Branching Process Simulator: Galton-Watson Extinction & Growth

simulator intermediate ~10 min
Loading simulation...
q ≈ 0.42 — 42% extinction probability

With mean offspring μ=1.5, the extinction probability is approximately 42%. The expected population grows as 1.5ⁿ per generation, but any single lineage has substantial chance of dying out due to stochastic fluctuations.

Formula

E[Zₙ] = μⁿ (expected population at generation n)
Var[Zₙ] = σ²μⁿ⁻¹(μⁿ − 1)/(μ − 1)
q = G(q), smallest non-negative root (extinction probability)

The Extinction of Family Names

In 1874, Francis Galton posed a question to readers of the Educational Times: what is the probability that a family surname goes extinct? Henry Watson provided a mathematical framework using probability generating functions, founding the theory of branching processes. Their model — where each individual independently produces a random number of offspring — revealed a fundamental threshold: if the average number of offspring is at most one, extinction is certain; above one, there is hope for survival.

Criticality and Phase Transitions

The branching process exhibits a sharp phase transition at μ = 1. In the subcritical regime (μ < 1), the population decays geometrically and extinction occurs quickly. At criticality (μ = 1), the population performs an unbiased random walk toward zero — extinction is still certain but takes much longer. In the supercritical regime (μ > 1), the population either goes extinct early or grows exponentially without bound, with nothing in between.

Stochastic Genealogies

Each realization of a branching process creates a random genealogical tree. This simulation visualizes these trees, showing how different parameter choices produce dramatically different structures — sparse, quickly-dying lineages when subcritical versus bushy, exponentially expanding trees when supercritical. The randomness means that even supercritical populations can go extinct if the first few generations are unlucky.

Modern Applications

Branching processes are ubiquitous in modern science. In cancer biology, they model tumor initiation — a single mutant cell must survive stochastic fluctuations to establish a tumor. In nuclear physics, they describe neutron chain reactions (the basis of criticality in reactors). In epidemiology, the early phase of an outbreak is a branching process where R₀ is the mean offspring number. In computer science, they analyze recursive algorithms and random tree data structures.

FAQ

What is a Galton-Watson branching process?

A Galton-Watson process is a stochastic model where each individual independently produces a random number of offspring according to a fixed distribution. Introduced by Francis Galton and Henry Watson in 1874 to study the extinction of family surnames, it is the simplest model of population growth with randomness.

When does a branching process go extinct?

If the mean offspring μ ≤ 1, extinction is certain (probability 1). If μ > 1, there is a positive probability of eternal survival, but extinction can still occur with probability q < 1, where q is the smallest root of the probability generating function equation G(s) = s.

How is it applied in biology?

Branching processes model cell division in tumors, neutron chain reactions, viral replication within a host, spread of mutations through a population, and the fate of endangered species lineages. They are also used in phylogenetics to model speciation and extinction.

What is the criticality threshold?

The process is subcritical (μ < 1, certain extinction), critical (μ = 1, certain extinction but slowly), or supercritical (μ > 1, positive survival probability). At criticality, the population survives for a long time but eventually dies out — a phase transition in the mathematical sense.

Sources

Embed

<iframe src="https://homo-deus.com/lab/mathematical-biology/branching-process/embed" width="100%" height="400" frameborder="0"></iframe>
View source on GitHub