Random methods: 4 Flavors
Pure Random: use fixed probability p. Simplicity is a plus.
Waxman: P(u,v) = ae-d/ bL, where 0 < a,b < 1
- Increase a => increase number of nodes in graph
- Increase b => increase ration of long to short edges
Exponential: P(u,v) = ae-d/(L- d)
- Tries to capture locality properties
- Relates edge probability to distance
- P(u,v) goes to 0 and distance between u and v approaches L
Locality:
- Partitions potential edges based on lengths
- P(u,v) = a, if d < r, and P(u,v) = b, if d >= r
1. Every pair of nodes is treated equally
2. No control on edge configuration
3. Problem when generating large sparse graphs