Homepage Solution manuals Terence Tao An Introduction to Measure Theory Exercise 1.4.29 (Properties of measurable functions)

Exercise 1.4.29 (Properties of measurable functions)

Let (X,) be a measurable space. Show that

(i)
a function f : X [0,+] is measurable if and only if the level sets {x X : f(x) > λ} are -measurable.
(ii)
the indicator function 1E : X {0,1} is measurable if and only if E X is -measurable.
(iii)
a function f : X is measurable if and only if for every Borel measurable subset E the set f1(E) is -measurable.
(iv)
a function f : X is measurable if and only if its real and imaginary parts are measurable.
(v)
a function f : X is measurable if and only if its positive and negative magnitudes are measurable.
(vi)
if (f )n : X is a sequence of measurable functions converging to another function f : X , then f is also measurable.
(vii)
if f : X is measurable, and ϕ : is continuous, then ϕ f is measurable.
(viii)
the sum of product of two measurable functions is measurable.

Answers

(i)
If f is measurable, then the sets { x X : f ( x ) > λ } = f 1 ( ( λ , + ] ) must be measurable too since the latter is the inverse image of an open set in [ 0 , + ] .
Conversely, suppose that for any λ > 0 the set f 1 ( ( λ , + ] ) is B -measurable. Then also
  • sets of the form f 1 ( [ , λ ] ) = f 1 ( ( λ , + ] ) = X f 1 ( ( λ , + ] ) must be measurable.
  • sets of the form f 1 ( [ , λ ) ) = f 1 ( n = 1 [ , λ + 1 n ] ) = n = 1 f 1 ( [ , λ + 1 n ] ) must be measurable.
  • sets of the form f 1 ( a , b ) = f 1 ( a , + ] f 1 [ , b ) are measurable.

Pick an arbitrary open set U [ 0 , + ] . Then U is a countable union of open intervals n I n (cf. Analysis I). Recall that an open interval of the extended positive real line either contains + (in which case it can be written as ( a , + ] or [ 0 , + ] for a [ 0 , + ] ) or it does not (in which case it can be written as ( a , b ) for a , b [ 0 , + ] ). Both of the cases result in measurable inverse images under f by the previous argument. We then have f 1 ( n I n ) = n f 1 ( I n ) , and the last term is measurable as a countable union of measurable sets.

(ii)
Suppose the 1 E is measurable. Then by definition E = ( 1 E ) 1 ( { 1 } ) = ( 1 E ) 1 ( n = 1 ( 1 n + 1 , 1 + 1 n ) ) = n = 1 ( 1 E ) 1 ( 1 n + 1 , 1 + 1 n )

and the last term is measurable as a countable intersection of measurable sets.suppose conversely that E is measurable. Now pick an arbitrary open set U [ 0 , + ] . Notice that only the values 0 , 1 [ 0 , + ] get mapped into. Thus, we have four cases:

  • 0 U , 1 U , then f 1 ( U ) = f 1 ( { 0 } ) f 1 ( U { 0 } ) = X E = X E which is measurable.
  • 0 U , 1 U . Then f 1 ( U ) = f 1 ( { 1 } ) f 1 ( U { 1 } ) = E = E which is measurable.
  • 0 , 1 U or 0 , 1 U , in both cases the inverse image f 1 ( U ) is measurable as it is either or X respectively.
(iii)
Suppose that the inverse image of every Borel measurable set in is B -measurable. Then, in particular the inverse image of every open set must me B -measurable too, and we are done.suppose conversely that f is measurable. We use the σ -induction from the Remark 1.4.15.
  • f 1 ( ) = is B -measurable.
  • If f 1 ( E ) is B -measurable, then f 1 ( E ) = X f 1 ( E ) is B -measurable, too.
  • If f 1 ( E n ) is B -measurable for ( E ) n , then f 1 ( n = 1 E n ) = n = 1 f 1 ( E n ) is B -measurable, too.
  • Finally, the property holds for every element of the generating set O ( ) of open sets.

Thus, this property must hold for all elements of σ ( O ( ) ) too.

(iv)
Suppose that f = [ ℜf , ℑf ] is measurable. We verify that the real part ℜf : X is measurable. Pick an arbitrary open set U (for the real part). Then the inverse image is ( ℜf ) 1 ( U ) = f 1 ( U × ) . The latter is a B -measurable set since it is an inverse image of a product of two open sets which is open (cf. Topology). Similarly the imaginary part is measurable as well.suppose that the real and the imaginary part of f are measurable. Pick an arbitrary open set U . Let c U . Notice that f 1 ( c ) = ( ℜf ) 1 ( ℜc ) ( ℑf ) 1 ( ℑc ) = ( f 1 1 ) ( c ) ( f 1 1 ) ( c )

This is easy to verify. (If x f 1 ( c ) then f ( x ) = c and therefore ℜf ( x ) = ℜc hence f ( x ) 1 ( ℜc ) and x f 1 1 ( ℜc ) , similarly with the imaginary part. Conversely, x f 1 1 ( ℜc ) means ℜf ( x ) = ℜc and x f 1 1 ( ℑc ) means ℑf ( x ) = ℑc . Since f ( x ) = ℜf ( x ) + ℑf ( x ) = c , the result follows.) But both terms are B -measurable by assumption, and so is their intersection.

(v)
Let f : X be measurable, and consider f + , f : X [ 0 , + ) given by f + = max { f , 0 } and f = min { f , 0 } . Then any open set U [ 0 , + ] we have ( f + ) 1 ( U ) = f 1 ( U ) is B -measurable, and ( f ) 1 ( U ) = f 1 ( U ) which is also B -measurable, since U is open.suppose that f + , f : X [ 0 , + ) given by f + = max { f , 0 } and f = min { f , 0 } are measurable. Pick an arbitrary open set U . We than have f 1 ( U ) = f 1 ( [ , 0 ] U ) f 1 ( U [ 0 , + ] ) = ( f ) 1 ( [ , 0 ] U ) ( f + ) 1 ( U [ 0 , + ] ) = ( f ) 1 ( U [ 0 , + ] ) ( f + ) 1 ( U [ 0 , + ] )

Both sets are open in [ 0 , + ] , and the union of B -measurable inverses is again B -measurable.

(vi)
We argue that f 1 ( E ) is a countable union/intersection of B -measurable sets. In particular, f 1 ( U ) = n = 1 N = n f N 1 ( U ) .

  • Pick an arbitrary x f 1 ( U ) . Then f ( x ) U is an interion point, and we thus can find an 𝜖 > 0 such that B ( f ( x ) , 𝜖 ) U . Since ( f ) n f , we can find a n such that N > n we have f N ( x ) B ( f ( x ) , 𝜖 ) and thus f N ( x ) U . This implies x n = 1 N = n f N 1 ( U )
  • Pick an arbitrary x n = 1 N = n f N 1 ( U ) , i.e., there exists a n such that x N = n f N 1 ( U ) , i.e., for all N n we have f N ( x ) U . Furthermore, there exists a n ′′ such that for all N n ′′ we have d ( f N ( x ) , f ( x ) ) 𝜖 . Thus, after n = max { n , n ′′ } the values f N ( x ) are both 𝜖 -close to f ( x ) and contained in U . By triangle inequality f ( x ) must be contained in an 𝜖 -ball in U as well.
(vii)
The inverse image ( ϕ f ) 1 ( U ) = f 1 ϕ 1 ( U ) is measurable, since U and therefore also ϕ 1 ( U ) are open, and f is measurable.
(viii)
Let f , g : X be measurable. Consider two further functions l : X 2 , x [ f ( x ) , g ( x ) ] ϕ : 2 , [ c 1 , c 2 ] c 1 + c 2

Then l is measurable. To see why, pick an arbitrary open U 2 . Then l 1 ( U ) = ( π 1 l ) 1 ( π 1 U ) ( π 2 l ) 1 ( π 2 U ) where π i is the projection function in the i -th co-ordinate. But π i l is a composition of a continuous function with a measurable function, and is thus by part (vii) itself measurable. Taking the intersection of B -measurable sets we see that l is measurable. ϕ is continuous (cf. Analysis II). Thus, by part (vii) the function f + g = l ϕ is measurable.
A similar argument proves that the product is measurable as well.

User profile picture
2020-08-07 00:00
Comments