STATISTICS GRADUATE LEVEL • PROBABILITY FOUNDATIONS

Sigma-Algebras & Probability Axioms — Sigma-algebras and probability axioms (conceptual)

The rigorous measure-theoretic foundations that transformed probability from intuition into a consistent mathematical framework.

Historical Context & Motivation

For centuries, probability was guided by intuition and combinatorial reasoning—counting favorable outcomes and dividing by totals. While this approach sufficed for card games and dice, it became increasingly clear by the late nineteenth century that a rigorous, axiomatic foundation was needed. Paradoxes arose when mathematicians attempted to assign probabilities to continuous sample spaces or to infinite collections of events. The question loomed: which subsets of outcomes should be considered "events" at all, and what rules must a probability function obey? Answering these questions required importing ideas from measure theory and set algebra, ultimately culminating in one of the most influential contributions to twentieth-century mathematics.

1874
Cantor's Set Theory
Georg Cantor establishes the foundations of set theory, introducing rigorous tools for reasoning about infinite collections—a prerequisite for defining σ-algebras on uncountable sample spaces.
1901
Lebesgue Measure
Henri Lebesgue defines his integral and measure, providing a way to assign lengths, areas, and volumes to a vast class of sets. His framework shows that not every subset of the real line can be measured consistently, motivating the need for σ-algebras.
1902
Vitali's Non-Measurable Set
Giuseppe Vitali demonstrates the existence of a set that cannot be assigned a Lebesgue measure without contradiction. This result underscores that restricting attention to a well-chosen collection of "measurable" sets is essential, not optional.
1933
Kolmogorov's Axioms
Andrey Kolmogorov publishes Grundbegriffe der Wahrscheinlichkeitsrechnung (Foundations of the Theory of Probability), laying out the axiomatic framework that defines probability as a measure on a σ-algebra. This work unifies probability with the broader edifice of measure theory.
1950s
Modern Probability Theory
Doob, Itô, and others build on Kolmogorov's axioms to develop stochastic processes, martingale theory, and stochastic calculus, demonstrating the power and necessity of the measure-theoretic foundation.

The central insight driving this history is deceptively simple: before you can assign probabilities to events, you must first decide which collections of outcomes qualify as events. A σ-algebra provides exactly this structure, and Kolmogorov's axioms tell us how a probability function should behave on such a collection. Together, they resolve centuries of conceptual ambiguity and furnish the language in which all of modern probability is written.

Core Principles & Definitions

The entire probabilistic framework rests on three interrelated objects: a sample space Ω, a σ-algebra ℱ of subsets of Ω, and a probability measure P defined on ℱ. The triple (Ω, ℱ, P) is called a probability space. Understanding why each component is necessary—and what constraints each must satisfy—is the core intellectual content of this lesson.

1

Sample Space (Ω)

The set of all possible outcomes of a random experiment. It may be finite (coin flips), countably infinite (number of emails per day), or uncountable (continuous measurements). Every probabilistic statement ultimately references subsets of Ω.
2

σ-Algebra (ℱ)

A collection of subsets of Ω that is closed under complementation and countable unions, and contains Ω itself. It specifies which subsets of Ω are deemed "events" and therefore eligible to receive a probability. The symbol σ reflects the countable ("sigma") nature of the closure operations.
3

Probability Measure (P)

A function P : ℱ → [0, 1] satisfying Kolmogorov's three axioms: non-negativity, normalization (P(Ω) = 1), and countable additivity. It assigns a numerical "weight" to each event in ℱ, quantifying the likelihood of that event.
4

Closure Properties

A σ-algebra is closed under complement (if A ∈ ℱ then Aᶜ ∈ ℱ), countable union (if A₁, A₂, … ∈ ℱ then ⋃Aₙ ∈ ℱ), and by De Morgan's laws, countable intersection. These guarantees ensure any "reasonable" combination of events is still an event.
5

Probability Space (Ω, ℱ, P)

The complete triple. Without Ω we have no outcomes; without ℱ we cannot specify events; without P we cannot quantify uncertainty. All three must be specified together, and they are inseparable in rigorous probability.
KEY TAKEAWAY
Think of a σ-algebra as the rulebook for a filing system. Imagine an office where certain folders are "official"—if a folder about event A is official, then the folder about "not A" must also be official, and if you merge any countable stack of official folders into one combined folder, that result must be official too. A probability measure is then a clerk who stamps each official folder with a number between 0 and 1, obeying the rule that the entire filing cabinet sums to exactly 1. Without the filing system (σ-algebra), the clerk (probability measure) would have nowhere consistent to work.

Visual Explanation — Structure of a σ-Algebra

The following diagram illustrates the σ-algebra on a simple sample space Ω = {a, b, c}. The power set 2Ω contains all 2³ = 8 subsets and is itself a valid σ-algebra (the largest possible one on this Ω). The diagram shows how each element connects to its complement and how unions close the structure.

The lattice of all 8 subsets of Ω = {a, b, c}. Solid lines connect a set to its subsets (containment); dashed amber lines link each set to its complement. Every valid σ-algebra on Ω is a sub-lattice of this power set that includes Ω, ∅, and is closed under complement and countable union.

Notice that the trivial σ-algebra {∅, Ω} is the smallest possible σ-algebra on any Ω: it contains only the certain event and the impossible event. At the other extreme, the power set 2^Ω is the largest—every subset is an event. For finite sample spaces, using the full power set is standard. However, for uncountable spaces like ℝ, the power set is too large (some subsets cannot be assigned a consistent measure), so we work with the Borel σ-algebra, generated by all open intervals.

Mathematical Framework — The Axioms of Probability

Kolmogorov's 1933 axioms are elegant in their economy: three conditions suffice to build the entire theory of probability. Let (Ω, ℱ) be a measurable space. A function P : ℱ → ℝ is called a probability measure if and only if the following three axioms hold.

AXIOM 1 — NON-NEGATIVITY
∀ A ∈ ℱ : P(A) ≥ 0
Every event receives a non-negative probability. This reflects the intuition that negative likelihood is meaningless. Combined with normalization, this confines probabilities to the interval [0, 1].
AXIOM 2 — NORMALIZATION (UNITARITY)
P(Ω) = 1
The certain event—the event that some outcome occurs—has probability exactly 1. This anchors the entire probability scale.
AXIOM 3 — COUNTABLE ADDITIVITY (σ-ADDITIVITY)
If A₁, A₂, A₃, … ∈ ℱ are pairwise disjoint, then P(⋃ₙ₌₁^∞ Aₙ) = Σₙ₌₁^∞ P(Aₙ)
For any countable collection of mutually exclusive events, the probability of their union equals the sum of their individual probabilities. This is far stronger than mere finite additivity: it permits passage to limits, which is essential for defining distributions on continuous spaces and for proving limit theorems.

Key Consequences of the Axioms

From these three axioms alone, a rich collection of results follows by purely logical deduction. First, setting A₁ = Ω and A₂ = A₃ = … = ∅ in Axiom 3 yields P(∅) = 0. Second, because any event A and its complement Aᶜ partition Ω, we get P(Aᶜ) = 1 − P(A). Third, for any two events A and B, the inclusion–exclusion principle states P(A ∪ B) = P(A) + P(B) − P(A ∩ B). These are not additional axioms but theorems derived from the axioms, illustrating how much structure the three conditions encode.

MONOTONICITY (DERIVED)
A ⊆ B ⟹ P(A) ≤ P(B)
If every outcome in event A is also in event B, then B is at least as likely as A. Proof: B = A ∪ (B \ A) with disjoint union, so P(B) = P(A) + P(B \ A) ≥ P(A) by non-negativity.

Detailed Breakdown — Examples & Generated σ-Algebras

A key concept in practice is the σ-algebra generated by a collection of sets. Given an arbitrary family 𝒞 of subsets of Ω, the generated σ-algebra σ(𝒞) is defined as the intersection of all σ-algebras containing 𝒞. Because the intersection of any family of σ-algebras is again a σ-algebra, σ(𝒞) is well-defined and is the smallest σ-algebra containing every set in 𝒞. The most important example in all of probability is the Borel σ-algebra ℬ(ℝ), generated by the open intervals of ℝ.

Three nested σ-algebras on Ω = {1, 2, 3, 4}. The trivial σ-algebra (innermost, cyan) is the coarsest. The σ-algebra generated by {1,2} (middle, violet) can distinguish whether the outcome is in {1,2} or {3,4}. The power set (outermost, red) distinguishes every individual outcome.
Common σ-algebras and their interpretive meaning
σ-AlgebraMembersIntuition
{∅, Ω}∅, Ω"Something happens or nothing does." No outcomes are distinguishable—no interesting questions can be asked.
σ({A})∅, A, Aᶜ, ΩDistinguishes exactly one binary question: "Did A occur?" Equivalent to knowing one yes/no fact.
2^ΩAll subsets of ΩMaximum resolution. Every singleton is an event, so you know exactly which outcome occurred.
ℬ(ℝ)Generated by open intervalsThe standard σ-algebra on the real line. Contains all open sets, closed sets, countable unions and intersections thereof. Excludes exotic non-measurable sets.

Worked Example — Verifying a σ-Algebra and Computing Probabilities

Consider a sample space Ω = {1, 2, 3} and the collection ℱ = {∅, {1}, {2, 3}, Ω}. We will verify that ℱ is a σ-algebra, then define a valid probability measure on it and compute several probabilities using the axioms.

Verifying σ-Algebra Properties & Computing Probabilities
1
Step 1 — Check Ω ∈ ℱBy inspection, Ω = {1, 2, 3} is listed in ℱ. The first σ-algebra requirement is satisfied.
✓ Ω ∈ ℱ
2
Step 2 — Check Closure Under ComplementationWe need to verify that for every set in ℱ, its complement relative to Ω is also in ℱ. The complement of ∅ is Ω ∈ ℱ. The complement of Ω is ∅ ∈ ℱ. The complement of {1} is {2, 3} ∈ ℱ. The complement of {2, 3} is {1} ∈ ℱ. Every complement is present.
✓ Closed under complement
3
Step 3 — Check Closure Under Countable UnionsSince ℱ has only four elements, we check all pairwise unions. ∅ ∪ A = A for any A ∈ ℱ. {1} ∪ {2, 3} = {1, 2, 3} = Ω ∈ ℱ. {1} ∪ Ω = Ω ∈ ℱ. {2, 3} ∪ Ω = Ω ∈ ℱ. All unions are in ℱ, and by induction any countable union of members of ℱ is again in ℱ.
✓ Closed under countable unions → ℱ is a σ-algebra
4
Step 4 — Define a Probability MeasureSet P({1}) = 0.4. Since {1} and {2, 3} are disjoint and their union is Ω, Axiom 3 (finite additivity) gives P({1}) + P({2, 3}) = P(Ω) = 1 by Axiom 2. Therefore P({2, 3}) = 1 − 0.4 = 0.6. Also P(∅) = 0 (derived from the axioms). We have now fully specified P on ℱ.
P(∅) = 0, P({1}) = 0.4, P({2, 3}) = 0.6, P(Ω) = 1
5
Step 5 — Verify AxiomsAxiom 1 (non-negativity): 0, 0.4, 0.6, 1 are all ≥ 0. Axiom 2 (normalization): P(Ω) = 1. Axiom 3 (countable additivity): The only non-trivial disjoint decomposition is {1} ∪ {2, 3} = Ω, and P({1}) + P({2, 3}) = 0.4 + 0.6 = 1 = P(Ω). All three axioms are satisfied.
(Ω, ℱ, P) is a valid probability space
6
Step 6 — Observe What Is Not an EventNotice that {2} is not in ℱ. In this probability space, we cannot ask "What is the probability that the outcome is 2?" because {2} is not a measurable event. This is the precise sense in which a σ-algebra determines which questions have probabilistic answers.
{2} ∉ ℱ → P({2}) is undefined in this probability space

Finite Additivity vs. Countable Additivity — Why It Matters

One might ask: why does Kolmogorov's third axiom demand countable additivity rather than the seemingly more natural condition of finite additivity? This distinction is not pedantic—it is the fulcrum on which the entire power of modern probability rests. Finite additivity alone cannot guarantee that limits of convergent sequences of events behave well, and without well-behaved limits, the law of large numbers, the central limit theorem, and indeed all of asymptotic statistics would collapse.

Finite vs. countable additivity
PropertyFinite AdditivityCountable Additivity (Kolmogorov)
StatementP(A₁ ∪ … ∪ Aₙ) = Σᵢ P(Aᵢ) for finitely many disjoint eventsP(⋃ₙ₌₁^∞ Aₙ) = Σₙ₌₁^∞ P(Aₙ) for countably many disjoint events
Continuity of PNot guaranteed. P may fail at limits.Implies continuity from below and above: if Aₙ ↑ A then P(Aₙ) → P(A).
Law of Large NumbersCannot be proved; the concept of convergence of infinite sequences is inaccessible.Both the weak and strong laws are provable.
Existence of distributions on ℝCannot consistently define Gaussian or other continuous distributions.Lebesgue integration and density functions are well-defined.
Notable advocatesde Finetti argued for finitely additive foundations in Bayesian probability.Kolmogorov, Doob, and the mainstream of modern probability.
KEY TAKEAWAY
Countable additivity is like a bridge engineer's requirement that the load capacity be maintained not just for any finite number of trucks, but for an unending convoy. Finite additivity would only guarantee that any specific group of trucks is safe, without ensuring the bridge holds as the convoy grows without bound. In probability, that "bridge" is the passage to the limit, and countable additivity is what keeps it standing.

Connections to General Measure Theory and Advanced Probability

A probability measure is simply a special case of a measure in the sense of Lebesgue–Carathéodory theory, namely one where the total measure is normalized to 1. Recognizing this connection opens the door to applying the full toolkit of measure theory—dominated convergence, Fubini's theorem, Radon–Nikodym derivatives—to probabilistic problems. The table below contrasts the probabilistic language with general measure-theoretic terminology.

Probability ↔ Measure theory dictionary
Probability ConceptMeasure Theory ConceptRelationship
Sample space ΩMeasurable space XΩ is an X with the interpretation of "all possible outcomes."
σ-algebra ℱσ-algebra 𝒜Identical mathematical object; ℱ is the conventional symbol in probability.
Probability measure PMeasure μP is a measure with the constraint μ(X) = 1.
Random variable XMeasurable function fX: Ω → ℝ is measurable iff X⁻¹(B) ∈ ℱ for every B ∈ ℬ(ℝ).
Expectation E[X]Lebesgue integral ∫ f dμE[X] = ∫_Ω X dP, a special case of integration against a measure.
Conditional expectation E[X | 𝒢]Radon–Nikodym derivativeDefined via the Radon–Nikodym theorem as a 𝒢-measurable function satisfying an integral identity.
🔭 Looking Ahead
Once you internalize the probability space (Ω, ℱ, P), the next major step is defining random variables as measurable functions and constructing product spaces for joint distributions. The σ-algebra generated by a random variable (the "σ-algebra of information") is the conceptual bridge to conditional expectation, filtrations, and martingale theory—central tools in stochastic processes and mathematical finance.

Practice Problems

PROBLEM 1CONCEPTUAL
Explain in your own words why the collection ℱ = {∅, {1}, {2}, Ω} on Ω = {1, 2, 3} is not a σ-algebra. Identify the specific closure property that fails and give a concrete example demonstrating the failure.
PROBLEM 2BASIC CALCULATION
Let Ω = {a, b, c, d} and ℱ = 2^Ω (the power set). Define P({a}) = 0.1, P({b}) = 0.3, P({c}) = 0.2, P({d}) = 0.4. Compute P({a, c}), P({b, c, d}), and P({a}ᶜ), justifying each step by referencing a specific axiom or derived property.
PROBLEM 3INTERMEDIATE
Let Ω = {1, 2, 3, 4, 5, 6} represent the outcomes of a fair die. Find the σ-algebra generated by the collection 𝒞 = {{1, 2, 3}, {3, 4}}. List all elements of σ(𝒞) explicitly.
PROBLEM 4APPLIED
A medical diagnostic test partitions the patient population Ω into four groups based on disease status (D or Dᶜ) and test result (T⁺ or T⁻). The lab's σ-algebra ℱ_lab = σ({T⁺}) = {∅, {T⁺ outcomes}, {T⁻ outcomes}, Ω} only reflects test results—it cannot distinguish D from Dᶜ within the same test result group. Explain why P(D | ℱ_lab) cannot be computed as a single number, and describe conceptually what the conditional expectation E[1_D | ℱ_lab] represents as a random variable.
PROBLEM 5CRITICAL THINKING
Prove that countable additivity implies continuity from below: if {Aₙ} is an increasing sequence of events (A₁ ⊆ A₂ ⊆ …) with A = ⋃ₙ₌₁^∞ Aₙ, then P(Aₙ) → P(A) as n → ∞. Hint: construct a disjoint sequence whose union equals A.

Lesson Summary

Modern probability rests on the probability space (Ω, ℱ, P). The sample space Ω catalogs every possible outcome of a random experiment. The σ-algebra ℱ specifies which subsets of Ω qualify as measurable events, enforcing closure under complement and countable union to guarantee logical consistency. The probability measure P maps events to the interval [0, 1] subject to Kolmogorov's three axioms: non-negativity, normalization (P(Ω) = 1), and countable additivity.

Countable additivity—not merely finite additivity—is essential because it ensures continuity of P and the validity of limit theorems that underpin all of statistical inference. The Borel σ-algebra on ℝ resolves the technical obstacle of non-measurable sets, while generated σ-algebras provide a systematic way to build the "right" event space for any problem. Together, these structures unify probability with Lebesgue measure theory and form the indispensable language for random variables, expectation, conditioning, and stochastic processes.

Varsity Tutors • Statistics Graduate Level • Sigma-Algebras & Probability Axioms