Question:
Suppose that $N\in\mathbb{N}$ immigrants (nodes) are connected by an Erdős–Rényi random graph and that the average degree is $\lambda$. The graph is fixed at the beginning of time (before anyone has immigrated).
One by one, all $N$ immigrants move to a new country. As they move to the country, they make native acquaintances according to the following timeline (the acquaintances are not incorporated into the graph):
- A new immigrant arrives.
- The immigrant randomly meets one native, who becomes an acquaintance.
- The immigrant talks to all the immigrants in his neighborhood who have already arrived in the country. All of their native acquaintances become his acquaintances (but not vice-versa).
- The next immigrant arrives.
Assume that the native population is large so that the chance that two immigrants randomly meet the same native is 0.
As $N\to \infty$, what is the distribution of acquaintances over immigrants, $\{A_n\}_{n\in\mathbb{N}}$?
Notes:
- An immigrant will only acquire acquaintances between when he arrives and the next immigrant arrives.
- The first immigrant will always only have a single acquaintance.
- Any immigrant who is not connected will always only have a single acquaintance.
- An immigrant connected to $k$ other immigrants who have already immigrated must have at least $k+1$ acquaintances.
- If the network has no edges ($\lambda=0$), then all immigrants will only have a single friend.
- Since this is a limit, the distribution should be supported over the natural numbers: $A_n>0$ for all $n \in \mathbb{N}$, as long as $\lambda>0 $
- A well-known property of Erdős–Rényi graphs, is that the degree distribution, $\{d_n\}_{n\in\mathbb{N}}$, converges to a Poisson distribution as the number of nodes, $N$, becomes large.
Progress:
So far, I have been thinking about this a series of differential equations $\{A_n(t)\}_{n\in\mathbb{N} \cup \{0\}}$. Then $\alpha_n(t)$ is the share of immigrants with $n$ acquaintances after fraction $t$ of the immigrants have moved to the new country. Then $\alpha_n(1)=A_n $. Here's what I know so far:
- $\alpha_0=1-t$ is simply the share who have not moved.
- $\dot{\alpha}_1=\sum_{n\in \mathbb{N}}d_n\cdot (1-t)^n$. I only have a single acquaintance if none of my neighborhood has immigrated.
- Things get quite complicated after that.