Skip to main content

On the second largest eigenvalue of networks

Abstract

From predicting the epidemic threshold of a disease outbreak to anticipating the stability of a complex system, analysis of spectra of the adjacency matrices of the underlying networks play a pivotal role. Despite spectra of networks considered as fingerprints of the corresponding complex systems, most works and review articles have circumscribed around the largest eigenvalue (\(\lambda _1\)) only. The second largest eigenvalue of a network that admits many applications in diverse fields, including mathematics and computer science, has not been thoroughly contemplated. This article first reviews existing literature on \(\lambda _2\), predominantly confined to the random regular graphs, followed by the results for various popular model networks. We emphasize the aspect that \(\lambda _2\) shows an entirely different behavior than \(\lambda _1\).

Introduction

Since the resolution of seven bridges of Königsberg problem (Euler 1741), the theory of networks, which deals with the study of complex interacting units represented as graphs, has demonstrated remarkable applications in various real-world systems (Albert and Barabási 2002). With an accelerated development in network science, various communities belonging to different branches of science, economics, biology, and sociology have witnessed growing applications of results and techniques of networks in their respective fields. The spectra of the adjacency matrix of a network have been shown to provide information about various structural properties of the network, as well as the dynamical behavior of the corresponding complex system. For example, the degeneracy of the zero and one eigenvalues provide a clue to the structural symmetries of the underlying networks (Yadav and Jalan 2015). The largest eigenvalue, in addition to capturing the information of the largest degree, is related to the synchronization phenomena of diffusively coupled dynamical units on the network (Restrepo et al. 2005). Further, spectra plays a pivotal role in determining the epidemic threshold of a disease outbreak (Van Mieghem et al. 2008), stability of a system (May 1972), drug identification in complex diseases (Rai et al. 2018), for studying international trade networks (Alhomaidhi et al. 2019) etc. Coupled dynamical evolution on networks can be better understood by the spectra of corresponding Laplacian matrices (Liu et al. 2015). Further, spectra of complex networks have also been studied under the Random matrix theory framework (Jalan and Bandyopadhyay 2007; Erdős et al. 2012). Recently, in Susca et al. (2020), a technique has been proposed to compute the statistics of the second largest eigenvalue and the components of the corresponding eigenvector for sparse symmetric matrices. It is to be noted that various structural properties of networks are computationally hard to determine nevertheless, often, the spectral measures cater good insights into a network structure, and these spectral measures are computationally easier to calculate. For example, calculating various expansion properties of networks is computationally NP-hard in nature. However, fortunately, these properties are closely related to the second largest eigenvalue that we can calculate very fast (in \(O(n^3),\) where n is the number of nodes in the network) (Alon 1995). Note that, though calculating spectra analytically is a troublesome task, there exist analytical approaches to calculate spectra for a few specific networks (Newman et al. 2019; Peixoto 2013; Newman 2019). Further, the expansion properties of graphs have found their applications in computational complexity theory (com 1980), for the determination of robustness of computer networks (Motwani and Raghavan 1995), graph pebbling, parallel sorting algorithms, and theory of error-correcting codes (Hoory et al. 2006), derandomization of random algorithms, analysis of algorithms in computational group theory, quantum cryptography, etc. Further, the second eigenvalue of the Laplacian matrix of a network is equivalent to its algebraic connectivity, also known as the Fiedler eigenvalue (Bapat 2010). Random regular networks form one very important class of networks and have been investigated intensively (Bender 1974; Bender and Canfield 1978; Bollobás 1980; McKay 1984; Wormald 1981), and for these networks, the second largest eigenvalue of the adjacency matrices is equivalent to the algebraic connectivity. If the second eigenvalues of these networks is lesser than their degrees, these networks are connected (Fiedler 1973). To get an overview of second largest eigenvalue, it is worth to see Dehmer and Emmert-Streib (2009), Van Mieghem (2010), Chung (1994).

Most of the studies on the second largest eigenvalue (\(\lambda _2\)) are done for random regular graphs, with sporadic studies for other networks. All these studies suggest that \(\lambda _2\) can yield a good insight into properties of the underlying network structure (Marcus et al. 2013). Particularly, the suitability of a network for a specific application depends on \(\lambda _2\), and often a small value of \(\lambda _2\) is desirable (Hoory et al. 2006). Several classes of networks (graphs) are defined by the value the second largest eigenvalue take, for instance, reflexive graphs or hyperbolic graphs (\(\lambda _2 \le 2\)), Salem graphs (\(\lambda _2 \in [-2, 2]\)), k-regular Ramanujan graphs (\(\lambda _2 \le 2\sqrt{d-1}\)) (Hoory et al. 2006; Maxwell 1978). Further, the second eigenvalue is crucial in the dynamics of the Markov chain. It is found that networks with small \(\lambda _{2}\) rapidly mix for reversible Markov chain yielding a high rate of information diffusion in large networks (Desai and Rao 1993).

The article is organized as follows. “Relationship of spectra with structure of networks” section presents results for various analytical relationships between the network’s structural properties and their spectral measures. “Results on various network models” section focuses on the results on few popular model networks, namely, 1D lattice, Erdös-Renýi (ER) random, Scale-free networks and small-world networks. Exact analytical results are presented for 1D lattice model and \(k-\)random regular graphs, and numerical results are for other model networks. “Conclusion” section concludes the article.

Relationship of spectra with structure of networks

Let \(G=(V, E)\) be a network on N nodes, where VE are the vertex-set and edge-set, respectively. If two distinct vertices \(i, j \in V\) are adjacent, we write \(i \sim j\), otherwise, \(i \not \sim j\). We represent network G by its adjacency matrix \(A=(a_{ij})_{N \times N}\) defined as follows.

$$\begin{aligned} a_{ij} = {\left\{ \begin{array}{ll} 1~~\text{ if } i \sim j, \\ 0 ~~ \text{ otherwise. } \end{array}\right. } \end{aligned}$$

The eigenvalues of the adjacency matrix A are denoted by \(\left\{ \lambda _{1}, \lambda _{2}, \lambda _{3},\ldots ,\lambda _{N}\right\} \) where \(\lambda _{1} \ge \lambda _{2} \ge \)...\( \ge \lambda _{N}\). For un-directed and unweighted networks, A is a symmetric matrix with all eigenvalues being real. The set of the eigenvalues \(\{\lambda _1, \lambda _2, \ldots , \lambda _N\}\) is known as the spectra of the network G. It is important to note that \(\lambda _1\) for adjacency matrix having all non-negative entries is always positive, and for connected networks, \(\lambda _1 > \lambda _2\) due to Perron-Frobenius Theorem. Let \(\langle k \rangle \) denote the average degree of G, and \(k_{\max }\) denote the maximum degree of G. The largest eigenvalue \(\lambda _1\) satisfies \(\langle k \rangle \le \lambda _1 \le k_{\max }\) (Sarkar and Jalan 2018). Also \(\lambda _1 \le \big (\frac{2M(N-1)}{N}\big )^{\frac{1}{2}},\) where MN are the number of edges and nodes in the graph, respectively. If G is connected, then \(\chi (G) \le 1+\lambda _1(G),\) where \(\chi (G)\) is the chromatic number of G (Bapat 2010). The chromatic number \(\chi (G)\) of graph G is the minimum number of colors required to color the vertices such that adjacent vertices get distinct colors. However, finding the chromatic number of a graph is an NP-hard problem.

Next, we see that along with \(\lambda _1\), other eigenvalues of A implicate the following information about the structure of graph G (Bapat 2010).

  1. 1

    \(\lambda ^2_1 + \lambda ^2_2 + \dots + \lambda ^2_N = 2M.\)

  2. 2

    \(\sum _{i, j, i\ne j} \lambda _i\lambda _j= -M\).

  3. 3

    \(\lambda ^3_1 + \lambda ^3_2 + \dots + \lambda ^3_N=6|T|,\)

where T is the set of all the cycles of length 3 in G (also referred as triangles). In general, the number of closed walks of length k in G equals \(\sum _{i}^N \lambda _i^k\). A walk of length k in a graph G is a sequence of vertices \((v_1, v_2,\ldots , v_{k+1})\) such that there is an edge \(v_i \sim v_{i+1}\) for \(i= 1, 2, \ldots , k.\) The walk is closed if \(v_1 = v_{k+1}.\)

In particular when G is random regular graph, \(\lambda _2\) shares an interesting relationship with the expansion coefficient of G and its diameter. The expansion coefficient of G, denoted as h(G) is defined as follows:

$$\begin{aligned} h(G)= \text {min}_{0<|S| \le N/2} \frac{E(S, \bar{S})}{|S|}, \end{aligned}$$
(1)

where \(S \subset V,\) and \(E(S, \bar{S})\) are the set of the edges with one end in S and other in \(\bar{S}\). Alon and Milman (Hoory et al. 2006) proved that if G is a k-random regular graph,

$$\begin{aligned} \frac{k-\lambda _2}{2}\le h(G)\le \sqrt{2k(k-\lambda _2)}. \end{aligned}$$
(2)

This immediately gives the following bounds on \(\lambda _2\).

$$\begin{aligned} k-2h(G) \le \lambda _2 \le k-\frac{h(G)^2}{2k}. \end{aligned}$$
(3)

Nilli (Nilli 1991) proved that if a k-random regular graph has a diameter D, then

$$\begin{aligned} \lambda _2 \ge 2 \sqrt{k-1} \bigg (1-\frac{2}{D-2}\bigg )+\frac{2}{D-2} \end{aligned}$$
(4)

Results on various network models

The relation (Eqs. 24) is valid for random regular networks, i.e., for networks with each node having the same degree. However, the second largest eigenvalue of the adjacency matrices of non-regular networks has not been given much attention. We numerically investigated the behavior of \(\lambda _{2}\) for various model networks, namely, Erdös-Renýi (ER) random, scale-free, 1D lattice (for completeness), and small-world networks. ER random networks are constructed using the ER model ? as follows. Starting with N nodes, each pair of the nodes is connected with a probability \(p=\langle k \rangle /N\). Our analysis has considered connected networks only without self-loops and multiple edges. The degree distribution of ER random networks follows a binomial distribution. Scale-free networks are constructed using BA preferential attachment method (Barabási and Albert 1999) in which each node prefers to connect with the existing higher degree nodes. Starting with a network with \(m_{0}\) nodes, we start adding a new node at each time step having m connections such that \(m \le m_{0}\). The new nodes can be connected to any existing node, and to incorporate preferential attachment, the probability of it getting connected to the ith nodes is proportional to \(k_{i}\) where \(k_{i}\) is the degree of the ith node. Thus, after t time steps, the network will have \(t + m_{0}\) nodes and mt edges. The network evolves into a scale-free network with degree distribution following a power law, that is, \(P(k) \sim k^{-\gamma }\) where \(\gamma = 3\). The 1D lattice considered here consists of a ring lattice with each node connecting exactly with the same number of neighbors on both sides. We construct small-world networks using the Watts and Strogatz algorithm (Watts and Strogatz 1998) as follows. Starting with a 1D lattice in which all the nodes have an equal degree, we rewire each edge of the network with a probability \(p_r\). This procedure of the rewiring allows to transform a 1D lattice with \(p_r = 0\), to a random network with \(p_r = 1\). In the intermediate \(p_{r}\) values, the network manifests the small-world behavior, which is quantified by a very small average shortest path length.

Fig. 1
figure 1

Comparison of \(\lambda _{1}\) and \(\lambda _{2}\) for various model networks as a function of \(\langle k \rangle \) with \(N = 500\) for 20 different network realizations

Fig. 2
figure 2

Plot of \(\lambda _{2} / N\) as a function of \(\langle k \rangle / N\) for various model networks. a 1D, b ER, c SF

First, we discuss available analytical results for \(\lambda _1\) and \(\lambda _2\), followed by numerical results for different model networks for various network parameters. The variation on \(\lambda _2\) is compared with that of \(\lambda _1\). As discussed earlier that \(\lambda _1\) is bounded below by the average degree, and above by \(k_{max}\), hence, an increase in the average degree will result in an increase in the lower bound of \(\lambda _1\). In particular, for a k-random regular network \(\lambda _1\) increases linearly with the average degree as \(\langle k \rangle = k_{max}=\lambda _1\). However, \(\lambda _2\) behaves quite differently. Figure 1 presents the behavior of \(\lambda _{1}\) and \(\lambda _{2}\) with respect to the average degree for various model networks. The value of \(\lambda _1\) increases with the average degree for all the networks. In particular, for 1D lattice in (Fig. 1a) \(\lambda _1=\langle k \rangle \), it is easy to see that the all-one vector \({\textbf {1}}=[1, 1, \ldots , 1]^T\) of order \(N \times 1\) satisfy \(A{\textbf {1}}=\langle k \rangle {\textbf {1}}\). Since A is an non-negative matrix and \({\textbf {1}}\) is an non-negative vector, using Perron-Frobenius theorem, \(\lambda _1\) is the largest eigenvalue which is equal to \(\langle k \rangle \). The second largest eigenvalue of a 1D lattice G on N nodes and having degree k is given by \( \sum _{j=1}^{\frac{k}{2}} 2 \cos {\frac{2 \pi j}{N}}\). This can be seen as follows. Each node in G has an edge with each of immediate previous, next \(\frac{k}{2}\) nodes. The first row of the adjacency matrix A of G is the following

$$\begin{aligned} 0, \overset{\frac{k}{2}}{\overbrace{1, \dots , 1}}, 0, \ldots , 0, \overset{\frac{k}{2}}{\overbrace{1, \dots , 1}}. \end{aligned}$$

The i-th row of A is the row-vector resulting after \(i-1\) right shift of the first row. Thus A is a circulant matrix (for more detail on circulant matrices see (Gray 2006)). The eigenvalues of circulant matrix of order N with the first row as \(c_0, c_1, \dots , c_{n-1}\) are given by \(\lambda _i= \sum _{j=0}^{N-1} c_k e^{\frac{- 2 \pi \iota i j}{N}},\) for \(i=0, 1, \ldots , N-1.\) The imaginary part cancels out because the adjacency matrix is symmetric, hence it will have only real eigenvalues. So we are left with only \(\cos \) terms, which implies, \(\lambda _i= \sum _{j=1}^{\frac{k}{2}} 2 \cos {\frac{2 \pi ji}{N}},\) for \(i=0, 1, \ldots , N-1.\) Thus \(\lambda _2=\sum _{j=1}^{\frac{k}{2}} 2 \cos {\frac{2 \pi j}{N}},\) which follow the trend as is seen in Fig. 1b shown in blue.

For k-random regular networks, \(\lambda _1=k\). For ER random networks, \(\lambda _{1}\) \(\approx \) \([1+o(1)]\langle k \rangle \) (Chung et al. 2004). For scale-free networks, \(\lambda _{1}\) \(\approx \) \(\sqrt{k_{\max }}\) if \(\gamma \) > 2.5 and \(\lambda _{1}\) \(\approx \) \(\frac{\langle k^{2} \rangle }{\langle k \rangle } \)=\( \frac{\sum _{i=1}^N (k_{i})^{2}}{\sum _{i=1}^N (k_{i})}\) if \(2< \gamma < 2.5\) (Sarkar and Jalan 2018). With increase in the average degree, \(\lambda _1\) exhibits a consistent increase for all these networks. However, by keeping the network size fixed, as we increase the average degree of a network, \(\lambda _{2}\) first exhibits an increase until \(\langle k \rangle \sim N/2\), which is not surprising as \(\lambda _1\) also exhibits an increase and one can expect of a similar trend followed by the second largest eigenvalue. However, rather intriguingly, if we increase the average degree further, \(\lambda _2\) reflects a decreasing behaviour with \(\langle k \rangle \). The trend of \(\lambda _2\) is symmetric with respect to \(\langle k \rangle = N/2\) (Fig. 1b).

We consider k-random regular networks for analytically substantiating this behaviour observed numerically for ER and SF networks. Let G be a k-random regular network on N nodes and M edges. Let \(\left\{ \lambda _{1}, \lambda _{2}, \lambda _{3},\ldots ,\lambda _{N}\right\} \) be the eigenvalues of the adjacency matrix of G such that \(k=\lambda _1 \ge \lambda _2 \ge \cdots \ge \lambda _N\) hold. Since

$$\begin{aligned} \lambda ^2_1+\lambda ^2_2+\cdots +\lambda ^2_N=2M. \end{aligned}$$

Dividing both the sides by N we get

$$\begin{aligned} \frac{\lambda ^2_1}{N}+\frac{\lambda ^2_2}{N}+\cdots + \frac{\lambda ^2_N }{N}=\frac{2M}{N}=\langle k \rangle =k. \end{aligned}$$
(5)

We have

$$\begin{aligned} \frac{\lambda ^2_2}{N}+\cdots + \frac{\lambda ^2_N }{N}=\frac{2M}{N}=k-\frac{k^2}{N}. \end{aligned}$$
(6)

Since N is constant, the quantity \(k-\frac{k^2}{N}\) depends on the higher-order terms in L.H.S of the Eq. 6. The quantity \(k-\frac{k^2}{N}\) first increases till k becomes \(\frac{N}{2}\), followed by a decrease, and so does the higher-order terms. Therefore, \(\lambda _2\) after reaching the threshold value starts decreasing.

For any other random network, the explanation is similar to what we deduced for random regular networks. Let G be any random network, for which Eq. 6 becomes

$$\begin{aligned} \frac{\lambda ^2_2}{N}+\dots + \frac{\lambda ^2_N }{N}=\frac{2M}{N}=\langle k \rangle -\frac{\lambda _1^2}{N}. \end{aligned}$$
(7)

Since \(\langle k \rangle \le \lambda _1 \le \Delta \), where \(\Delta \) is the largest degree in G. The RHS of Eq. 7 first takes increasing value till k becomes \(\frac{N}{2}\), followed by a decrease, and so does \(\lambda _2\) and the higher order terms.

Further, in Fig. 2, the plots are shown for \(\frac{\lambda _2}{N}\) vs \(\frac{\langle k \rangle }{N}\) for various network models. The behaviour is similar to what it is for \(\lambda _2\) vs \(\langle k \rangle \) as shown in Fig. 1. The ratio \(\frac{\lambda _2}{N}\) increases when \(\frac{\langle k \rangle }{N}\) increases from 0 to 0.5 and after that it shows a decreasing trend from 0.5 to 1. In every iteration, for 1D lattice, we have plot for double the size of previous network instance, but keeping \(\frac{\langle k \rangle }{N}\) fixed. We see that these curves overlaps which can be explained as follows. The second largest eigenvalue of 1D lattice with degree k is \(\lambda _2=\sum _{j=1}^{\frac{k}{2}} 2 \cos {\frac{2 \pi j}{N}}\). For a given N and k, expanding the summation, we get the following expression

$$\begin{aligned} \lambda _2 =2 \Big ( \cos {\frac{2 \pi }{N}} + \cos {\frac{4 \pi }{N}} + \cos {\frac{6 \pi }{N}}+ \dots + \cos {\frac{ \pi k}{N}}\Big ) \end{aligned}$$
(8)

Now let us change the parameters N, k to \(N'\), \(k'\), respectively such that \(\frac{ k }{N}=\frac{ k' }{N'}\). Let the corresponding second largest eigenvalue for these parameters be \(\lambda _2'\). We can express \(\lambda _2'\) in terms of N and k as follows. We have \(N' = 2N\) and so \(k'=2k\), then

$$\begin{aligned} \lambda _2' = 2 \Big (\Big (\cos {\frac{2 \pi }{N}}+\cos {\frac{\pi }{N}}\Big ) + \Big (\cos {\frac{4 \pi }{N}} + \cos {\frac{3 \pi }{N}}\Big )+ \dots + \Big (\cos {\frac{ \pi k}{N}} + \cos {\frac{ \pi (k-1)}{N}}\Big ) \Big ) \end{aligned}$$
(9)

Next, we have to prove that \(\frac{\lambda _2'}{N'} = \frac{\lambda _2}{N}\), that is, R.H.S of Eq. 9 should be equal to 2 times R.H.S of Eq. 8 as \(N' = 2N\). The consecutive terms of Eq. 9 (enclosed in parentheses) can be written as \(\cos {\frac{ \pi j}{N}} + \cos {\frac{ \pi (j-1)}{N}}\) where \(j = 2,4,6, \dots , k\). The second term can be further written as \(\cos ({\frac{\pi j}{N} - \frac{\pi }{N}})\). Since \(\cos {\theta }\) is a continuous function and \(\cos ({\theta -\delta })\) \(\approx \) \(\cos {\theta }\) when \(\delta \rightarrow 0\), we have \(\cos ({\frac{\pi j}{N} - \frac{\pi }{N}})\) \(\approx \) \(\cos ({\frac{\pi j}{N}})\) for \(N \rightarrow \infty \). Thus, R.H.S of Eq. 9 becomes \(4\times (\cos {\frac{2 \pi }{N}} + \cos {\frac{4 \pi }{N}} + \cos {\frac{6 \pi }{N}}+ \dots +\cos {\frac{\pi k}{N}}\)), which proves that \(\frac{\lambda _2'}{N'} = \frac{\lambda _2}{N}\).

Fig. 3
figure 3

Numerical value of \(\lambda _{2}\) (blue) for 1D lattice with k/2 nearest neighbour couplings, its upper (green) and lower bound (red) as a function of the average degree. Here, \(N = 100\)

Expansion coefficient h(G) of random regular graphs and \(\lambda _2\)

The expansion coefficient h(G) of a network G reveals how well the network is connected. For a disconnected network \(h(G)=0\), and higher the expansion coefficient better well-connected a network is. This can be directly spotted from the expression for the expansion coefficient

$$\begin{aligned} h(G)= \min _{0<|S| \le N/2} \frac{E(S, \bar{S})}{|S|}. \end{aligned}$$
(10)

This means that for any subset S, the number of edges from S to the rest of the vertices is at least \(h(G) \times \min (|S|, |\bar{S}|)\). Therefore, the higher the value of h(G) more well-connected the network is. However, in general calculating h(G) is an intractable problem (NP-hard) (Hoory et al. 2006). Nevertheless, for a 1D lattice, we can easily calculate the expansion coefficient as follows. Consider a 1D lattice of degree k on N nodes. By Eq. 10, the minimum ratio is achieved for \(|S|=\frac{N}{2}\). It is easy to observe that the total number of edges from S to rest of the vertices are

$$\begin{aligned} E(S, \bar{S}) = 2\times \Big (1+2+\dots +\frac{k}{2}\Big ). \end{aligned}$$

Hence,

$$\begin{aligned} h(G)= \frac{k}{N}\times \Big (\frac{k+2}{2}\Big ). \end{aligned}$$

Figure 3 plots \(\lambda _2\) of 1D lattice against the lower and upper bounds in terms of h(G) (Expression 3). It can be seen that the lower bound almost matches the behaviour of \(\lambda _2\) given by the Inequality 2

$$\begin{aligned} \frac{k-\lambda _2}{2}\le h(G)\le \sqrt{2k(k-\lambda _2)}. \end{aligned}$$

by Definition 1, in general for any k-random regular network G, the expansion coefficient h(G) will increase as k increases, thus the quantity, \(k-\lambda _2\) (it is present in both the sides of inequality, also known as spectral gap) must also increase with k. However, in general, calculation of h(G) is NP-hard, it will be interesting to check how \(\lambda _2\) behaves with respect to h(G) for other networks (Hoory et al. 2006).

Conclusion

This article reviews the properties of the second largest eigenvalue for 1D lattice, random regular graphs and networks capturing properties of real-world complex systems. The intriguing conclusion of future interests is that \(\lambda _2\) behaves in a completely different demeanor from the largest eigenvalues. The largest eigenvalue \(\lambda _{1}\) is more dependent on the highest degree of the corresponding networks. The larger the value of \(k_{max}\), the higher will be the value of \(\lambda _{1}\), however \(\lambda _{2}\) though having an upper bound of \(k_{max}\) turns out to be independent of it. For model networks, an increase in the average degree leads to an increase in the value of \(\lambda _{1}\). However, contrarily \(\lambda _{2}\) first manifests an increase till \(\langle k \rangle = \frac{N}{2}\) followed by a decreasing trend. Further, \(\lambda _{1}\) of the scale-free networks is always higher than the 1D lattice and random network since it has higher \(k_{max}\) while \(\lambda _{2}\) is higher for 1D lattice than that of the random and scale-free networks. We have analytically obtained the lower bound of \(\lambda _{2}\) for 1D lattice using its relation with the expansion coefficient and found that the numerical value of \(\lambda _{2}\) follows the same behavior as its lower bound.

Availability of data and materials

Not applicable.

Abbreviations

ER:

Erdös–Renýi

SF:

Scale-free

RRG:

Random regular graph

BA:

Barabási and Albert

L.H.S:

Left hand side

R.H.S:

Right hand side

References

  • Albert R, Barabási A-L (2002) Statistical mechanics of complex networks. Rev Mod Phys 74(1):47

    Article  MathSciNet  Google Scholar 

  • Alhomaidhi A, Al-Thukair F, Estrada E (2019) Gaussianization of the spectra of graphs and networks theory and applications. J Math Anal Appl 470(2):876–897

    Article  MathSciNet  Google Scholar 

  • Alon N (1995) Tools from higher algebra. Handb Comb 2:1749–1783

    MathSciNet  MATH  Google Scholar 

  • Bapat RB (2010) Graphs and matrices, vol 27. Springer

  • Barabási A-L, Albert R (1999) Emergence of scaling in random networks. Science 286(5439):509–512

    Article  MathSciNet  Google Scholar 

  • Bender EA (1974) The asymptotic number of non-negative integer matrices with given row and column sums. Discret Math 10(2):217–223

    Article  MathSciNet  Google Scholar 

  • Bender EA, Canfield ER (1978) The asymptotic number of labeled graphs with given degree sequences. J Comb Theory Ser A 24(3):296–307

    Article  MathSciNet  Google Scholar 

  • Bollobás B (1980) A probabilistic proof of an asymptotic formula for the number of labelled regular graphs. Eur J Comb 1(4):311–316

    Article  MathSciNet  Google Scholar 

  • Chung FRK (1994) Spectral graph theory. regional conference series in mathematics, issn 0160-7642; no. 92. In: CBMS conference on recent advances in spectral graph theory, California State University at Fesno

  • Chung F, Linyuan L, Van V (2004) The spectra of random graphs with given expected degrees. Internet Math 1(3):257–275

    Article  MathSciNet  Google Scholar 

  • Dehmer M, Emmert-Streib F (2009) Analysis of complex networks: from biology to linguistics. Wiley

  • Desai MP, Rao VB (1993) On the convergence of reversible markov chains. SIAM J Matrix Anal Appl 14(4):950–966

    Article  MathSciNet  Google Scholar 

  • Erdős L, Knowles A, Yau H-T, Yin J (2012) Spectral statistics of erdős-rényi graphs ii: eigenvalue spacing and the extreme eigenvalues. Commun Math Phys 314(3):587–640

    Article  Google Scholar 

  • Euler L (1741) Solutio problematis ad geometriam situs pertinentis. Commentarii academiae scientiarum Petropolitanae, pp 128–140

  • Fiedler M (1973) Algebraic connectivity of graphs. Czechoslov Math J 23(2):298–305

    Article  MathSciNet  Google Scholar 

  • Gray RM (2006) Toeplitz and circulant matrices: a review

  • Hoory S, Linial N, Wigderson A (2006) Expander graphs and their applications. Bull Am Math Soc 43(4):439–561

    Article  MathSciNet  Google Scholar 

  • Jalan S, Bandyopadhyay JN (2007) Random matrix analysis of complex networks. Phys Rev E 76(4):046107

    Article  MathSciNet  Google Scholar 

  • Liu H, Dolgushev M, Qi Y, Zhang Z (2015) Laplacian spectra of a class of small-world networks and their applications. Sci Rep 5(1):1–7

    Google Scholar 

  • Marcus A, Spielman DA , Srivastava N (2013) Interlacing families i: bipartite ramanujan graphs of all degrees. In: 2013 IEEE 54th annual symposium on foundations of computer science. IEEE, pp 529–537

  • Maxwell G (1978) Hyperbolic trees. J Algebra 54(1):46–49

    Article  MathSciNet  Google Scholar 

  • May RM (1972) Will a large complex system be stable? Nature 238(5364):413–414

    Article  Google Scholar 

  • McKay BD (1984) Asymptotics for 0-1 matrices with prescribed line sums. Enumeration and Design. Academic Press, pp 225–238

  • Motwani R, Raghavan P (1995) Randomized algorithms. Cambridge University Press

  • Newman MEJ (2019) Spectra of networks containing short loops. Phys Rev E 100(1):012314

    Article  Google Scholar 

  • Newman MEJ, Zhang X, Nadakuditi RR (2019) Spectra of random networks with arbitrary degrees. Phys Rev E 99(4):042309

    Article  Google Scholar 

  • Nilli A (1991) On the second eigenvalue of a graph. Discret Math 91(2):207–210

    Article  MathSciNet  Google Scholar 

  • Peixoto TP (2013) Eigenvalue spectra of modular networks. Phys Rev Lett 111(9):098701

    Article  Google Scholar 

  • Rai A, Shinde P, Jalan S (2018) Network spectra for drug-target identification in complex diseases: new guns against old foes. Appl Netw Sci 3(1):1–18

    Article  Google Scholar 

  • Restrepo JG, Ott E, Hunt BR (2005) Onset of synchronization in large networks of coupled oscillators. Phys Rev E 71(3):036151

    Article  MathSciNet  Google Scholar 

  • Sarkar C, Jalan S (2018) Spectral properties of complex networks. Chaos Interdiscip J Nonlinear Sci 28(10):102101

    Article  MathSciNet  Google Scholar 

  • STOC ’80: Proceedings of the Twelfth Annual ACM Symposium on Theory of Computing, New York, NY, USA, 1980. Association for Computing Machinery. ISBN 0897910176

  • Susca VAR, Vivo P, Kühn R (2020) Second largest eigenpair statistics for sparse graphs. J Phys A Math Theor 54(1):015004

    Article  MathSciNet  Google Scholar 

  • Van Mieghem P (2010) Graph spectra for complex networks. Cambridge University Press

  • Van Mieghem P, Omic J, Kooij R (2008) Virus spread in networks. IEEE/ACM Trans Netw 17(1):1–14

    Article  Google Scholar 

  • Watts DJ, Strogatz SH (1998) Collective dynamics of ‘small-world’networks. Nature 393(6684):440–442

    Article  Google Scholar 

  • Wormald NC (1981) The asymptotic connectivity of labelled regular graphs. J Comb Theory Ser B 31(2):156–167

    Article  MathSciNet  Google Scholar 

  • Yadav A, Jalan S (2015) Origin and implications of zero degeneracy in networks spectra. Chaos Interdiscip J Nonlinear Sci 25(4):043110

    Article  MathSciNet  Google Scholar 

Download references

Acknowledgements

Not applicable.

Funding

S.J. acknowledges Govt of India, BRNS Grant No. 37(3)/14/11/2018-BRNS/37131 and CSIR grant 25(0293)/18/EMR-II for financial support.

Author information

Authors and Affiliations

Authors

Contributions

AM, RS and SJ designed the study and the methodologies, AM and RS processed and analysed the results. All the Authors wrote and approved the manuscript.

Corresponding author

Correspondence to Sarika Jalan.

Ethics declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare that they have no competing interests.

Additional information

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Rights and permissions

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Mishra, A., Singh, R. & Jalan, S. On the second largest eigenvalue of networks. Appl Netw Sci 7, 47 (2022). https://doi.org/10.1007/s41109-022-00484-w

Download citation

  • Received:

  • Accepted:

  • Published:

  • DOI: https://doi.org/10.1007/s41109-022-00484-w

Keywords