Ch1 Flashcards
(16 cards)
Sigma Algebra
1) empty set in F; 2) if A\in F, A^c\in F; 3) countable union of sets is in F
Measure Space
Pair (\Omega, F) where F is the sigma algebra.
sigma algebra generated by A
Let A\subseteq \Omega, the smallest sigma algebra containing A
Borel sigma-algebra
Omega needs to be endowed with a topology. Then the borel sigma algebra is the sigma algebra generated by open sub sets of Omega.
Borel Sigma Algebra on R
Also generated by all open, half-open, closed, and (-\infty,x) intervals.
Restrictions on Algebras
Let (\Omega, F) be a measurable space. Let \Omega’\in F, then the restriction F’={A\cap \Omega’:A\in F} is a sigma algebra, and generated by A’={A\cap\Omega’:A\in scriptA} if scriptA generated F.
Measures
Measures are defined on measurable spaces, and have the following two properties: 1) \mu(\emptyset)=0; 2) countable additivity (sum of disjoint equals measure of union)
Pi system
Let \Omega be a set. P is a pi-system if it closed under finite intersection.
Lambda system
Let \Omega be a set. L is a lambda system if L is a collection of \Omega subsets and satisfies: 1) \emptyset \in L; 2) complements in L; 3) union of countable disjoint is in set.
Dynkin’s Pi-Lambda theorem
Let \Omega be a set. If there is a P pi-system contained within a L lambda-system, then the sigma algebra generated by P is within L.
When are measures unique?
if \mu_1 and \mu_2 are both measures that agree on P, and there is an increasing sequence of sets in P that converge to \Omega, then they also agree on \sigma(P).
Outer Measure vs. Measure
1) Defined on power set; 2) \phi(union) \leq sum \phi(A-i)
B is phi measurable if
phi(A)=phi(A\cap B)+phi(A\cap B^c)
Sigma Algebra and Phi Measurability
Let Omega be a set, and F be the set of phi-measurable sets in Omega. Then F is a sigma algebra, and phi is the measure on (Omega, F).
Caratheodory’s Extension Theorem
Let A be an algebra on Omega, and let there be a measure on A. Then mu has an extension on the sigma algebra generated by the algebra, and moreover, if mu is sigma finite, the measure is unique.
In particular, we define the extension as the infimum of the cover over A, where A_i is taken from the algebra which approximates A.
Lebesgue Measure
For any half open interval (a,b], define the Lebesgue measure in the usual sense. It has a unique extension on the Borel Sigma algebra on R, which is generated by the collection of scriptA={(a,b]:a<b></b>