A normal branching process $Z_n$ initialized with $Z_0=1$ and offspring generated from $Pois(p),p<1,$ has a total progeny / total off spring distribution

$$X=\sum_{n=0}^\infty Z_n$$ $X\in \mathbb{N}$ almost surely, because $Z_n$ goes extinct eventually when $p<1$. $X$ follows the Borel Distribution. In particular $\mathbb{E}\exp(cX)<\infty$ for small $c$.

Question: does the total offspring $X$ also satisfy $\mathbb{E}\exp(cX)<\infty$ for a multitype branching process with poisson offspring?

Elaboration: The relevant multitype branching process can be described as a Markov chain $Z_n\in \mathbb{N}^m,$ with a $m\times m$ matrix $A$ as parameter. The distribution $Z^{i}_n\vert Z_n=v$ is equal in law to

$\sum_{j=1}^m\sum_{k=1}^{v_k} Pois(a_{ij})$

where all poisson summands are independent.

Again, this process goes extinct when $A$ has spectral radius strictly smaller than 1.

I am interested whether the variable

$$X=\sum_{n=0}^\infty\sum_{j=1}^m Z^{j}_n$$

satisfies $\mathbb{E}\exp(cX)<\infty$ for small $c$.


Yes, with essentially the same proof. In the scalar case, just notice that $e^{cX_n+bZ_n}$ is a supermartingale as long as $p(e^{c+b}-1)\le b$. In the vector case you will need to find positive vectors $c,b\in(0,\infty)^m$ with the property $(e^{c+b}-1)A\le b$ (I use the notation $f(c)=(f(c_1),\dots,f(c_m))$ and compare the vectors entry-wise), which is possible if and only if the spectral radius of the matrix $A$ with non-negative entries is strictly less than $1$.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.