1 Measurability
1.1 Special kinds of set collections
1.1.1 p-system
A collection of sets that is closed under intersection.
1.1.2 d-system
A collection of sets \mathcal D that:
- contains the universal set E.
- is closed under subset subtraction i.e. if A, B \in \mathcal D such that B \subset A then B \setminus A \in \mathcal D.
- is closed under monotone limits i.e. if (A_n) \subset \mathcal D, A_n \nearrow A then A \in \mathcal D
1.1.3 Sigma algebra
A collection of sets that is:
- non empty
- closed under complementation
- closed under countable union
1.1.3.1 Properties of sigma algebras
- A collection of sets is a sigma algebra if and only if it is both a p-system and a d-system.
- The intersection of an arbitrary (even uncountable) collection of sigma algebras is a sigma algebra.
- A sigma algebra always contains the null set and the universal set.
- A sigma algebra is closed under countable union, countable intersection, set difference.
- The intersection of all sigma algebras containing \mathcal C is the smallest sigma algebra containing \mathcal C.
1.2 Monotone class theorem
If \mathcal D is a d-system that contains a p-system \mathcal C, then \mathcal D actually contains all of \sigma \mathcal C.
1.2.1 Good sets technique
Let’s say we want to prove some condition for a sigma algebra \mathcal F. But since proving everything for every element of a sigma algebra is impossible, we can reduce the problem to finding p-system \mathcal C that generates that sigma-algebra, proving that property for \mathcal C, and then proving that the collection of good sets.
Steps:
Define the collection of all good sets \mathcal G = \{B \subset E : B \text{ satisfies the condition}\}.
- Prove \mathcal G is a d-system.
- Find a generator \mathcal C and prove that \sigma \mathcal C = \mathcal F
- Prove that \mathcal C is a p-system.
- Prove that \mathcal C \subset \mathcal G i.e. all members of \mathcal C satisfy the condition.
By the monotone class theorem, \sigma \mathcal C = \mathcal F \subset \mathcal G and hence the condition holds for all sets in \mathcal F.
1.2.2 Example
Given two set functions, f and g, if they agree on all sets A \in \mathcal C, they agree for all sets in \sigma \mathcal C.
1.2.3 Simpler but less general form of good sets technique
Define the collection \mathcal A = \{ B \in \mathcal F : B \text{ satisfies the condition} \}
- Prove \mathcal A is a sigma algebra.
- Find a generator \mathcal C and prove that \sigma \mathcal C = \mathcal F
- Prove that \mathcal C \subset \mathcal A i.e. all members of \mathcal C satisfy the condition (they are in the sigma algebra by step 2).
Since \mathcal A is a sigma algebra it is also a d-system. Hence by the monotone class theorem \mathcal C \subset \mathcal A implies \sigma C = \mathcal F \subset \mathcal A.
This implies \mathcal A = \mathcal F and thus the condition holds for all members of \mathcal F.
This can be used for example to prove that a function is measurable if the preimages of all sets in a generator collection belong to the domain sigma algebra.
1.2.4 Borel sigma algebra
1.2.4.1 Examples of Borel sets
- Singletons
- [a, b], (a,b), (a,b] and [a,b) where a, b \in \bar{\mathbb R}
1.3 Measurable functions
1.3.1 Operations that preserve measurability
- Supremum, infimum of countable collections of functions
- Limit superior, limit inferior of countable sequences of functions
- Composition of measurable functions
- Concatenation (“product”) of measurable functions
1.3.2 Useful results on measurability
Continuous functions are measurable: a function f: S \rightarrow T between two topological spaces (S,\tau_S) and (T,\tau_T) is continuous iff for every A \in \tau_T, f^{-1} A \in \tau_S.
A function f: E \rightarrow \bar{\mathbb R} is measurable iff f^+(x) = \max (0, f(x)) and f^-(x) = \max (0, -f(x)) are both measurable.
Simple functions are measurable: a function f: E \rightarrow \bar{\mathbb R} is simple iff it can be expressed in the form \sum_{i=1}^n a_i \mathbb 1 (x \in A_i) where a_i \in \mathbb R, n is finite and (A_i)_{i=1}^n \subset \mathcal E.
Any non-negative function f: E \rightarrow \mathbb R_+ is measurable iff it is the pointwise limit of a monotone increasing sequence of simple functions.
1.3.3 Techniques for proving measurability
Non-obvious ways of proving a function f: E \rightarrow F is \cal E \backslash F measurable.
- Show that f is the limit of a sequence of simple functions.
- Show that for all sets A \in \mathcal C such that \sigma \mathcal C = \mathcal F, f^{-1} A \in \mathcal E.