Vol.
423). Cambridge University Press.
Stevanoviâc, D., 2015. Spectral Radius of Graphs.
Elsevier.
Brualdi, R.A. and Hoffman, A.J., 1985. On the spectral radius of (0,
1)-matrices. Linear Algebra and its Applications, 65,
pp.133-146.
Rowlinson, P., 1988. On the maximal index of graphs with a prescribed
number of edges. Linear Algebra and its Applications, 110,
pp.43-53.
Bollobás, B. and Nikiforov, V., 2007. Cliques and the spectral
radius. Journal of Combinatorial Theory, Series B, 97(5),
pp.859-865.
Min, G., Lou, Z. and Huang, Q., 2022. A sharp upper bound on the
spectral radius of C5-free/C6-free graphs with given size. Linear
Algebra and its Applications, 640, pp.162-178.
Lin, H., Ning, B. and Wu, B., 2021. Eigenvalues and triangles in
graphs. Combinatorics, Probability and Computing, 30(2),
pp.258-270.
Lou, Z., Gao, M. and Huang, Q., 2022. On the spectral radius of
minimally 2-(edge)-connected graphs with given size. arXiv preprint
arXiv:2206.07872.
Rowlinson, P., 1989. On Hamiltonian graphs with maximal index.
European Journal of Combinatorics, 10(5), pp.489-497.
Zhai, M., Lin, H. and Shu, J., 2021. Spectral extrema of graphs with
fixed size: cycles and complete bipartite graphs. European Journal
of Combinatorics, 95, p.103322.
Zhai, M., Xue, J. and Lou, Z., 2020. The signless Laplacian spectral
radius of graphs with a prescribed number of edges. Linear Algebra
and its Applications, 603, pp.154-165.
Lou, Z., Guo, J.M. and Wang, Z., 2021. Maxima of L-index and Q-index:
graphs with given size and diameter. Discrete Mathematics,
344(10), p.112533.
Guo, S.G. and Zhang, R., 2022. Sharp upper bounds on the Q-index of
(minimally) 2-connected graphs with given size. Discrete Applied
Mathematics, 320, pp.408-415.
Jia, H., Li, S. and Wang, S., 2022. Ordering the maxima of L-index
and Q-index: Graphs with given size and diameter. Linear Algebra and
its Applications, 652, pp.18-36.
Zhai, M., Xue, J. and Liu, R., 2022. An extremal problem on
Q-spectral radii of graphs with given size and matching number.
Linear and Multilinear Algebra, 70(20), pp.5334-5345.
Zhang, R. and Guo, S.G., 2022. Maxima of the Laplacian spectral
radius of (minimally) 2-connected graphs with fixed size. Linear
Algebra and its Applications, 651, pp.390-406.
Li, D. and Qin, R., 2021. The Aα-spectral radius of graphs with a
prescribed number of edges for 12 ≤ α≤ 1. Linear Algebra and its
Applications, 628, pp.29-41.
Feng, Z. and Wei, W., 2022. On the A α-spectral radius of graphs with
given size and diameter. Linear Algebra and its Applications,
650, pp.132-149.
Huang, P., Li, J. and Shiu, W.C., 2022. Maximizing the Aα-spectral radius of graphs with
given size and diameter. Linear Algebra and its Applications,
651, pp.116-130.
Chen, X. and Guo, L., 2019. On minimally 2-(edge)-connected graphs
with extremal spectral radius. Discrete Mathematics, 342(7),
pp.2092-2099.
Fan, D., Goryainov, S. and Lin, H., 2021. On the (signless Laplacian)
spectral radius of minimally k-(edge)-connected graphs for small k.
Discrete Applied Mathematics, 305, pp.154-163.
Nikiforov, V. and Rojo, O., 2018. On the α-index of graphs with pendent
paths. Linear Algebra and its Applications, 550, pp.87-104.
Li, D., Chen, Y. and Meng, J., 2019. The Aα-spectral radius of trees and
unicyclic graphs with given degree sequence. Applied Mathematics and
Computation, 363, p.124622.
Yu, G., Wu, Y. and Shu, J., 2011. Sharp bounds on the signless
Laplacian spectral radii of graphs. Linear Algebra and Its
Applications, 434(3), pp.683-687.
Zhang, R. and Guo, S.G., 2023. An upper bound on the Aα-spectral
radius of Hamiltonian graphs with given size, Advances in
Mathematics,(China), accepted.