Mathwords logoMathwords

Kolmogorov Axioms — Definition, Formula & Example

The Kolmogorov axioms are three basic rules that every probability assignment must follow. Probabilities are never negative, the whole sample space has probability 1, and the probability of non-overlapping events combined equals the sum of their probabilities.

Let Ω\Omega be a sample space and F\mathcal{F} a σ\sigma-algebra of events on Ω\Omega. A function P:F→RP: \mathcal{F} \to \mathbb{R} is a probability measure if (1) P(E)≥0P(E) \ge 0 for every E∈FE \in \mathcal{F}, (2) P(Ω)=1P(\Omega) = 1, and (3) for any countable sequence of pairwise disjoint events E1,E2,…E_1, E_2, \ldots, P(⋃iEi)=∑iP(Ei)P\left(\bigcup_{i} E_i\right) = \sum_{i} P(E_i). The triple (Ω,F,P)(\Omega, \mathcal{F}, P) is called a probability space.

Key Formula

(1) P(E)≥0(2) P(Ω)=1(3) P(⋃i=1∞Ei)=∑i=1∞P(Ei) for pairwise disjoint Ei\text{(1) } P(E) \ge 0 \qquad \text{(2) } P(\Omega) = 1 \qquad \text{(3) } P\left(\bigcup_{i=1}^{\infty} E_i\right) = \sum_{i=1}^{\infty} P(E_i) \text{ for pairwise disjoint } E_i
Where:
  • E,EiE, E_i = Events, meaning subsets of the sample space that belong to the event collection
  • Ω\Omega = The sample space, the set of all possible outcomes
  • PP = The probability measure, which assigns a number to each event
  • ⋃\bigcup = Union: the event that at least one of the EiE_i occurs

Worked Example

Problem: A carnival spinner has four colored sectors of unequal size. The game designer claims P(Red) = 0.40, P(Blue) = 0.25, and P(Green) = 0.15. The remaining sector is Yellow. Use the Kolmogorov axioms to find P(Yellow), confirm the model is valid, and find P(Red or Blue).
Step 1: The four single-color outcomes are pairwise disjoint, and their union is the whole sample space. Axiom 3 says their probabilities add to P(Ω)P(\Omega), and Axiom 2 says P(Ω)=1P(\Omega) = 1. Solve for the missing value.
P(Yellow)=1−(0.40+0.25+0.15)=1−0.80=0.20P(\text{Yellow}) = 1 - (0.40 + 0.25 + 0.15) = 1 - 0.80 = 0.20
Step 2: Check Axiom 1. All four values are nonnegative, so the assignment satisfies every axiom and is a valid probability model.
0.40, 0.25, 0.15, 0.20≥00.40,\ 0.25,\ 0.15,\ 0.20 \ge 0
Step 3: Red and Blue cannot both occur on one spin, so the events are disjoint. Apply Axiom 3 directly.
P(Red∪Blue)=0.40+0.25=0.65P(\text{Red} \cup \text{Blue}) = 0.40 + 0.25 = 0.65
Answer: P(Yellow) = 0.20, the model is valid, and P(Red or Blue) = 0.65.

Why It Matters

Every rule in an intro statistics course traces back to these three axioms, including the complement rule and the general addition rule. In upper-level probability and measure-theory courses, you prove results by reducing them to the axioms. The axioms also give you a test for whether a proposed model, such as a machine-learning classifier's output, is a legitimate probability distribution.

Common Mistakes

Mistake: Listing P(E)≤1P(E) \le 1 or P(∅)=0P(\emptyset) = 0 as extra axioms.
Correction: Both facts are theorems, not axioms. For example, EE and EcE^c are disjoint with union Ω\Omega, so P(E)+P(Ec)=1P(E) + P(E^c) = 1. Since P(Ec)≥0P(E^c) \ge 0, it follows that P(E)≤1P(E) \le 1.

Related Terms