Unformatted text preview: ﬀects.
Consider a discrete-time model and let F (t ) be the fraction of agents
infected at time t .
The Bass model is described by the diﬀerence equation: F (t ) = F (t − 1) + p (1 − F (t − 1)) + q (1 − F (t − 1))F (t − 1). The term p (1 − F (t − 1)) is the infection rate times the fraction of
uninfected agents. The term q (1 − F (t − 1))F (t − 1) is the contagion rate
times the frequency of encounters between healthy and infected agents.
4 Networks: Lecture 8 Bass Model–2
A continuous time version of this model is described by
d F (t )
= (p + qF (t ))(1 − F (t )),
with F (0) = 0.
This is a nonlinear diﬀerential equation, but admits a closed form solution
F (t ) = 1 − e − (p +q )t
1 + q e − (p +q )t
p Note that the levels of p and q scale time, the ratio of q to p determines the
overall shape of the curve. 5 Networks: Lecture 8 Bass model–3 Figure: Diﬀusion curves: left is for p < q and right is for p > q .
Many empirical studies have found diﬀusion patterns that are S-shaped (e.g.
adoption of hybrid corn seeds among Iowa farmers).
Let us interpret this in the context of adoption of new technologies.
First adopters are almost entirely those who adopt from their spontaneous
innovation (when F (t ) is close to 0, F (t ) = p ).
As process progresses, there are more agents to be imitated leading to an
increase in the rate of diﬀusion, which eventually slows down since there are
fewer agents to do the imitating.
6 Networks: Lecture 8 Diﬀusion in a Network with Immune Nodes The problems of modeling contagion or the spread of information through a
society involve determining “when paths exist that connect diﬀerent nodes”,
i.e., understanding the component structure.
Let us consider the following problem.
There is a society of n individuals. Initially one of them is infected with a
disease. Each individual is immune with probability π .
The question of whether the disease can spread to a nontrivial fraction of
the population amounts to whether the infected...
View Full Document
- Fall '09