Measure Theory 1 | Sigma Algebras [dark version]

Measure theory starts by replacing “length of an interval” with a more flexible idea: a way to assign a generalized volume to subsets of a set X. On the real line, length is easy for intervals—an interval from a to b naturally has length b − a—but complicated subsets force a broader framework. The goal becomes abstract: given an arbitrary set X, identify which subsets are allowed to be “measurable,” so that a consistent notion of size can be defined later.

The first step is to look at the power set of X, the collection of all subsets. Not every subset can realistically be measured in a way that behaves nicely under the operations needed for integration. So measure theory introduces a special family of subsets, called a sigma algebra (written as Σ). The sets inside Σ are precisely the ones that will be measurable.

A sigma algebra must satisfy three closure rules. First, it must contain the simplest sets: the empty set ∅ and the whole space X. This ensures there is a baseline notion of size for “nothing” and “everything.” Second, it must be closed under complements: if a set A is measurable (A ∈ Σ), then the complement Ac = X \ A must also be measurable. Intuitively, if the inside of A has a well-defined generalized volume, then the outside should too.

Third, Σ must be closed under countable unions. If A1, A2, A3, … are measurable sets, then their union ⋃_{i=1}^∞ Ai must also be measurable. This rule is crucial because measures are designed to interact predictably with limits of increasingly fine approximations—exactly the kind of structure that integration relies on. The transcript links this to the idea that the “volume” of a union should correspond to the sum (or limit of sums) of the volumes of the pieces, which only makes sense if the union stays within the measurable collection.

With these rules, the transcript frames sigma algebras as the backbone of measure theory: they define the universe of sets on which a measure can be consistently defined. Every measurable set is measurable relative to a particular sigma algebra, so “measurable” is not absolute—it depends on Σ.

Two extreme examples anchor the concept. The smallest sigma algebra is {∅, X}, which contains only the trivial measurable sets. The largest is the power set P(X), which would allow every subset to be measurable; it automatically satisfies all three sigma algebra rules because it contains everything. In practice, important sigma algebras sit between these extremes: large enough to support the constructions needed for integration, but restricted enough to avoid pathologies that prevent a consistent measure from existing on all subsets.

The takeaway is that sigma algebras are not a side definition—they determine what “measurable” means and set the stage for defining measures and, eventually, integrals. Next steps naturally move from this set-theoretic structure to the definition of a measure itself.