From a follow-up read from "I'll have what she's having", a paper by Bently called "Evolving social influence in large populations".
He presents a number of applications of preferential attachment in sociology. He then presents a model which is essentiallly a Dirichlet process/Chinese restaurant scenario, with the twist that:
a fixed number of people enter at each step (not just one)
Table choice is based on people who have sat at that table in the last m rounds; thus the process has limited memory (and may no longer be exchangeable!!).
He makes analogy to the neurtral drift aspect of evolution, and comments that evolutionary models are inherently dynamic.
His model:
N initial population
n enter the restaurant at each time-step
mu agent chooses a new table
1-mu agent chooses an existing table, with probability proportional to # at the table
So, can I tie this idea into the quasispecies as memory concept?
No comments:
Post a Comment