Extinction probability of binomial branching process tends to poisson one.

368 Views Asked by At

The folowing is stated and proved in the random graphs book by Luczak, Janson, Rucinski and this is on page 108 in the Giant component section.

I can't understand why the conclusion follows from the last statement in the proof.

Question: We consider a series of branching processes with immediate offspring distribution $X_n \sim \mathrm{Bin}(n,p)$ where $np \to c$ as $n\to \infty$ for some constant $c>1$.

We also consider a random variable $X \sim \mathrm{Poiss}(c)$.

If we denote by $\rho_{X_i}$ the extinction probability.

Show $\rho_{X_i} \to \rho_X$.

Solution: This is the solution given in my textbook, it seems wrong to me and I don't seem to be able to fix it, it is quite important to me as the remainder of the chapter is based on it.

We know that generating functions are $$f_{X_n}(x)=(1-p+xp)^n$$ and

$$f_X(x)=e^{c(x-1)}$$

It is now a standard result that for a fixed $x$ $f_{X_n}(x) \to f_X(x)$ so $f_n$ converge pointwise to $f$.

As $c<1$ we have $f_{X_n}(\rho_{X_n})=\rho_{X_n}$ and $f_{X}(\rho_{X})=\rho_{X}$

Now the book just states that this implies $\rho_{X_n} \to \rho_X$. And I just don't see why.