Download Chapters 3-6 (.PDF) - ODBMS
Download Chapters 3-6 (.PDF) - ODBMS
Download Chapters 3-6 (.PDF) - ODBMS
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
30 4. PATTERNS IN WEIGHTED GRAPHS<br />
4.3 DW-2: LWPL: WEIGHTED PRINCIPAL EIGENVALUE<br />
OVER TIME<br />
Given that unweighted (0-1) graphs follow the λ1 Power Law (LPL, pattern D-5), one may ask if<br />
there is a corresponding law for weighted graphs.The answer is ’yes:’ let λ1,w be the largest eigenvalue<br />
of the weighted adjacency matrix Aw, where the entries wi,j of Aw represent the actual edge weight<br />
between node i and j (e.g., count of phone-calls from i to j). Notice that λ1,w increases with the<br />
number of edges, following a power law with a higher exponent than that of its λ1 Power Law (see<br />
Fig. 4.3).<br />
Observation 4.3 λ1,w Power Law (LWPL) Weighted real graphs exhibit a power law for the<br />
largest eigenvalue of the weighted adjacency matrix λ1,w(t) and the number of edges E(t) over time.<br />
That is,<br />
λ1,w(t) ∝ E(t) β<br />
In the experiments in [200, 202], the exponent β ranged from 0.5 to 1.6.<br />
λ 1, W<br />
10 10<br />
10 8<br />
10 6<br />
10 4<br />
10 2<br />
10 0<br />
10 1<br />
0.91595x + (2.1645) = y<br />
10 2<br />
10 3<br />
|E|<br />
10 4<br />
10 5<br />
10 6<br />
λ 1, W<br />
10 4<br />
10 3<br />
10 2<br />
10 1<br />
10 0<br />
10 0<br />
10 -1<br />
0.69559x + (−0.40371) = y<br />
10 1<br />
10 2<br />
10 3<br />
|E|<br />
10 4<br />
10 5<br />
10 6<br />
λ 1,w<br />
3<br />
10<br />
10 2<br />
10 1<br />
10 1<br />
10 0<br />
0.58764x + (-0.2863) = y<br />
(a) CampIndiv (b) BlogNet (c) Auth-Conf<br />
Figure 4.3: Illustration of the LWPL. 1 st eigenvalue λ1,w(t) of the weighted adjacency matrix Aw versus<br />
number of edges E(t) over time. The vertical lines indicate the gelling point.<br />
10 2<br />
3<br />
10<br />
|E|<br />
10 4<br />
10 5