Measure Theory-5: Operations on Measurable functions

Course 4.1
May 2, 2022

Theorem 1. Let f and g be measurable real valued functions on E, and c be a constant. Then
each of the following functions is measurable on E:
(a) f ±c
(b) cf

(c) f +g
(d) f −g
(e) |f |
(f) f2

(g) fg
(h) f /g (g vanishes nowhere on E).
Proof. Let α be an arbitrary real number.
(a) Since f is measurable and
E(f ± c > α ) = E(f > α ∓ c), the function f ± c is measurable.
(b) Assume that c 6= 0 since in case c = 0, the measurability of the function cf follows immediately.
The measurability of cf for c 6= 0 follows from the obvious relation:
(
E(f > α /c) if c > 0
E(cf > α ) =
E(f < α /c) if c < 0

(c) Consider the set

E(f + g > α ) = {x ∈ E : f (x) > α − g(x)}.
Since g is a measurable function, α − g is measurable in view of (a) and (b). Hence f + g is
measurable in view of the previous theorem.
(d) It is obvious in view of (b) and (c), since
f − g = f + (−g).
(e) Observe that (
E if α < 0
E(|f | > α ) =
E(f > α ) ∪ E(f < −α ) if α ≥ 0
Both the sets on the right-hand side are measurable and hence |f | is measurable.
(f) From the relation
(
2 E if α < 0
E(|f | > α ) = √
E(|f | > α ) if α ≥ 0

and using (e), it follows that f 2 is a measurable function.

1
 1 2 2
(g) This follows from the identity f g = 4 [(f +g) −(f −g) ], in view of (b), (c), (d) and (f) above.

(h) In view of the identity

f 1
=f
g g
6 0 on E, is
and (g) above, it is sufficient to prove that 1/g, where g = measurable. We note that

1 E(g > 0)
 α =0
E( > α ) = E(g < α1 ) ∩ E(g < 0) α >0
g
E(g > 0) ∪ E(g < 0) ∩ E(g < α1 )

α <0


1
This proves that in either case g is a measurable function. The proof of the theorem is now
complete.
Remark 2. The results in the above theorem hold good for extended real-valued functions, except
that f + g is not defined when f = ∞ and g = −∞ or vice-versa, for then,

∞
[ \ [
E(f + g > α ) = [E(f > ri ) E(g > α − ri )] [E(f = ∞)∆E(g = −∞)]
i=1

which is a measurable set.

The converse of Theorem (e) is not true as can be seen from the following example.
Example 3. Let P be a non-measurable subset of E = [0, 1). Define a function f : E → R by
(
1 if x ∈ P
f (x) =
−1 if x 6∈ P .

The function f is not measurable, since E(f > 0) (=P) is a non-measurable set. But |f | is
measurable as the set (
E if α < 1
E(|f | > α ) =
∅ if α ≥ 1
is measurable. Now we define two new functions from a pair of existing functions and show that
thy are also measurable. This shows that we can generate new measurable functions from existing
ones.
Definition 4. Let f1 and f2 be real-valued functions with common domain E. Then the functions
f ∗ = max(f1 , f2 ) and f∗ = min(f1 , f2 ) are defined to be the real-valued functions on E, where
values at any point x ∈ E are given by
f ∗ (x) = max(f1 (x), f2 (x))
and
f∗ (x) = min(f1 (x), f2 (x))
respectively.
Theorem 5. If f1 and f2 are measurable functions, then the functions f ∗ and f∗ are measurable.
Proof. Let α be any real number. We note that
E(f ∗ > α ) = E(f1 > α ) ∪ E(f2 > α )
E(f∗ > α ) = E(f1 > α ) ∩ E(f2 > α ).
Since f1 and f2 are measurable functions, the sets E(f1 > α ) and E(f2 > α ) are measurable.
Also, the union and intersection of two measurable sets are again measurable sets. This proves
that f ∗ and f∗ are measurable functions.

2
Definition 6. Let f be a function. Then its positive part, written f + , and its negative part,
written f − , are defined to be the non-negative functions given by

f + (x) = max(f (x), 0)

and
f − (x) = − max(−f (x), 0)
respectively.
We observe that f = f + − f − and |f | = f + + f − .

In this way one can define new functions from the old, in infinitely many different ways.
Let {fi } be a sequence of functions defined on a common domain E. Then define the functions

max(f1 , f2 , ..., fn )(x) = max{f1 (x), ..., fn (x)}

and
min(fl , f2 , ..., fn )(x) = min{f1 (x), ..., fn (x)}
for any finite n.
When n is infinite, for each x ∈ E, define the functions


sup fn (x) = sup {fn (x)} ,
n n

and

inf fn (x) = inf {fn (x)} .
n n

Similarly define,

lim sup fn = inf (sup fk ) (1)

n n k≥n
inf fn = − sup(−fn ) (2)
n n
lim inf fn = − lim sup(−fn ) = sup( inf fk ). (3)
n n n k≥n

Alternative notations for lim sup is lim and for lim inf is lim respectively.
Theorem 7. Let {fn } be a sequence of measurable functions (with the same domain E of defini-
tion). Then the functions max {f1 , f2 , ..., fn }, min {f1 , ..., fn } , supn fn , inf n fn , lim supn fn and
lim inf n fn are all measurable.
Proof. In view of the above definitions, it will suffice to prove that max {f1 , f2 , ..., fn } and supn fn
are measurable functions. Let h = max {f1 , f2 , ..., fn }. We note that,
n
[
E(h > α ) = E(fi > α ).
i=1

Measurability of each fi implies that the function h is measurable.

Similarly, by writing g as supn fn and observing that
∞
[
E(g > α ) = E(fi > α ),
i=1

the measurability of the function g follows.

Corollary 8. If {fn } is a sequence of measurable functions converging to f on E, then f is a
measurable function.

3
Proof. Since fn → f , we have,

limfn = limfn = f
In view of the above theorems, f is a measurable function.
Corollary 9. The set of points on which a sequence {fn } of measurrable functions converges is
measurable.
Proof. The required set 
x : limfn (x) = limfn (x)
is, clearly, measurable.
Note We may conclude that all operations of analysis, including limit operations, when applied to
measurable functions lead to a measurable function.

Various Measurable Functions

In this section you are going to get various important functions which are indeed the backbones
of the development of measure and integration.

Characteristic Functions
Definition 10. Let E be a given set. If A is a subset of E, then the characteristic function χA
of A is a real-valued function defined on E by
(
1 if x ∈ A
χA (x) =
0 if x 6∈ A

Remark 11. The characteristic function χA of the set A is also called the indicator function of
A. A few simple properties of a characteristic function are given in the following theorem, the
proof of which is straightforward and hence left for you.
Theorem 12. Let A and B be subsets of E. Then
(a) χ∅ = 0 and χE = l.
(b) A ⊂ B ⇒ χA ≤ χB .
(c) χA∪B = χA + χB − χA∩B
(d) χA∩B = χA χB
(e) χAc =P1 − χA
∞ S∞
(f) χA = i=1 χAi , where A = i=1 Ai and the sequence {Ai } consists of disjoint
subsets of E.
(g) χA−B = χA − χA∩B .
We define the measurability of a function f over a measurable set E. If E is a non-measurable
set, we cannot define a measurable function f on E. But it does not mean that if E is a measurable
set, the function f is always measurable. In fact, we prove the following theorem.
Theorem 13. Let E be a measurable set. Then the set A ⊂ E and its characteristic function χA
are simultaneously measurable or non-nmeasurable.
Proof. Let α be any real number. Then,

∅
 if α ≥ 1
E(χA > α ) = A if 0 ≤ α ≤ 1

E if α < 0


4
If A is a measurable set, then the set E(χA > α ) is measurable for each real number α and hence
χA is a measurable function. On the other hand, if χA is a measurable function, the set A turns
out to be measurable by choosing an α such that 0 ≤ α < 1.
Remark 14. The above theorem asserts that the characteristic function of non-measurable sets
are non-measurable even though the domain set is measurable.

Simple Functions
A very useful concept in the theory of measurable functions is that of a simple function. A function
is said to be a simple function if it assumes only a finite number of values, each of them being
finite. We put it more precisely as follows.

Definition 15. A function f : E → R is saidSto be a simple function if there is a finite disjoint

∞
class {E1 , E2 , ..., En } of measurable sets with i=1 Ei = E and a finite set {a1 , a2 , ..., an } of real
numbers such that

f (x) = ai , x ∈ Ei , 1 ≤ i < n.
Let f be a simple function as described above. Then we have,
∞
X
f (x) = ai χEi (x)
i=1

where χEi is the characteristic function of the measurable set Ei .

Example 16. (a) Each step function is a simple function.
(b) Each characteristic function of a measurable set is a simple function.
Pn
(c) Any linear combination
S of the form i=1 ai χEi represents a simple function φ defined on
the set E = Ei . (The sets Ei ’s are measurable but not necessarily disjoint.) Note that the
above representation of a simple function can be considered as a canonical representation
if each Ei ’s are disjoint and ai ’s are distinct and non zeros.

Continuous Functions
We begin with the following interesting and important theorem.
Theorem 17. A continuous function defined on a measurable set is measurable.
Proof. Let f be a function defined and continuous on E(measurable). Let α be any arbitrary real
number. Consider the set
A = {x ∈ E : f (x) ≤ α }
Let xo ∈ E be a limit point of A. Then, ∃ a sequence {xn } of points in A with xn → xo . Therefore,

f (xn ) ≤ α ∀n.

By the continuity of f at x0 , we have

f (x0 ) ≤ α
and, as such, x0 ∈ A. Therefore A is a closed set and, every closed set being the complement of
an open set is measurable since every open set is measurable. Thus A is also measurable. This
proves that f is a measurable function.
Alternative Proof:

5
Proof. Consider the set {x ∈ E|f (x) ≥ α }. Now,

{x ∈ E|f (x) ≥ α } = f −1 [α , ∞)
Now, f is continuous and so f −1 [α , ∞) is open since [α , ∞) is open in R . Thus f −1 [α , ∞) is a
Borel set and therefore is measurable. This completes the proof.
Remark 18. The converse of the above Theorem is not true; more precisely, a measurable function
need not be continuous. You can see this fact from the following example.
Example 19. Consider the function f : R → {0, 1} given by,
(
1 if 0 ≤ x < 1
f (x) =
0 otherwise

Clearly, the function f is measurable but not continuous. The point at x = 0 is a point of
discontinuity of f .

Sets of Measure zero

Sets of measure are just the sets that are negligible in the theory of Lebesque integration. However,
their uses are immense in tackling problems on measure theory.
Definition 20. A property P is said to hold good almost everywhere (abbreviated a.e.) on a set
S if the set of points of S where P fails to hold has measure zero.
Example 21.
(a) If a function f defined on a set E is discontinuous on a subset of E whose measure is zero
and is continuous elsewhere on E, then we say that the function f is continuous a.e. on E.

(b) Let f be a function defined on R by

(
0 if x is irrational
f (x) =
1 if x is rational

Then f = 0 a.e.
Definition 22. Two functions f and g defined on the same domain E are said to be equivalent
on E, written f ∼ g on E, if f = g a.e. on E ; i.e., f (x) = g(x) for all x ∈ E \ E1 where E1 ⊂ E
with m(E1 ) = 0.
Theorem 23. Let f and g be two functions defined on a common domain E such that f = g a.e.
and g is measurable. Then f is measurable.
Proof. Let E1 = {x ∈ E : f (x) = g(x)} and E2 = {x ∈ E : f (x) 6= g(x)}. Then E = E1 ∪ E2 and
m(E2 ) = 0. Let α be any arbitrary real number. Consider the set

A = {x ∈ E : f (x) > α } .

Since A ∩ E2 ⊂ E2 , it follows that m∗ (A ∩ E2 ) = 0. Also,

A ∩ E1 = {x ∈ E : g(x) > α } ∩ E1

whence A ∩ E1 is measurable.
The measurability of f follows from the relation

A = (A ∩ E1 ) ∪ (A ∩ E2 )

6
Remark 24. Let f be the Dirichlet’s function defined on R by
(
0 if x is irrational
f (x) =
1 if x is rational

Recall that f (x) = 0 a.e. Since 0 is a measurable function as a constant function therefore, f is
also measurable.
Definition 25. A sequence {fn } of functions defined on E is said to converge a.e. to a function
f if
lim fn (x) = f (x),
n→∞

for all x ∈ E \ E1 , where E1 ⊂ E with m(E1 ) = 0.

Example 26. The sequence {fn } given by fn (x) = (−1)n xn , x ∈ [0, 1] converges a.e. to the
function f ≡ 0, in fact, everywhere except at point x = 1.

Borel Measurable Functions

Following the definition of a measurable function, the Borel measurable function is defined as
follows.
Definition 27. A function f defined on a Borel set E is said to be a Borel measurable function,
or more precisely, Borel function E if the set E(f > α) is a Borel set for all real numbers α .
Therefore, every Borel measurable function is Lebesgue measurable. But the converse is not true
as can be seen from the following example.
Example 28. The characteristic function of a set which is Lebesaue measurable and non-Borel
is Lebesgue measurable but not a Borel measurable function.
Remark 29. If we replace a Lebesgue measurable set by a Borel set and a Lebesgue measurable
function by a Borel measurable function we can prove the following results.
Theorem 30. Let f be an extended real-valued function defined on a Borel set E (of finite or
infinite measure). Then the following statements are equivalent:
(a) E(f > α) is a Borel set for all α ∈ R .
(b) E(f ≥ α) is a Borel set for all α ∈ R .
(c) E(f < α) is a Borel set for all α ∈ R .
(d) E(f ≤ α) is a Borel set for all α ∈ R .

Remark 31. Theorem 23 cannot be extended to Borel measurable functions.

Theorem 32. Let f and g be Borel measurable functions on E, and let c be any constant. Then
each of the following functions is Borel measurable on E:
(a) f ±c
(b) cf
(c) f +g
(d) f −g
(e) |f |
(f) f2
(g) fg
(h) f /g (g vanishes nowhere on E).