Professional Documents
Culture Documents
The graphs derived from the U.S. power grid, the Hollywood graph of actors (where there is an edge
between two actors if they have appeared together in a movie), the food web (links for ecological
dynamics among diverse assemblages of species), cellular and metabolic networks, and various social networks all obey a power law. Thus, a power law fit for the degree distribution appears to be a
generic and robust property for many massive real-world graphs. Here we work with a model G(p)
where, in each step, with probability p we add a new vertex and an edge between it and an existing
vertex chosen following the preferential attachment rule, otherwise we choose two existing vertices
in the same way as before and we add an edge between them. We shall show that this model follows
a tunable power law as the time approaches infinity.
The model
(2 p)
m1,t = m1,t1 (1
2t
+ p,
thus,
(2 p)
E(m1,t) = E(m1,t1) (1
2t
+ p.
We proceed by induction on k to show that limt E(mk,t)/t has a limit Mk for each k. To
do so,
we usea result which asserts that for a sequence {at} satisfying the recursive relation
at+1 = 1 btt at + ct, then
at
c
lim
=
t t
1+b
provided that limt bt = b > 0 and limt ct = c. In the end of the induction, we get
The model we work is known as GLP (Generalized Linear Preference) [1]. We start with the graph
G0 formed by one vertex having one loop. We have two operations we can do on a graph:
Vertex-step Add a new vertex v, and add an edge {u, v} from v by randomly and independently
choosing u with probability proportional to the degree of u in the current graph.
Edge-step Add a new edge {r, s} by independently choosing vertices r and s with probability
proportional to their degrees.
The random graph model G(p) is defined as follows:
1. Begin with the initial graph G0.
2. For t > 0, at time t, the graph Gt is formed by modifying Gt1 as follows:
with probability p, take a vertex-step,
otherwise, take an edge-step.
2p
M1 =
4p
2
(k)
2
+
2p
2p
Mk =
4 p k + 1 + 2
2p
(1)
We wish to show that the graph G generated by G(p) is a power law graph with Mk k for large
k. If Mk k , then
k
Mk
=
=
Mk1 (k 1)
1
1
k
1
=1 +O
.
2
k
k
From 1 we have
2
1 + 2p
Mk
k1
=
=1
=1
2
2
Mk1 k + 2p
k + 2p
2
1 + 2p
1
+O
.
2
k
t
X
st
i=1
where
P (sj = 1) = p = 1 P (sj = 0).
Let mk,t denote the number of vertices of degree k at time t. We wish to derive the recurrence for
the expected value E(mk,t). Let Ft be the -algebra associated with the probability space at time t.
Thus, for t > 0 and k > 1, we have
(2 p)k
(2 p)(k 1)
E(mk,t|Ft1) = mk,t1 1
+ mk1,t1
.
2t
2t
If we take the expectation on both sides, we get the following recurrence formula
(2 p)k
(2 p)(k 1)
+ E(mk1,t1)
.
E(mk,t) = E(mk,t1) 1
2t
2t
References
[1] T. Bu and D. Towsley. On distinguishing between internet power law topology generators. Proc.
of IEEE / INFOCOM, December 2002.
[2] Fan Chung and Linyuan Lu. Complex Graphs and Networks (Cbms Regional Conference Series
in Mathematics). American Mathematical Society, Boston, MA, USA, 2006.