8/3/2019 Fractual Boundary of Complex Network
http://slidepdf.com/reader/full/fractual-boundary-of-complex-network 1/8
OFFPRINT
Fractal boundaries of complex networks
Jia Shao, Sergey V. Buldyrev, Reuven Cohen, Maksim
Kitsak, Shlomo Havlin and H. Eugene Stanley
EPL, 84 (2008) 48004
Please visit the new websitewww.epljournal.org
8/3/2019 Fractual Boundary of Complex Network
http://slidepdf.com/reader/full/fractual-boundary-of-complex-network 2/8
Europhysics Letters (EPL) has a new online home at www.epljournal.org
Take a look for the latest journal news and information on:
• reading the latest articles, free!
• receiving free e-mail alerts
• submitting your work to EPL
TAKE A LOOK AT
THE NEW EPL
www.epljournal.org
8/3/2019 Fractual Boundary of Complex Network
http://slidepdf.com/reader/full/fractual-boundary-of-complex-network 3/8
November 2008
EPL, 84 (2008) 48004 www.epljournal.org
doi: 10.1209/0295-5075/84/48004
Fractal boundaries of complex networks
Jia Shao1(a), Sergey V. Buldyrev1,2, Reuven Cohen3, Maksim Kitsak1, Shlomo Havlin4 andH. Eugene Stanley1
1 Center for Polymer Studies and Department of Physics, Boston University - Boston, MA 02215, USA2 Department of Physics, Yeshiva University - 500 West 185th Street, New York, NY 10033, USA3 Department of Mathematics, Bar-Ilan University - 52900 Ramat-Gan, Israel 4 Minerva Center and Department of Physics, Bar-Ilan University - 52900 Ramat-Gan, Israel
received 12 May 2008; accepted in final form 16 October 2008published online 21 November 2008
PACS 89.75.Hc – Networks and genealogical treesPACS 89.75.-k – Complex systemsPACS 64.60.aq – Networks
Abstract – We introduce the concept of the boundary of a complex network as the set of nodes at distance larger than the mean distance from a given node in the network. We studythe statistical properties of the boundary nodes seen from a given node of complex networks. Wefind that for both Erdos-Renyi and scale-free model networks, as well as for several real networks,the boundaries have fractal properties. In particular, the number of boundaries nodes B follows apower law probability density function which scales as B−2. The clusters formed by the boundarynodes seen from a given node are fractals with a fractal dimension df ≈ 2. We present analyticaland numerical evidences supporting these results for a broad class of networks.
Copyright c EPLA, 2008
Many complex networks are “small world” due to thevery small average distance d between two randomlychosen nodes. Often d∼ ln N , where N is the number of nodes [1–6]. Thus, starting from a randomly chosen nodefollowing the shortest path, one can reach any other nodein a very small number of steps. This phenomenon is called“six degrees of separation” in social networks [4]. That is,for most pairs of randomly chosen people, the shortest“distance” between them is not more than six. Manyrandom network models, such as Erdos-Renyi network(ER) [1], Watts-Strogatz network (WS) [5] and scale-free
network (SF) [3,6–8], as well as many real networks, havebeen shown to possess this small-world property.
Much attention has been devoted to the structuralproperties of networks within the average distance d from agiven node. However, almost no attention has been givento nodes which are at distances greater than d from agiven node. We define these nodes as the boundaries of the network and study the ensemble of boundaries formedby all possible starting nodes. An interesting question is:how many “friends of friends of friends etc. . . . ” has one ata distance greater than the average distance d? What istheir probability distribution and what is the structureof the boundaries? The boundaries have an important
(a)E-mail: [email protected]
role in several scenarios, such as in the spread of virusesor information in a human social network. If the virus(information) spreads from one node to all its nearestneighbors, and from them to all next nearest neighborsand further on until d, how many nodes do not get thevirus (information), and what is their distribution withrespect to the origin of the infection?
In this letter, we find theoretically and numericallythat the nodes at the boundaries, which are of orderN , exhibit similar fractal features for many types of networks, including ER and SF models as well as several
real networks. Song et al. [9] found that some networkshave fractal properties while others do not. Propertiesof fractal networks were also studied [10,11]. Here weshow that almost all model and real networks includ-ing non-fractal networks have fractal features at theirboundaries.
Figure 1 demonstrates our approach and analysis. Foreach “root” node, we call the nodes at distance ℓ from it“nodes in shell ℓ”. We choose a random root node andcount the number of nodes Bℓ at shell ℓ. We see thatB1 =10, B2 = 11, B3 = 13, etc. . . . We estimate the aver-age distance (diameter) d≈ 2.9 by averaging the distancesbetween all pairs of nodes. After removing nodes withℓ < d≈ 2.9, the network is fragmented into 12 clusters,with sizes s3={1, 1, 2, 5, 1, 3, 1, 1, 8, 1, 2, 3}. Note that the
48004-p1
8/3/2019 Fractual Boundary of Complex Network
http://slidepdf.com/reader/full/fractual-boundary-of-complex-network 4/8
Jia Shao et al.
cluster of 8 nodes
shell 2
shell 1
Fig. 1: (Color online) Illustration of shells and clusters origi-nating from a randomly chosen root node, which is shown inthe center (red). Its neighboring nodes are defined as shell 1(green), the nodes at distance ℓ are defined as shell ℓ. Whenremoving all nodes with ℓ < 3, the remaining network (purple)becomes fragmented into 12 clusters.
boundary of the network is always seen from a given node,thus not a unique set of nodes in the network.
We begin by simulating ER and SF networks, andthen we present analytical proofs. Figure 2a showssimulation results for the number of nodes Bℓ reached
from a randomly chosen origin node for an ER network.The results shown are for a single network realizationof size N = 106, with average degree k=6 and d≈ 7.9(see footnote 1). For ℓ < d, the cumulative distributionfunction, P (Bl), which is the probability that shellℓ has more than Bℓ nodes, decays exponentially forBℓ > B∗
ℓ , where B∗ℓ is the maximum typical size of shell ℓ
(see footnote 2). However, for ℓ > d, we observe a cleartransition to a power law decay behavior, whereP (Bℓ)∼B−β
ℓ , with β ≈ 1 and the pdf of Bℓ is
P (Bℓ)≡ dP (Bℓ)/dBℓ ∼B−2ℓ . For different networks, the
emergence of the power law can occur at shell ℓ = d +1 orℓ = d + 2. Thus, our results suggest a broad “scale-free”
distribution for the number of nodes at distances largerthan d. This power law behavior demonstrates that thereis no characteristic size and a broad range of sizes canappear in a shell at the boundaries.
In SF networks, the degrees of the nodes, k, followa power law distribution function q(k)∼ k−λ, where theminimum degree of the network, kmin, is chosen to be 2.
Figure 2b shows, for SF networks with λ = 2.5, similarpower law results, P (Bℓ)∼B−β
ℓ , with β ≈ 1 for ℓ > d
1Different realizations yield similar results. In one realization,a certain fraction of nodes are randomly taken to be origin. The
histogram is obtained fromBℓ belonging to different origin nodes.2The behavior of the pdf of Bℓ for ℓ < d will be discussed later
and is shown in fig. 3c.
010
012
014
016
B
014-
013-
012-
011-
010
P
( B )
RE)a(
0.1=β
11=
21=
31=
l
l
l
l
l
010
012
014
016
B
014-
013-
012-
011-
010
P ( B )
FS)b(9=
01=
8=
11=
β 0.1=
l
l
l
l
l
l
010
011
012
013
014
B
01 3-
012-
011-
010
P ( B )
0.1=β
PEH)c(
9=
01=l
l
l
l 11=ll
21=l
01 0 01 1 01 2 01 3 01 4
B
013-
012-
011-
010
P ( B )
0.1=β
SA)d(
6=
7=
4=5=
l
l ll
l
l
Fig. 2: (Color online) The cumulative distribution function,P (Bℓ), for two random network models: (a) ER networkwith N = 106 nodes and k= 6, and (b) SF network withN = 106 nodes and λ = 2.5, and two real networks: (c) theHigh Energy Particle (HEP) physics citations network and(d) the Autonomous System (AS) Internet network. The shellswith ℓ > d are marked with their shell number. The thin linesfrom left to right represent shells ℓ= 1, 2, . . . , respectively,with ℓ < d. For ℓ > d, P (Bℓ) follows a power law distributionP (Bℓ)∼B−βℓ , with β ≈ 1 (corresponding to P (Bℓ)∼B−2ℓ forthe pdf). The appearance of a power law decay only happensfor ℓ larger than d≈ 7.9 for ER and d≈ 4.7 for the SF network.The straight lines possess slopes of −1.
48004-p2
8/3/2019 Fractual Boundary of Complex Network
http://slidepdf.com/reader/full/fractual-boundary-of-complex-network 5/8
Fractal boundaries of complex networks
6- 4202-4- 6 8
)> k< nl / N nl ( -l 01
5-
014-
01 3-
012-
011-
010
< B
> / N
002=N004=N008=N
0008=N00023=N
000821=N
01=N6
RE)a(
l
4321 5 6 7 3121110198
l 01
0
01
1
012
013
014
k +
000821=NRE
01=NRE6
000821=NFS
01=NFS6
)b(
l
1
~
Fig. 3: (Color online) (a) Normalized average number of nodesat shell ℓ, Bℓ/N , as a function of ℓ− lnN/ln k for ERnetwork with k= 6. For different N , the curves collapse.(b) kℓ + 1, which is k2ℓ /kℓ, as a function of ℓ shown for bothER and SF networks with different N .
which is similar with ER. We find similar results also for
λ > 3 (not shown).To test how general is our finding, we also study
several real networks (figs. 2c, d), including the HighEnergy Particle (HEP) physics citations network [12] andthe Autonomous System (AS) Internet network [13,14].For HEP network and AS network, d≈ 4.2 and 3.3,respectively. The degree distribution of HEP network isnot a power law (see fig. 2c), while the AS network showsa power law degree distribution with λ≈ 2.1 (see fig. 2d).Our results suggest that the power law decay behaviorappears also in both networks, with similar values of β ≈ 1for ℓ > d (see footnote 3).
Next we ask: how many nodes are on average atthe boundaries? Are they a nonzero fraction of N ?We calculate the mean number Bℓ in shell ℓ, and infig. 3a plot Bℓ/N as a function of ℓ− ln N/ln k fordifferent values of N for ER network. The term ln N/ln krepresents the diameter d of the network [2]. We find that,for different values of N , the curves collapse, supportinga relation independent of network size N . Since Bℓ/N isapparently constant and independent of N , it follows thatBℓ ∼N , i.e., a finite fraction of N nodes appears at eachshell including shells with ℓ > d. We find similar behaviorfor SF network with λ = 3.5 (not shown).
3We also find similar results (not shown here) for other realnetworks.
010
012
014
016
s
010
012
014
016
018
n
( s )
5=6=7=
FS)a(
0.3=θ
l
ll
l
l
010
011
012
013
014
s
010
012
014
016
018
n ( s )
5=7=8=0.3=θ
PEH)b(
lll
l
l
Fig. 4: (Color online) The number of clusters of sizes sℓ, n(sℓ),as a function of sℓ after removing nodes within shell ℓ for:(a) SF network with N = 106 and λ= 2.5, (b) HEP citationsnetwork. The relation between n(sℓ) and sℓ is characterizedby a power law, n(sℓ)∼ s−θℓ , with θ≈ 3. In order to show allcurves clearly, vertical shifts are made. Note that the pointsin the tail of the distributions represent the rare occurences of large clusters which are formed by nodes outside shell ℓ− 1.
The branching factor [15] of the network is
k = k2/k− 1,
where the averages are calculated for the entire network.For ER network, k can be proved to be equivalent tok. Similarly, we define kℓ ≡ k
2ℓ /kℓ− 1, where the
averages are calculated only for nodes in shell ℓ. Abovethe diameter, kℓ + 1 decreases with ℓ for both ER andSF networks (fig. 3b). Thus, at the shells where powerlaw behavior of P (Bℓ) appears (fig. 2), the nodes havemuch lower kℓ + 1 compared with the entire network.
The approaching of kℓ +1 to 1 (ER network) and 2 (SFnetwork) is consistent with a critical behavior at theboundaries of the network [15].
Next, we study the structural properties of the bound-aries. Removing all nodes that are within a distance ℓ (notincluding shell ℓ) for ℓ > d, the network will become frag-mented into several clusters (see fig. 1). We denote thesize of those clusters as sℓ, the number of clusters of sizesℓ as n(sℓ), and the diameter of the cluster as dℓ. We findn(s)∼ s−θ, with θ ≈ 3.0 (figs. 4a and b). The points inthe tails of figs. 4a and b represent the rare appearancesof the large clusters. We find similar results for ER andother real networks.
The relation between the sizes of the clusters sℓ andtheir diameters dℓ is shown as scatter plots in figs. 5a
48004-p3
8/3/2019 Fractual Boundary of Complex Network
http://slidepdf.com/reader/full/fractual-boundary-of-complex-network 6/8
Jia Shao et al.
011 d
l
010
012
014
S l
5=6=
0.2=ϕ
PEH)b(
l l
001011 d
l
01
0
011
012
013
014
S l
S8
dsv8
3.6=.tsiD.evA9>k
S8
dsv8
1.7=.tsiD.evA4<k
RE)c(000,001=N 6=>k <
8.1=ϕ
001011 d
010
011
012
013
014
S
6=7=
9.1=ϕ
FS)a(
l
l l
l
Fig. 5: (Color online) The size of clusters, sℓ, shown as scatterplots of the diameters dℓ of the clusters for (a) SF networkwith N = 106 and λ= 2.5, (b) HEP citations network for nodesoutside shell ℓ− 1. Vertical shifts of the curves are made forclarity. sℓ scales with dℓ as sℓ ∼ dϕℓ , with ϕ≈ 2. (c) For ERnetwork with N = 105 and k= 6, sℓ as a function of dℓ forsmall degree (k < 4) and large degree (k > 9) nodes chosen asroots. Here, sℓ as a function of dℓ is shown for ℓ=8. Thediameter of the entire network is d≈ 6.6. Depending on thedegree of the root node, the average distance of all nodes fromthe root may change. Large degree roots (k > 9) have smallaverage distances (≈ 6.3) while small degree roots (k < 4) havelarge average distances (≈ 7.1). However, sℓ ∼ dϕℓ , with ϕ≈ 2,can be observed for both small and large degree roots. Weignore the large clusters which appear in the flat regions of fig. 4.
and b, for SF (λ = 2.5) and HEP citations networks,respectively. In order to show all curves, vertical shifts aremade. Figures 5a and b show a power law relation, sℓ∼ dϕℓ ,with ϕ≈ 2, suggesting that the clusters at the boundariesare fractals with fractal dimension df = 2 like percolationclusters at criticality [16,17]. Here, we ignore the non-fractal large clusters which appear in the flat regions of fig. 4. We find that the fractal dimension is df = ϕ≈ 2 also
for ER, SF with λ = 3.5 and several other real networks.Root nodes with different degree yield different averagedistances of the rest of the nodes from the root [18,19].However, using our definition of boundaries the fractalclusters can be observed for both large and small degree
roots (see fig. 5c).Next we present analytical derivations supporting the
above numerical results. We denote the degree distributionof a network as q(k). For infinitely large networks we canneglect loops for ℓ < d and approximate the forming of anetwork as a branching process [20–23]. The probabilityof reaching a node with k outgoing links through arandomly selected link is q(k) = (k + 1)q(k + 1)/k. Wedefine G0(x)≡
∞k=0 q(k)xk as the generating function of
q(k), G1(x) =∞
k=0 q(k)xk = G′0(x)/k as the generating
function of q(k). For ER networks we have G0(x) =G1(x) = ek(x−1) and k = k. The generating function for
the number of nodes, Bm, at the shell m is [23]Gm(x) = G0(G1(. . .(G1(x)))) = G0(Gm−1
1 (x)), (1)
where G1(G1(. . .))≡Gm−11 (x) is the result of applying
G1(x), m− 1 times. P (Bm), which is the probabilitydistribution of Bm, is the coefficient of xBm in the Taylorexpansion of Gm(x).
For shells with large m which is still smaller than d,it is expected [23] that the number of nodes will increaseby a factor of k. It is possible to show [21] that Gm−1
1 (x)converges to a function of the form f ((1−x)km) for largem (m≪ d), where f (x) satisfies the Poincare functionalrelation
G1(f (y)) = f (yk), (2)
where y = 1−x. The function form of f (y) can be uniquelydetermined from eq. (2).
It is known [21] that f (x) has an asymptotic func-tional form, f (y) = f ∞ + ay−δ + 0(yδ), where f ∞ satisfiesG1(f ∞) = f ∞. It can be shown [22] that f ∞ also givesthe probability that a link is not connected to the giantcomponent of the network by one of its ends. Expandingboth sides of eq. (2), we obtain
G1(f ∞) + G′1(f ∞)ay−δ = f ∞ + ak−δy−δ + 0(yδ). (3)
Since G1(f ∞) = f ∞, we have δ =− ln G′1(f ∞)/ ln k.
If q(1)=0 and q(2) = 0, from G1(f ∞) = f ∞, we havef ∞ = 0 a n d G′
1(f ∞) = G0′′(0)/k= 2q(2)/k. If q(2)=
q(1)=0 (Bottcher case [21]), then δ =∞, which indi-cates that f (y) has an exponential singularity. Therefore,networks with minimum degree kmin 3 do not have thepower law distribution of Bℓ shown in fig. 2, and thereforehave no fractal boundaries.
Applying Tauberian-like theorems [21,24] to f (y), whichhas a power law behavior for y →∞, Dubuc [25] concludedthat the Taylor expansion coefficient of Gm(x), P (Bm),behaves as Bμ
m with an exponential cutoff at B∗m ∼ km,
where
μ =
δ− 1, q(1) = 0 a n d q(2) = 0;
2δ− 1, q(1) = 0 and q(2) = 0.
48004-p4
8/3/2019 Fractual Boundary of Complex Network
http://slidepdf.com/reader/full/fractual-boundary-of-complex-network 7/8
Fractal boundaries of complex networks
010
012
014
016
B
018-
016-
014-
012-
010
P
D F
1=2=3=4=5=6=7=8=
1.04.1=µ
)d<(RE)a( l
lll
ll
l
ll
l
_
-+
4.02.00 6.0 18.0 r
013-
012-
011-
010
r)6(.qE
RE)b(
1=m
2=m
3=m
l
l + m
Fig. 6: (Color online) (a) For ER network, the probabilitydistribution function P (Bℓ) of number of nodes Bℓ in shellsℓ d. For small values of Bℓ, P (Bℓ)∼Bμ
ℓ , where μ depends onthe k of the network (eq. (4)). The slopes of the least-squarefit represented by the straight lines give μ= 1.4± 0.1, which isin good agreement with the theoretically predicted value μ=1.34. (b) The fraction of nodes outside shell ℓ+m− 1, rℓ+m, asa function of rℓ for ER network, where rℓ = 1− (
ℓ−1i=1 Bi)/N
is calculated for any possible ℓ. The (red) lines represent the
theoretical iteration function (eq. (6)).
Thus the probability distribution function of the numberof nodes in the shell m with m≪ d has a power law tailfor small values of Bm
P (Bm)∼Bμm. (4)
For an ER network, eq. (4) is supported by simulationsfor m d in fig. 6a. Figure 6a shows for ER networkthat P (Bℓ) for ℓ < d and small values of Bℓ increaseas a power law, P (Bℓ)∼Bμ
ℓ . For ER network, we have
k = k, μ = δ− 1, and δ =− ln G′
1
(f ∞)/ ln k. Thus μ =−ln (kf ∞)/ln k− 1, where f ∞ can be obtained numeri-cally from f ∞ = ek(f ∞−1). In the case of k= 6, μ≈ 1.34,which is close to the result shown in fig. 6a.
The above considerations are correct only for m < d,for which the depletion of nodes with large degree in thenetwork is insignificant. In a large network, the shells withm≫ 1 behave almost deterministically, and one can applythe mean-field approximation for the number of nodes andlinks in each shell. Writing down the master equation forthe degree distribution in the outer shells, one can obtaina system of ordinary differential equations, which canbe solved analytically using the apparatus of generating
functions. Using this solution one can show that
rn = G0(Gn−m1 (G−1
0 (rm))), (5)
where rn = 1− (n−1
i=1 Bi)/N is the fraction of nodesoutside shell n− 1. Note that eq. (5) has almost thesame structure as eq. (1). It can also be shown that thebranching factor of nodes outside shell n− 1 is k(rn) =uG′′
0(u)/G′0(u), where u = G−1
0 (rn).
For ER networks, eq. (5) yields
rℓ+1 = ek(rℓ−1) =
∞ℓ=0
q(k)rkℓ , (6)
which is valid for all possible ℓ. We test it in fig. 6b forER network. The relation between rℓ+m and rℓ can beobtained by applying eq. (6) m times on rℓ. In fig. 6b weshow the fraction of nodes outside shell ℓ + m− 1, rℓ+m,as a function of rℓ for ER network. Different values of mare tested in the plot.
When m≪ d and n≫ d, using the same considerations
as we used in eq. (1), one can show that
rn = [ak(1− rm)]−μ−1 + r∞, (7)
where r∞ = G0(f ∞) is the fraction of nodes not belongingto the giant component of the network, a is a constant.
Based on eqs. (4) and (7), expressing rm and rn interms of Bm and Bn, we find that for m≪ d and n≫ d,Bn ∼B−μ−1
m . Using P (Bn)dBn = P (Bm)dBm, we obtain
P (Bn)∼B−1−μ/(μ+1)−1/(μ+1)n = B−2
n , (8)
supporting the numerical findings in fig. 2.
These results are rigorous when˜k exists and when theminimum degree kmin 2. For SF networks with λ < 3, k
diverges for N →∞. But for finite N , k still exists. Thusthe above results can also be applied to the case of λ < 3.For both ER and SF networks with kmin 3, the powerlaw of P (Bn) with n≫ d cannot be observed, as we indeedconfirm by simulations.
Relating our problem to percolation theory, we canexplain the simulation results of the probability distri-bution of cluster size sℓ. The cluster size distributionin percolation at some concentration p close to pc isdetermined by the formula [15]
P p(s > S )∼ S −τ +1 exp(−S | p− pc|1/σ). (9)
In the case of random networks the percolation thresholdis given by pc = 1/k. In the exterior of the shell n− 1(n≫ d), we can estimate | p− pc| ∼ (k(rn)− 1)/k, wherek(rn) decreases and reaches the critical percolation valueof 1. Near the percolation threshold the nodes outsideshell n− 1 are split into a number of finite clusters, and if k > 1 a giant component. These finite clusters have fractaldimension df = 2 [16,17]. This theoretical prediction isconfirmed in fig. 5.
The cluster size distribution can be estimated byintroducing a sharp exponential cutoff at s = S ∗
n
∼
|k(rn)− 1|−1
σ , so that P n(s > S )∼S −τ +1P (S ∗n > S ),where P (S ∗n > S ) is the probability for a given shell to have
48004-p5
8/3/2019 Fractual Boundary of Complex Network
http://slidepdf.com/reader/full/fractual-boundary-of-complex-network 8/8
Jia Shao et al.
S ∗n > S . Since rn− r∞ has a smooth power law distributionand k(r∞) < 1, the probability that |k(rn)− 1|< S −σ = εis proportional to ε. Thus P (S ∗n > S )∼S −σ andP n(s > S ) = S −τ +1−σ [26]. Therefore the cluster sizedistribution follows n(s)∼ s−(τ +σ) = s−θ, thus θ = τ + σ.
For ER networks and SF networks with λ > 4, τ = 2.5and σ = 0.5, the above derivations lead to n(s)∼ s−3.For SF networks with 2 < λ < 4, τ = (2λ− 3)/(λ− 2) andσ = |λ− 3|/(λ− 2) [16]. Thus, for SF network with λ > 3,there will be n(s)∼ s−3. We conjecture n(s)∼ s−3 evenfor 2 < λ < 3, although in this case k(rn) does not existand the above derivations are not valid. Our numericalsimulations support these results in fig. 4.
In summary, we find empirically and analytically thatthe boundaries of a broad class of complex networksincluding non-fractal networks [9] have fractal features.The probability distribution function of the number of
nodes in these shells follows a power law, P (Bℓ)∼B−2ℓ ,
and the number of clusters of size sℓ, n(sℓ), scales asn(sℓ)∼ s−3ℓ . The clusters at the boundaries are fractalswith a fractal dimension df ≈ 2. Our findings can beapplied to the study of epidemics. They imply that astrong decay of the epidemic will happen in the boundariesof human network, due to the low degree of nodes.
∗ ∗ ∗
We thank ONR and Israel Science Foundation forfinancial support. SVB thanks the Office of the Academic
Affairs of Yeshiva University for funding the YeshivaUniversity high-performance computer cluster andacknowledges the partial support of this research throughthe Dr. Bernard W. Gamson Computational ScienceCenter at Yeshiva College.
REFERENCES
[1] Erdos P. and Renyi A., Publ. Math., 6 (1959) 290; Publ.Math. Inst. Hung. Acad. Sci., 5 (1960) 17.
[2] Bollobas B., Random Graphs (Academic, London) 1985.
[3] Albert R. and Barabasi A.-L., Rev. Mod. Phys., 74(2002) 47.
[4] Milgram S., Psychol. Today , 2 (1967) 60.[5] Watts D. J. and Strogatz S. H., Nature (London), 393
(1998) 440.
[6] Cohen R. and Havlin S., Phys. Rev. Lett., 90 (2003)058701.
[7] Dorogovtsev S. N. and Mendes J. F. F., Evolution of Networks: from Biological Nets to the Internet and WWW (Oxford University Press, New York) 2003.
[8] Pastor-Satorras R. and Vespignani A., Evolution and Structure of the Internet: Statistical Physics Approach (Cambridge University Press) 2004.
[9] Song C. et al., Nature (London), 433 (2005) 392; Nat.Phy., 2 (2006) 275.
[10] Goh K. I. et al., Phys. Rev. Lett., 96 (2006) 018701.[11] Kitsak M. et al., Phys. Rev. E., 75 (2007) 056115.[12] Derived from the HEP section of arxiv.org; http://
vlado.fmf.uni-lj.si/pub/networks/data/hep-th/hep-
th.htm (website of Pajek).[13] Carmi S. et al., Proc. Natl. Acad. Sci. U.S.A., 104 (2007)
11150.[14] Shavitt Y. and Shir E., DIMES - Letting the Inter-
net Measure Itself , http://www.arxiv.org/abs/cs.NI/
0506099.[15] Cohen R. et al., Phys. Rev. Lett., 85 (2000) 4626.[16] Cohen R. et al., Phys. Rev. E , 66 (2002) 036113;
Bornholdt S. and Schuster H. G. (Editors), Handbook of Graphs and Networks (Wiley-VCH) 2002, Chapt. 4.
[17] Bunde A. and Havlin S., Fractals and Disordered System (Springer) 1996.
[18] Holyst J. et al., Phys. Rev. E , 72 (2005) 026108.
[19] Dorogovtsev S. N. et al., Phys. Rev. E ,73
(2006)056122.[20] Harris T. E., Ann. Math. Stat., 41 (1948) 474; The
Theory of Branching Processes (Springer-Verlag, Berlin)1963.
[21] Bingham N. H., J. Appl. Probab. A, 25 (1988) 215.[22] Braunstein L. A. et al., Int. J. Bifurcat. Chaos, 17
(2007) 2215; Phys. Rev. Lett., 91 (2003) 168701.[23] Newman M. E. J. et al., Phys. Rev. E , 64 (2001) 026118.[24] Weiss G. H., Aspects and Applications of the Random
Walk (North Holland Press, Amsterdam) 1994.[25] Dubuc S., Ann. Inst. Fourier , 21 (1971) 171.[26] Barabasi A. L. et al., Phys. Rev. Lett., 76 (1996) 2192.
48004-p6