# Idea Basic Urn model: An Urn contains balls of different colors. The outcome equals the color of the ball selected. Simplest model: [[Bernoulli distribution|Bernoulli]] model where outcomes are independent. No path dependence. Polya process (Phat-dependence, see [[path dependence]]): Select ball and return. Add a new ball that is the same color as the ball selected. Path dependence. - Any probability of red balls is an equilibrium and equally likely. That is, anything could happen, and they are all equally likely to happen. - Any history of blue and red balls are equally likely. Knowing the frequency of balls doesn't tell us anything about the sequence. Like [[Bayes theorem]]^[[[Adam Bear]]], where the initial state of the world is the [[prior]], and data are being added constantly, which updates the state of the world. ```python def polya(steps=100): urn = [0, 1] # "prior" for _ in range(steps): np.random.shuffle(urn) # "likelihood" urn.append(urn[0]) # "data" return urn x = [np.mean(polya()) for i in range(1000)] np.mean(x) # close to 50 plt.hist(x) ``` Balancing: Select and return. Add a new ball that is the opposite color of the ball selected. - This process converges to equal percentages of the two colors of balls. - Equilibrium is always 50%. Sway process: Select and return. ![[Pasted image 20210615003046.png|800]] # Types of [[path dependence]] Path-dependent **outcomes**: color of ball in a given period depends on the path (Polya, balancing processes). Path-dependent **equilibrium**: percentage of red balls in long run depends on the path (Polya process). ![[Pasted image 20210615002238.png|800]] # References - https://www.coursera.org/learn/model-thinking/lecture/ym0VV/urn-models