MATH6142 Complex Networks
Hello, dear friend, you can consult us at any time if you have any questions, add WeChat: daixieit
MATH6142 Complex Networks
Exam 2016
● Solution Problem 1
a) The network is directed because the adjacency matrix is not sym- metric. (2 marks)[Unseen]
b) The network is shown in Figure 1. (4 marks)[Unseen]
Figure 1: The directed network with adjacency matrix A.
c) The network contains two weakly connected components formed re- spectivelly by the nodes {1, 2} and {3, 4}. (2 marks)[Unseen]
d) The network contains one strongly connected component formed by the nodes {3, 4}. (2 marks)[Unseen]
e) The network does not have any out-component. (2 marks)[Unseen]
f) The in-degree sequence is fiven by {1, 0, 1, 1} the out-degree sequence is given by {0, 1, 1, 1}.
(4 marks) [Unseen]
g) The in-degree distribution is given by Pin (0) = 1/4, Pin (1) = 3/4Pin (k) =
0 for k = 2, 3.
(2 marks) [Unseen]
The out-degree distribution is given by Pout (0) = 1/4, Pout (1) = 3/4Pout (k) = 0 for k = 2, 3.
(2 mark2) [Unseen]
h) The eigenvector centrality x can be found as following.
Given the initial guess x(0) = 1 given by
╱ 1 、
x(0) = !(!) 1(1) │(│)
. 1 !
let us is calculate the result of the iteration
xn = Axn _ 1 .
We have
(1)
(2)
╱ 0 x(1) = !(!) 0(0) . 0
╱ 0
x(2) = !(!) 0
.
╱ 0
x(3) = !(!) 0 . 0
Therefore
for every n > 3.
It follows that
1 0 0 0
1 0 0 0
1 0 0 0 |
0 0 0 1
0 0 0 1
0 0 0 1 |
0(0) 、│ 1 ╱! 1(1) 、│ 1 ╱! 0(1) 、│ 0(1) │! 4 !. 1(1) │! = 4 !. 1(1) │! . (2 marks)[Unseen]
0(0) 、│ 1 ╱! 0(1) 、│ 1 ╱! 0(0) 、│ 1 │! 4 !. 1 │! = 4 !. 1 │! . (2 marks)[Unseen]
0(0) 、│ 1 ╱! 0(0) 、│ 1 ╱! 0(0) 、│ 0(1) │! 4 !. 1(1) │! = 4 !. 1(1) │! . |
╱ 0 、
x(n) = !(!) 1(0) │(│)
(2 marks)[Unseen]
╱ 0 、
x = !(!) 1(0) │(│)
(2 marks)[Unseen]
i) The result of point h) is expected because nodes 1 and 2 are in an in-component of the network and therefore have zero eigenvector centrality. Nodes 3 and node 4 instead have the same eigevcetor centrality because they form a dimer. (2 marks)[Unseen]
– Solution Problem 2
a) The undirected network of N nodes with smallest diameter is the complete graph, formed by N nodes each one connected to all the others. The diameter of this network is D = 1. (2 marks)[Unseen] This network has the small-world distance property. In fact
lN = lN = 0. (5)
(3 marks)[Bookwork]
b) The undirected network of N nodes which is connected and has the largest diameter is the (open) linear chain of N nodes that has di- ameter D = N · 1. (2 marks)[Unseen] This network does not have the small-world distance property. In fact
lN = lN = o. (6)
(3 marks)[Bookwork]
c) i) The degree distribution P(k) of the Poisson network with average degree (k)= c = 4 is given by
P (k) = ck e_c . (7)
(1 marks)[Bookwork] The generating function G(x) is given by
→
G(x) = P (k)xk = e_c+cx . (8)
k=0
(2 marks)[Bookwork] The first and second moment of the Poisson degree distribution can be obtained by differentiating the generating function G(x). In particular we have
(k) = │x=1 = c (k(k · 1)) = │x=1 = c2 .
(9)
(2 marks)[Bookwork]
It follows that
(b)= = c = 4. (10)
(1 marks)[Bookwork]
ii) Using the result of point i) we get
Nd ~ (k) ╱ 、d _ 1 = (k)d (11)
(2 marks)[Bookwork]
Therefore
Nd ~ 4d . (12)
(1 marks)[Unseen]
iii) Since at distance d = 0 from the random node there is only one node (the node itself) we have
Nd ~ 4d (13)
for 0 < d. (1 marks)[Unseen]
The total number of nodes Nd<e at distance 0 < d < 1 from any random node of the Poisson network is given by
e 4e+1 · 1
Nd<e = 4d = 4 · 1 . (14)
d=0
(2 marks)[Unseen]
Therefore we have
N = ┌4e+1 · 1┐ (15)
(1 marks)[Unseen]
iv) By imposing that all the nodes of the network are found at dis- tance d < 1 from a random node we have
N = Nd<e = ┌4e+1 · 1┐ (1marks)
3N = 4e+1 · 1(1mark)
3N + 1 = 4e+1(1mark)
1 = ln [3N + 1] · 1.(1mark) (16)
v) Using Eq. (16) and considering the leading term for N > 1 we have
ln[3N] ln N
2 ln 2 2 ln 2 .
(4 marks) [Unseen]
vi) The local clustering coefficient Ci of a generic node i of a Poisson network can be estimated to be
Number of triangles passing through node i
Ci = ki (ki · 1)/2
~
pki (ki · 1)/2
ki (ki · 1)/2
(18)
where ki is the degree of node i and p = (k)/N is the probability of a link between any two nodes of the network. (2 marks) [Bookwork]
In fact each node i belongs in average to pki (ki · 1)/2 triangles since any pair of its neighbor is connected with probability p. (2 marks) [Bookwork]
● Solution Problem 3
a) The average number of links i added to node i in timestep t is given by
N N
i = π (i,r) = .
r=1 r=1
(19)
(2 marks) [Unseen]
Using
N
L = kj
j=1
N
ki = Air
r=1
we get
(20)
i =
(21)
(4 marks) [Unseen]
b)Since initially the number of links is m0 = 1 and at each time we add
two links we have L = 1 + 2t. (1 mark) [Unseen]
Since initially we have n0 = 2 nodes and at each time we add a node we
have N = 2 + t.
c)The average degree (k)is given by
= =
(1 mark) [Unseen]
(22)
(4 marks) [Bookwork]
d)In the mean-field approximation, the degree ki (t) of node i at time t satisfies the following differential equation
= i =
(23)
(2 marks)[Bookwork] In the limit t > 1, we have(j kj = 2L ~ 4t. Therefore we can write the dynamical mean-field equation for the degree ki (t) of node i, getting
dki 2ki ki
= =
dt 4t 2t ,
with initial condition ki (ti ) = 2.
This equation has solution
ki (t) = 2 ╱ 、1/2 .
(24)
(2 marks)[Bookwork]
(25)
(2 marks)[Bookwork]
e) The probability P(ki (t) > k) that a random node has degree ki (t) > k , in the mean-field approximation can be calculated as follows
P (ki (t) > k) = P ╱2 ╱ 、 (1/2) > k\ = P ╱ti < t ╱ 、 2 \ = ╱ 、 2 . (26)
(3 marks)[Bookwork] Therefore, the degree distribution P (k) is given by
P (k) = = · ╱ 、 2 = ╱ 、 3 . (27)
(3 marks)[Bookwork]
f) The master equation for Nk (t) reads
Nk (t + 1) = Nk (t) + Nk _ 1 (t)(1 · δk,2) · Nk (t) + δk,2
(28)
(3 marks)[Bookwork]
g) Assuming Nk (t) ~ (t + n0 )P (k) for large t
1 we get
P (k) = P (k · 1)(1 · δk,2) · P (k) + δk,2 .
(29)
giving
P (k) = P (k · 1)
for k > 2 and
P (2) = 1
Solving these equations we get
(3 marks)[Bookwork]
(30)
(31)
(3 marks)[Bookwork]
P (k) =
12
k(k + 1)(k + 2) .
(32)
(2 marks)[Bookwork]
2022-05-16