Information Technology Reference
In-Depth Information
Table 3.3
Simulation environment
Item
Components and performance
CPU
Intel Pentium D (2MB
2 L2 Cache, 3.20 GHz, 800 MHz FSB)
Memory
3 GB DDR (DDR2-SDRAM, Dual Channel)
OS
Linux 2.6.33.3-85.fc13.c86 64
Compiler
gcc 4.4.4
Option
Default optimization
3.6 Simulation
In this section, we will investigate the information diffusion on scale-free networks
that often appear in our realistic social networks. According to our model of
information diffusion, we can see the simulation of phenomena using computa-
tional calculations.
3.6.1 Settings
In our simulation, the hardware comprised an Intel Pentium D with 64 bit mode and
3 GB DDR. Our simulation program was a native C program, which was compiled
by gcc 4.4.4 on Fedora 13 (Linux 2.6.33.3-85.fc13.c86 64) with default optimiza-
tion. See Table
3.3
for components in detail.
At the beginning of the simulation, the program constructs a scale-free network
according to the method of preference selection proposed by Dorogovtsev et al.
(
2000
). Figure
3.2
shows the degree distribution of the generated network. In the
figure, the frequency of each degree, which is defined by the number of links of a
node, are plotted with log-log axis. It is easy to see from the figure that the degree
distribution forms like
Ck
g
;
p
ð
k
Þ¼
(3.17)
where
k
is the degree of each node,
g
is the scale factor and
C
is a positive constant.
We have obtained
g ¼
1
.
77 and
C
¼
333 by applying regression with determina-
tion coefficient
R
2
¼
0
.
904.
In the simulation,
U
¼f
u
1
;
u
2
; ...;
u
1024
g
(3.18)
is assumed because of the memory limitation of two-dimensional array alignments.
A part of the adjacency matrix is expressed by