Homepage Solution manuals Terence Tao An Introduction to Measure Theory Exercise 1.7.19 (Examples of product $\sigma$-algebras)

Exercise 1.7.19 (Examples of product $\sigma$-algebras)

(i)
Show that the product of two trivial σ-algebras (on two different spaces X,Y ) is again trivial.
(ii)
Show that the product of two finite σ-algebras is again finite.
(iii)
Show that the product of two Borel σ-algebras (on two Euclidean spaces Rd,Rd with d,d 1) is again the Borel σ-algebra (on Rd × Rd Rd+d ).
(iv)
Show that the product of two Lebesgue σ-algebras (on two Euclidean spaces Rd,Rd with d,d 1) is not the Lebesgue σ-algebra.
(v)
However, show that the Lebesgue σ-algebra on Rd+d is the completion of the product of the Lebesgue σ-algebras of Rd and Rd with respect to d + d-dimensional Lebesgue measure.
(vi)
This part of the exercise is only for students who are comfortable with cardinal arithmetic. Give an example to show that the product of two discrete σ-algebras is not necessarily discrete.
(vii)
On the other hand, show that the product of two discrete σ-algebras 2X,2Y is again a discrete σ-algebra if at least one of the domains X,Y is at most countably infinite.

Answers

(i)
Product of two trivial σ-algebras (on two different spaces X,Y ) is again trivial.
Using the generating set from the previous exercise, we case-by-case verify that {×,× Y,X ×,X × Y } = {,X × Y }.

By Exercise 1.4.9 (Recursive description of a generated Boolean algebra), we know that this equals to {,X × Y }, which is a trivial algebra.

(ii)
Product of two finite σ-algebras is again finite.
Let n := #BX and m := #BY . Then {E × FE BX,F BY } nm and we have BX ×BY = {E × FE BX,F BY } 2{E×FEBX,FBY }

and the cardinality of the right hand side is less than or equal to 2nm.

(iii)
Product of two Borel σ-algebras (on two Euclidean spaces Rd,Rd with d,d 1) is again the Borel σ-algebra (on Rd × Rd Rd+d ).
We demonstrate that B[Rd] ×B[Rd ] = B[Rd+d ].
  • Let E be a Borel measurable subset of d, and let F be a Borel measurable subset of d . By Exercise 1.4.17, E × F is a Borel measurable subset of d+d .
  • Let E be a Borel measurable subset of d+d . By Exercise 1.4.18, for any x1 d , the slice {x2 d : (x1,x2) E} is a Borel measurable subset of d .

Thus, every pair of sets in B[Rd] and B[Rd ] can be associated with a set in B[Rd+d ], and vice versa.

(iv)
Product of two Lebesgue σ-algebras (on two Euclidean spaces Rd,Rd with d,d 1) is not the Lebesgue σ-algebra.
Consider two one-dimensional Lebesgue σ-algebras L[R] and the Borel σ-algebra L[R2]. Since a Lebesgue σ-algebra is always complete, it must in particular contain the null set {0}× V , where V denotes the Vitali set. However, no two sets in R can result in this product since V is non-measurable. Thus, we have found a set {0}× V L[R2] such that {0}× V L[R] ×L[R].
(v)
The Lebesgue σ-algebra on Rd+d is the completion of the product of the Lebesgue σ-algebras of Rd and Rd with respect to d + d-dimensional Lebesgue measure.
  • Obviously L[d] ×L[d ] L[d+d ] since the former is the coarsest σ-algebra on d+d containing the product sets and L[d+d ] contains all product sets by Exercise 1.2.22. Furthermore, Lebesgue σ-algebra is complete; thus, the completion of L[d] ×L[d ] must be contained in L[d+d ] as well.
  • Recall that by Exercise 1.4.19, the Lebesgue σ-algebra is generated by the union of the Borel σ-algebra and the Borell null-algebra. Thus, by part (iii) of this exercise, we obtain

    L[Rd+d ] = B[Rd+d ]¯ = B[Rd] ×B[Rd ]¯.

    But B[Rd] ×B[Rd ] L[Rd] ×L[Rd ], and so we are done.

(vi)
The product of two discrete σ-algebras is not necessarily discrete.
this part is yet to be solved by a student who is comfortable with cardinal arithmetic.
(vii)
The product of two discrete σ-algebras 2X,2Y is again a discrete σ-algebra if at least one of the domains X,Y is at most countably infinite.
this part is yet to be solved by a student who is comfortable with cardinal arithmetic.
User profile picture
2021-09-13 00:00
Comments