DocumentCode :
2852
Title :
Scaling Laws for Connectivity in Random Threshold Graph Models with Non-Negative Fitness Variables
Author :
Makowski, A.M. ; Yagan, O.
Author_Institution :
Dept. of Electr. & Comput. Eng., Univ. of Maryland, College Park, MD, USA
Volume :
31
Issue :
9
fYear :
2013
fDate :
Sep-13
Firstpage :
573
Lastpage :
583
Abstract :
We explore the scaling properties for graph connectivity in random threshold graphs. In the many node limit, we provide a complete characterization for the existence and type of the underlying zero-one laws, and identify the corresponding critical scalings. These results are consequences of well-known facts in Extreme Value Theory concerning the asymptotic behavior of running maxima on i.i.d. random variables. In the important special case of exponentially distributed fitness, we show that the (essentially unique) critical scaling which ensures a power-law degree distribution, does not result in graph connectivity in the asymptotically almost sure (a.a.s.) sense.
Keywords :
graph theory; random processes; a.a.s; asymptotic behavior; asymptotically almost sure sense; extreme value theory; i.i.d. random variable; nonnegative fitness variable; power-law degree distribution; random threshold graph model; running maxima behavior; scaling law property; zero-one law; Context; Convergence; Diseases; Distribution functions; Exponential distribution; Manganese; Connectivity; Extreme Value Theory; Hidden variables; Random threshold graphs; Scale-free networks; Zero-one laws;
fLanguage :
English
Journal_Title :
Selected Areas in Communications, IEEE Journal on
Publisher :
ieee
ISSN :
0733-8716
Type :
jour
DOI :
10.1109/JSAC.2013.SUP.0513050
Filename :
6544546
Link To Document :
بازگشت