Extinction Probability
The extinction probability of a branching process with offspring probability generating function satisfies
and is the minimal non-negative solution of ; moreover if and only if .
Let be the PGF of . Then , and
Proof of the fixed point equation
Since and is continuous with , letting gives .
For directly: .
Alternatively, conditioning on and writing for independent copies of the process started from one individual,
Means of the generations
Let be a branching process with offspring distribution . Then,
Proof of the mean formula
By induction: , and assuming ,
so and taking expectation gives .
Generating functions across generations
Let and . Then
Minimal solution characterisation
Let be a branching process with offspring distribution . Assume that .
Then the extinction probability is the minimal non-negative solution to the equation .
Moreover, iff .
Proof of the characterisation
Let be the smallest non-negative solution of . Induction on shows : the case is clear, and increasing on gives
Hence as , while because itself solves the fixed point equation; so .
For the criterion that exactly when , first suppose . Then and , so
and since , necessarily .
Now assume and . Define , so and
because some with . Hence is strictly increasing on , so by Rolle’s theorem has at most one root other than .
If there is no other root then , and since and we have on , whence
that is .
If instead for some , that root must be the extinction probability . Rolle’s theorem gives with , and strict increase yields
so .
Related
Stated in
- Proposition 3.68§3.7.2 Extinction Probability
- Theorem 3.66§3.7.1 Generating Functions of Branching Processes
- Theorem 3.67§3.7.1 Generating Functions of Branching Processes
- Theorem 3.69§3.7.2 Extinction Probability
