CALLIOPE · BCOM

This working paper presents an interactive pedagogical exposition of the Galton–Watson branching process, a foundational stochastic model describing the probabilistic evolution of self-replicating populations across discrete generations. The framework centers on the offspring distribution and its probability generating function f(s), through which the critical extinction probability q is characterized as the smallest fixed point of f(s) on the unit interval. The theoretical treatment establishes the classical dichotomy: when the mean offspring number μ does not exceed one, extinction is certain, whereas supercritical regimes (μ > 1) admit a strictly positive probability of indefinite survival. The exposition integrates real-time simulation, allowing empirical extinction frequencies to be compared against theoretical predictions derived from the generating function. A prediction game further reinforces intuition by challenging the reader to estimate extinction probabilities for unknown offspring distributions before analytic confirmation. The work is aimed at students and practitioners seeking a hands-on, mathematically grounded introduction to branching processes and their phase-transition behavior.

An interactive teaching tool that lets you feel the extinction theorem by playing with a branching process until the math clicks.

The Galton–Watson process asks a simple question: if every individual in a population independently produces a random number of offspring, does the lineage survive forever or die out? The answer hinges almost entirely on one number — the mean offspring count μ. This paper (really an interactive web demo with an accompanying short PDF note) builds intuition for that fact by letting you manipulate the offspring distribution in real time and watch the family tree respond.

The mathematical core is the probability generating function f(s), which encodes the full offspring distribution in a single function. The extinction probability q — the chance the lineage eventually dies out — turns out to be the smallest fixed point of f(s) on the interval [0,1]. That's a clean, non-obvious result: you don't need to simulate anything, you just need to find where f(s) crosses the diagonal. When μ ≤ 1, the only such fixed point is s = 1, meaning extinction is guaranteed. When μ > 1, a second fixed point appears strictly below 1, and that's your survival probability.

The demo makes this tangible in two ways. First, a live simulation runs batches of family trees and reports empirical extinction frequencies, which you can compare against the theoretical q. Second, a "prediction game" shows you a mystery offspring distribution and asks you to guess q before the math is revealed — a clever way to force you to build intuition rather than just read formulas.

The source itself is sparse: it's a single-file browser demo with minimal prose, and the mathematical depth lives in the linked PDF note rather than the markdown. The source does not make the PDF's contents explicit, so the depth of the theoretical treatment there cannot be assessed from what's provided. What the demo does well is collapse the gap between the abstract fixed-point theorem and the visceral experience of watching a population blink out of existence.