diff --git a/probability/02_axioms.py b/probability/02_axioms.py new file mode 100644 index 0000000..4ab3bd6 --- /dev/null +++ b/probability/02_axioms.py @@ -0,0 +1,214 @@ +# /// script +# requires-python = ">=3.11" +# dependencies = [ +# "marimo", +# "matplotlib==3.10.0", +# "numpy==2.2.2", +# ] +# /// + +import marimo + +__generated_with = "0.11.2" +app = marimo.App(width="medium") + + +@app.cell +def _(): + import marimo as mo + return (mo,) + + +@app.cell(hide_code=True) +def _(mo): + mo.md( + r""" + # Axioms of Probability + + Probability theory is built on three fundamental axioms, known as the [Kolmogorov axioms](https://en.wikipedia.org/wiki/Probability_axioms). These axioms form + the mathematical foundation for all of probability theory[1](https://chrispiech.github.io/probabilityForComputerScientists/en/part1/probability). + + Let's explore each axiom and understand why they make intuitive sense: + """ + ) + return + + +@app.cell(hide_code=True) +def _(mo): + mo.md( + r""" + ## The Three Axioms + + | Axiom | Mathematical Form | Meaning | + |-------|------------------|----------| + | **Axiom 1** | $0 \leq P(E) \leq 1$ | All probabilities are between 0 and 1 | + | **Axiom 2** | $P(S) = 1$ | The probability of the sample space is 1 | + | **Axiom 3** | $P(E \cup F) = P(E) + P(F)$ | For mutually exclusive events, probabilities add | + + where the set $S$ is the sample space (all possible outcomes), and $E$ and $F$ are sets that represent events. The notation $P(E)$ denotes the probability of $E$, which you can interpret as the chance that something happens. $P(E) = 0$ means that the event cannot happen, while $P(E) = 1$ means the event will happen no matter what; $P(E) = 0.5$ means that $E$ has a 50% chance of happening. + + For an example, when rolling a fair six-sided die once, the sample space $S$ is the set of die faces ${1, 2, 3, 4, 5, 6}$, and there are many possible events; we'll see some examples below. + """ + ) + return + + +@app.cell(hide_code=True) +def _(mo): + mo.md( + r""" + ## Understanding Through Examples + + Let's explore these axioms using a simple experiment: rolling a fair six-sided die. + We'll use this to demonstrate why each axiom makes intuitive sense. + """ + ) + return + + +@app.cell +def _(event): + event + return + + +@app.cell(hide_code=True) +def _(mo): + # Create an interactive widget to explore different events + + event = mo.ui.dropdown( + + options=[ + + "Rolling an even number (2,4,6)", + + "Rolling an odd number (1,3,5)", + + "Rolling a prime number (2,3,5)", + + "Rolling less than 4 (1,2,3)", + + "Any possible roll (1,2,3,4,5,6)", + + ], + + value="Rolling an even number (2,4,6)", + + label="Select an event" + + ) + return (event,) + + +@app.cell(hide_code=True) +def _(event, mo, np, plt): + # Define the probabilities for each event + event_map = { + "Rolling an even number (2,4,6)": [2, 4, 6], + "Rolling an odd number (1,3,5)": [1, 3, 5], + "Rolling a prime number (2,3,5)": [2, 3, 5], + "Rolling less than 4 (1,2,3)": [1, 2, 3], + "Any possible roll (1,2,3,4,5,6)": [1, 2, 3, 4, 5, 6], + } + + # Get outcomes directly from the event value + outcomes = event_map[event.value] + prob = len(outcomes) / 6 + + # Visualize the probability + dice = np.arange(1, 7) + colors = ['#1f77b4' if d in outcomes else '#d3d3d3' for d in dice] + + fig, ax = plt.subplots(figsize=(8, 2)) + ax.bar(dice, np.ones_like(dice), color=colors) + ax.set_xticks(dice) + ax.set_yticks([]) + ax.set_title(f"P(Event) = {prob:.2f}") + + # Add explanation + explanation = mo.md(f""" + **Event**: {event.value} + + **Probability**: {prob:.2f} + + **Favorable outcomes**: {outcomes} + + This example demonstrates: + + - Axiom 1: The probability is between 0 and 1 + + - Axiom 2: For the sample space, P(S) = 1 + + - Axiom 3: The probability is the sum of individual outcome probabilities + """) + + mo.hstack([plt.gcf(), explanation]) + return ax, colors, dice, event_map, explanation, fig, outcomes, prob + + +@app.cell(hide_code=True) +def _(mo): + mo.md( + r""" + ## Why These Axioms Matter + + These axioms are more than just rules - they provide the foundation for all of probability theory: + + 1. **Non-negativity** (Axiom 1) makes intuitive sense: you can't have a negative number of occurrences + in any experiment. + + 2. **Normalization** (Axiom 2) ensures that something must happen - the total probability must be 1. + + 3. **Additivity** (Axiom 3) lets us build complex probabilities from simple ones, but only for events + that can't happen together (mutually exclusive events). + + From these simple rules, we can derive all the powerful tools of probability theory that are used in + statistics, machine learning, and other fields. + """ + ) + return + + +@app.cell(hide_code=True) +def _(mo): + mo.md( + r""" + ## 🤔 Test Your Understanding + + Consider rolling two dice. Which of these statements follow from the axioms? + +
+ 1. P(sum is 13) = 0 + + ✅ Correct! This follows from Axiom 1. Since no combination of dice can sum to 13, + the probability must be non-negative but can be 0. +
+ +
+ 2. P(sum is 7) + P(sum is not 7) = 1 + + ✅ Correct! This follows from Axioms 2 and 3. These events are mutually exclusive and cover + the entire sample space. +
+ +
+ 3. P(first die is 6 or second die is 6) = P(first die is 6) + P(second die is 6) + + ❌ Incorrect! This doesn't follow from Axiom 3 because the events are not mutually exclusive - + you could roll (6,6). +
+ """ + ) + return + + +@app.cell +def _(): + import numpy as np + import matplotlib.pyplot as plt + return np, plt + + +if __name__ == "__main__": + app.run()