9a)

Let E be measurable. Then, $ \exists $ an $ F_{\sigma} $ set B and a set of measure zero Z such that

$ E = B\cup Z $ (disjoint union).

Then $ F(E)=F(B)\cup F(Z) $.

Now F is A.C., thus, F(Z) has measure zero. To see this, realize if $ \epsilon > 0 $ then let $ \delta $ be number corresponding to $ \epsilon $ in the definition of A.C. Take a collection of open intervals covering Z the sum of whose lenghts is $ <\delta $, then the sum of the lengths of the image of these intervals under F will be $ < \epsilon $.

Also, since B is an $ F_{\sigma} $ it is a countable union of closed sets. Each closed set can be written as a countable union of compact sets, so B can be written as the countable union of compact sets. That is,$ B = \cup_n C_n ; \ C_n $ are compact.

Then $ F(B) = F( \cup_n C_n) = \cup_n F(C_n) $ which is the countable union of compact hence closed sets, so F(B) is an $ F_{\sigma} $.

So F(E) is the union of a two measurable sets, hence measurable.


9b)

Let $ [a,b] \subset [0,1] $. Since F is continuous, $ F([a,b]) $ is compact, thus $ \exists \alpha , \beta \in [a,b] $ such that $ F(\alpha) \leq F(x) \leq F(\beta) \ \forall \ x \in [a,b] $ or $ F(\alpha) \geq F(x) \geq F(\beta) \ \forall \ x \in [a,b] $ .


Then $ m(F(E))\leq |F(\alpha)-F(\beta)| = |\int_{\alpha}^{\beta} f(t) dt| \leq \int_{\alpha}^{\beta} |f(t)| dt = \int_{[a,b]} |f(t)| dt $ So the claim holds for a compact interval $ [a,b]. $

Now let G be any open set. Then we can write G as a countable union of nonoverlapping compact intervals:

$ G = \cup_{n=1}^\infty [a_n,b_n] $

Then $ m(F(G)) = m(F(\cup_{n=1}^\infty [a_n,b_n]))=m(\cup_{n=1}^\infty F([a_n,b_n])) $

$ = \sum_n m(F([a_n,b_n])) \leq \sum_n \int_{[a_n,b_n]} |f(t)| dt = \int_G |f(t)| dt $ since the intervals are nonoverlapping.

Finally for any E, let $ G_k $ be a collection of open sets $ \searrow E $. Then $ F(G_k) \searrow F(E)\Rightarrow m(F(G_k)) \searrow m(F(E)) $ so,

$ m(F(E)) = lim_{k \rightarrow \infty} m(F(G_k)) \leq lim_{k \rightarrow \infty} \int_0^1 |f(t)|\chi_{G_k} dt = \int_0^1 lim_{k \rightarrow \infty} |f(t)|\chi_{G_k} dt = \int_0^1 |f(t)|\chi_{E} dt = \int_E |f(t)| dt \Rightarrow $ claim.


Dat & Ben W

Alumni Liaison

Correspondence Chess Grandmaster and Purdue Alumni

Prof. Dan Fleetwood