Homepage › Solution manuals › Terence Tao › An Introduction to Measure Theory › Exercise 1.4.8 (Cardinality of the generated Boolean algebra)
Exercise 1.4.8 (Cardinality of the generated Boolean algebra)
Let be a collection of sets with . Show that is a finite Boolean algebra of cardinality at most . Show that this bound is the best possible.
Answers
Suppose that is a finite collection of subsets of a set , and let be an arbitrary Boolean algebra containing . By the properties of the intersection we have . In the following we induct on .
-
(induction base) Let , i.e., for some set . It is easy to verify that the set
is a Boolean algebra on , and it contains . Thus, .
-
(induction step) Now let (obviously all elements assumed to be distinct), and suppose inductively that we have proven this statement for some . By induction hypothesis we can find a Boolean algebra for the set of the cardinality , and we can find a Boolean algebra for the set also of the cardinality . The crucial step here is to employ Exercise 1.4.4 in order to use the fact that both sets can be generated using some atomic algebras or of the size bounded by . By Example 1.4.7 (Atomic algebra), let be the Boolean algebra generated by . Then is a Boolean algebra containing ( is contained in or and can thus be generated using one of the collections or ). We hence have
To demonstrate that this bound is sharp, consider the discrete set . Then, there exists a family of subsets such that . We prove the existence by inducting on .
-
(induction base) Let , i.e., . Set . Let be an arbitrary Boolean algebra containing . Then all of the atom-elements of are contained in . This is demonstrated by the closure under intersection and complements property of the Boolean algebra:
In other words, . Therefore, by the closure under unions property of the Boolean algebra, we also have . Since our choice of was arbitrary, taking intersections we obtain and therefore . Combining this with the previosly proven assertion, we see the sharpness of the bound .
-
(induction step) Now suppose inductively that there exists a family of subsets of with and . Now consider . Recall that we can bijectively associate with two subsets of : either those ending with 0, or those ending with 1:
Thus, consider extended collections
It is easy to verify that where . Similarly, we must now have and thus . But this implies as desired.
Comments
For the concrete bound:
The set has elements. Let be the set of singleton elements, which is an atomic base set for , generating the discrete algebra. Then by MTE4.4, the discrete algebra consists of sets.