Probability Theory

probability-theory

Definition

Probability Theory

Probability theory is the mathematical theory of randomness. It models uncertain outcomes using a probability space $(\Omega ,\mathcal{F},P)$, where $\Omega$ contains the possible outcomes, $\mathcal{F}$ contains the events, and $P$ assigns a probability to each event.

Direction of Reasoning

$$(\Omega ,\mathcal{F},P)\text{ known}⟶P(A)\text{ computed for }A\in \mathcal{F}.$$

The population or probability model is treated as known. Probability theory then applies logical, self-contained rules to calculate the probabilities of events concerning it.

Character

• It provides rules for calculating probabilities.
• Its conclusions follow logically from the specified model and assumptions.
• Once these are fixed, a well-posed probability question has one correct answer.

Empirical Grounding

Probability theory is deductive: its axioms determine what follows from a probability model, but they do not by themselves guarantee that the model describes a real process. Reality enters when the model is constructed and tested:

1. A random experiment, its possible outcomes, and relevant events are identified.
2. Probabilities are assigned using known mechanisms, symmetries, or observed data.
3. The resulting predictions are compared with further observations and the model is retained, refined, or rejected.

For an event $A$ observed over $n$ repetitions, its empirical frequency is

$${\hat{P}}_n(A)=\frac{1}{n}\sum _{i=1}^n{1}_A(X_i),$$

where $1_A(X_i)=1$ when $A$ occurs and $0$ otherwise. If the repetitions are independent and follow the same model with $P(A)=p$, the law of large numbers gives

$${\hat{P}}_n(A)\stackrel{P}{\to }p.$$

Predicted probabilities can therefore be checked against stable frequencies across repeated or comparable situations. Probability theory works well when real processes exhibit such regularities and the chosen model captures the mechanisms relevant to the question.

A die model

The symmetry of an ideal die motivates $P(\{i\})=1/6$ for each face $i$. Repeated rolls test this model: persistent frequencies far from $1/6$ suggest that the die or rolling mechanism is not adequately represented by the ideal model.

Warning

The law of large numbers is conditional on the probability model; it does not prove that nature follows that model. Empirical agreement supports the modelling assumptions, while systematic disagreement requires them to be revised.