Mathematics Department Stanford University

Math 171 Lecture Supplement

The Riemann and Lebesgue Integrals
L EON S IMON
A UTUMN Q UARTER , 2011

T HESE NOTES ARE MEANT AS A QUICK INTRODUCTION TO THE R IEMANN INTEGRAL USING STEP FUNCTION TERMI -
NOLOGY, FOLLOWED BY AN ALMOST- AS - QUICK INTRODUCTION TO THE L EBESGUE INTEGRAL , ALSO VIA STEP FUNC -
TIONS . W E INCLUDE A PROOF OF L EBESGUE ’ S THEOREM WHICH PRECISELY CHARACTERIZES THOSE FUNCTIONS
WHICH ARE R IEMANN INTEGRABLE .

Contents
1 Preliminaries: Step Functions . . . . . . . . . . . . . . . . . . . . . . 1
2 Riemann Integrable Functions . . . . . . . . . . . . . . . . . . . . . 3
3 Lebesgue measure zero . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
4 Definition and Properties of the Lebesgue Integral . . . . 8

1 Preliminaries: Step Functions

Here we work in Rn , where n  1 is given. By an interval I in Rn we mean the cartesian
product I1  I2 


 In , where each Ij is an interval in R, thus each Ij is one of the fol-
lowing: Œaj ; bj ; .aj ; bj /; Œaj ; bj /; .aj ; bj , where aj  bj are real. The volume jI j of an inter-
val I D I1  I2 


 In is of course defined to be the product of the edge lengths. Thus
jI j D .b1 a1 / 


 .bn an / in case Ij has end-points aj ; bj with aj  bj , and the volume
is zero if I is a degenerate interval (i.e. if aj D bj for some j ).
We say the interval I is closed if each factor Ij is a closed interval; thus a closed interval in Rn is
one that can be written Œa1 ; b1  


 Œan ; bn . Similarly, an open interval is one that can be written
I D .a1 ; b1 / 


 .an ; bn / (D ∅ if aj D bj for some j ).
From now on R D Œa1 ; b1  


 Œan ; bn  will be a fixed closed interval in Rn .
Definition: A partition P of R is the collection of closed intervals I  R obtained by partitioning
each of the edges of R; thus for each j D 1; : : : ; n we select points aj D tj;0 < tj;1 <


< tj;Nj D bj
and then P D {Œt1;i1 1 ; t1;i1   Œt2;i2 1 ; t2;i2  


 Œtn;in 1 ; tn;in  W ij 2 {1; : : : ; Nj } for each j D
1; : : : ; n}. The points tj;0 ; : : : ; tj;Nj are called “the j ’th edge points” of the partition P . For any
I D Œt1;i1 1 ; t1;i1   Œt2;i2 1 ; t2;i2  


 Œtn;in 1 ; tn;in  2 P we let IM denote the corresponding open
rectangle .t1;i1 1 ; t1;i1 /  .t2;i2 1 ; t2;i2 / 


 .tn;in 1 ; tn;in /, and @I D I n IM.
Definition: ' is a step function on R if it is bounded and there is a partition P of R such that for
each I 2 P there is a real constant aI such that '  aI on IM; thus
X
1:1 'D aI IM on [I 2P IM;
I 2P

1
2 M ATH . 171 L ECTURE S UPPLEMENT 2011

where we use the notation that, for any set A  Rn , A denotes the indicator function of A; thus
A .x/ D 1 if x 2 A and A .x/ D 0 if x 2 Rn n A.
Notice that in 1.1 we make no particular restriction on the values taken by ' on [I 2P @I beyond the
fact that '.x/ should be a real number for each x 2 [I 2P @I and 'j.[I 2P @I / should be bounded.
If ' and are two such step functions on R, say with ' D I 2P aI IM on [I 2P IM and
P
D
M
P
J 2Q bJ JM on [J 2Q J , then the sum and difference is also a step function, as are max{'; },
min{'; }. One checks this by taking a common refinement R of the partitions P; Q (thus R is
a partition of R with j ’th edge points including all the j ’th edge points of P and all the j ’th edge
points of Q, j D 1; : : : ; n); then
X X
1:2 'D eI IM and
a D beI IM on [I 2R IM
I 2R I 2R

for suitable aeI ; beI (in fact aeJ D aI whenever J 2 R with J  I and I 2 P and beJ D bI whenever
J 2 R with J  I and I 2 Q; this makes sense because every J in the common refinement R is
contained in a unique I 2 P and a unique Ie 2 Q). Evidently it is also true that any scalar multiple
of a step function is also a step function (indeed any product of step functions is a step function
by 1.2) so in particular we have shown that the step functions form a subspace of the vector space
of functions f W R ! R. The vector space of step functions on R will be denoted
S.R/:

Naturally the integral of a step function ' as in 1.1 is defined by

Z X
1:3 'D aI jI j;
R I 2P

where jI j is the volume of I (i.e. the product of the edge lengths of I ). Of course this definition
is independent of the particular representation of step function used; thus if instead of 1.1 we use
a different partition Q of R and if 'j [J 2Q JM D J 2Q bJ JM , then of course it turns out that
P
P P
I 2P aI jI j D J 2Q bJ jJ j; one can check this directly by using a common refinement R of P; Q
as in 1.2 above, as follows: Since each K 2 R is contained in a unique I 2 P and '  cK on KM ,
where cK D aI , we have
X X X X
cK jKj D aI jKj D aI jI j;
K2R I 2P K2K with KI I 2P

and similarly X X
cK jKj D bJ jJ j;
K2R J 2Q

so as claimed X X
aI jI j D bJ jJ j:
I 2P J 2Q

Using the formulae 1.2 it is evident that the integral so defined is linear on the step functions; that
is, if '; are both step functions on R and if ˛; ˇ are real constants then
Z Z Z
1:4 .˛' C ˇ / D ˛ 'Cˇ ; '; 2 S.R/:
R R R
 R IEMANN & L EBESGUE I NTEGRALS 3

It is also clear from the representation 1.2 that

Z Z
1:5 '; 2 S.R/ with   )  :
R R

Observe that
X Z
1:6 aI I 2 S.R/ on [I 2P IM with 0)˛
P
D {I 2PWaI ˛ } jI j 
I 2P R

for each ˛ > 0, which follows directly from 1.5 because ˛

P
{I 2PWaI ˛ } IM  .

2 Riemann Integrable Functions

Given a bounded function f W R ! R we can define the upper and lower Riemann integral of f ,
R R
denoted R f and R f respectively, by
Z Z
f D inf '
R '2S.R/;f ' R
Z Z
f D sup '
R '2S.R/;'f R

R R
and f is said to be Riemann integrable if R f D R f . Of course using 1.5 and the definitions we
have Z Z
f  f
R R

in any case, so to prove f is Riemann integrable it suffices to prove the reverse inequality.

We begin with a convenient criterion for checking Riemann integrability:

Lemma 1: Suppose f W R ! R is bounded. Then f is Riemann integrable ” for each given " > 0
R R R
there are step functions '; with '  f  on R and R < R ' C " (i.e. R . '/ < ").

Proof “(”: Let " > 0 be given and choose step functions '; with '  f  on R and
R R
R < R ' C ". Then
Z Z Z Z Z
' f  f   ' C ";
R R R R R
R R R R
whence R f R f < ", and hence (since this holds for each " > 0) R f  R f , and so f is
Riemann integrable as claimed.
R R R
Proof “)”: Suppose f is Riemann integrable. Then R f D R f D R f , and hence there are
R R R R
step functions '; with '  f  and R ' > R f "=2 and R  R f C "=2, and so
R R
R R ' < "=2 C "=2 D ".

We now check a few general facts about the set of Riemann integrable functions.
4 M ATH . 171 L ECTURE S UPPLEMENT 2011

Theorem 1: Every continuous function f W R ! R is Riemann integrable.

Proof: Let " > 0. R is compact and a continuous function on a compact set is uniformly contin-
uous, so there is ı > 0 such that jf .x/ f .y/j < " whenever x; y 2 R with jx yj < ı . So take a
partition P of R such that each I 2 P has diameter < ı , and observe that then maxI f  minI f C"
for each I 2 P . Define step functions '; on R as follows:

'jIM D min f; jIM D max f; I 2P

I I

and take ' D minR f and D maxR f on the remaining points of R (i.e. on [I 2P @I ). Then
Z Z X X
'f  and 'D .max f min f /jI j  " jI j D "jRj;
R R I I
I 2P I 2P

and hence f is Riemann integrable by Lemma 1.

Theorem 2: The set of Riemann integrable functions on R is a vector space and the Riemann integral is
a linear operator on that vector space (i.e. f; g Riemann integrable on R and ;  2 R ) f C g is
R R R
also Riemann integrable, and R .f C g/ D  R f C  R g ).
Proof: Let R.R/ be the set of Riemann integrable functions on R. It suffices to prove
Z Z Z
.1/ f; g 2 R.R/ ) f C g 2 R.R/ and .f C g/ D f C g
R R R

and
Z Z
.2/ f 2 R.R/ and  2 R ) f 2 R.R/ and f D  f:
R R

To prove (1), let " > 0 and, using Lemma 1, pick step functions '; ; 'e; e such that '  f 
R R
; 'e  g  e , and R . '/ < "=2; R . e 'e/ < "=2, and so by virtue of the linearity 1.4 we have
Z Z
. C e/ .' C 'e/ < "=2 C "=2 D ";
R R

and
' C 'e  f C g  C e;
R R
so we deduce that f C g is Riemann integrable by virtue of Lemma 1, and f C Rg "  R
R R R R R R R
R. C e/ "  R .' C 'e/  R .f C g/  R . C e /  R .' C 'e/ C "  R f C R g C ", so
Z Z Z Z Z
f C g "  .f C g/  f C gC"
R R R R R
R R R
for each " > 0, whence R .f C g/ D R f C R g .
The proof of (2) is left as an easy exercise, again based on Lemma 1 and the linearity 1.4 (one has
to break consideration into the cases  > 0;  D 0;  D 1).
The Riemann integral also preserves ordering of functions as follows:
 R IEMANN & L EBESGUE I NTEGRALS 5

Theorem 3: Suppose f; g are Riemann integrable on R and f  g (meaning f .x/  g.x/ 8x 2 R).
R R
Then R f  R g .
R R R R
Proof: Let " > 0. Since R f D R f we can find a step function ' with '  f and R f  R ' C",
R R R R
and since R g D R g we can find a step function with g  and R  R g C ". Then
R R R R R R
'  f  g  and by 1.5 R f  R ' C "  R C "  R g C " C "; that is R f  R g C 2",
R R
and since " is arbitrary this implies R f  R g .

3 Lebesgue measure zero

A key difference between Lebesgue theory and the theory of volume related to the Riemann integral
is in the notion of measure zero. In the Riemann theory a set A is said to have volume zero (or
“Jordan content zero”) if for each " > 0 there are open intervals I1 ; : : : ; IN with A  [jND1 Ij and
PN
j D1 jIj j < ". In the Lebesgue theory, we allow infinitely many intervals:

Definition: A set A  Rn is said to have Lebesgue measure zero if for each " > 0 there are open
intervals I1 ; I2 ; : : : such that A  [j1D1 Ij and j1D1 jIj j < ".
P

This makes a profound difference in the related theory; for example we have the following lemma:
Lemma 2. Suppose A1 ; A2 ; : : : is a sequence of subsets of Rn such that each Aj has Lebesgue measure
zero. Then [j1D1 Aj also has Lebesgue measure zero.
Proof: Let " > 0. For each j we can select open intervals Ij;1 ; Ij;2 ; : : : with Aj  [1 I and
kD1 j;k
P1 j 1
P1 j
kD1 jIj;k j < 2 ". Then [j D1 Aj  [j;k Ij;k and j;k jIj;k j < j D1 2 " D "
P

The following terminology is commonly used in the literature, and we will freely use it in subse-
quent discussion here:
Terminology: We say that a property holds “almost everywhere” or “for almost every point” (ab-
breviated “a.e.”) in a subset   Rn if there is a set S   of Lebesgue measure zero such that the
property in question holds at each point of  n S .
Thus for example fk .x/ ! 0 a.e. x 2  means that there is a set S   with Lebesgue measure
zero such that lim fk .x/ D 0 for every x 2  n S .
As our first results which depend intrinsically on the concept of measure zero, we have two lemmas
concerning sequences of step functions. The first lemma relates to decreasing sequences of non-
negative step functions:
R
Lemma 3. Suppose 'k 2 S.R/ with 0  'kC1  'k 8 k D 1; 2; : : :. Then R 'k ! 0 ” 'k .x/ ! 0
a.e. x 2 R.
Proof of “)”: Let ˛ > 0 and S D {x 2 R W lim 'k .x/  ˛ }. Then S  {x 2 R W 'k .x/  ˛ } for
R
each k , and we can apply 1.6 with D 'k , thus giving partitions P k such that ˛ I 2F k jI j  R 'k
P

for some subcollection F k  P k with 'k < ˛ on .[I 2P k nF k IM/. Thus for each " > 0 and each k
we have shown that we can find a finite collection of intervals which cover S and which have total
R R
volume < R 'k C ", hence (since R 'k < " for sufficiently large k ) S has measure zero. But then by
6 M ATH . 171 L ECTURE S UPPLEMENT 2011

Lemma 2 we have {x 2 R W lim 'k .x/ > 0} D [j1D1 {x 2 R W lim 'k > 1=j } also is a set of measure
zero as claimed.
Proof of “(”: Let " > 0. For each k D 1; 2; : : : let P k be a partition of R compatible with 'k (so
that 'k is constant on IM for each I 2 P k ) and let E be a set of Lebesgue measure zero such that
'k .x/ ! 0 for each x 2 R n E . Since E [ .[1 [
kD1 I 2P k
@I / has Lebesgue measure zero, we can select
open intervals I1 ; I2 ; : : : such that E [ .[kD1 [I 2P k @I /  [j1D1 Ij and j1D1 jIj j < ".
1
P

Now K D R n .[j1D1 Ij / is a compact set and by construction 'k .x/ ! 0 for each x 2 K and hence
for each given x 2 K we can find kx such that 'kx .x/ < " and also (since x … [I 2P kx @I ) there is
Jx 2 P kx with x 2 JMx and 'kx jJMx  const. < ". Now of course K  [x2K JMx so by compactness
there is a finite set of points x1 ; : : : ; xN 2 K such that K  [jND1 JMxj . Let L D max{kx1 ; : : : ; kxN }
and observe that if k  L then we have 'k < " on [jND1 JMxj because 'k ./  'kxj ./ < " for
each  2 JMxj and each k  kxj . So let Q be a partition of R with the property that, for i D
1; : : : ; n, the i ’th edge points of the partition Q include all the end points of all the i ’th edges of
the intervals in the collection Jx1 ; : : : ; JxN . Then each J 2 Q is either such that JM  JMxj for
some j or else J \ .[jND1 JMxj / D ∅. Let  be a step function which is " on the former intervals
and M D supR '1 on the latter (and keep in mind that each J of the latter kind is contained in
R n .[jND1 JMxj /  [j1D1 Ij ). So we have 'k   on R for all k  L, and hence R 'k  R  
R R

"jRj C M J 2QWJ Rn.[N JMx / jJ j  "jRj C M j1D1 jIj j < .jRj C M /" for all k  L. Thus we have
P P
R j D1 j
shown limk!1 R 'k D 0 as required.
The second lemma concerns increasing sequences of step functions:
Lemma 4. Suppose 'k 2 S.R/ with 'k  'kC1 for each k D 1; 2; : : : and suppose also that
R
{ R 'k }kD1;2;::: is bounded. Then {'k .x/}kD1;2;::: is bounded for a.e. x 2 R.

Proof: We have to prove that S has measure zero, where

S D {x 2 R W {'k }kD1;2;::: is unbounded}:

R R
We are given { R 'k }kD1;2;::: is bounded, and it is increasing by 1.5, hence lim R 'k 2 R and we
can select a subsequence {'kj }j D1;2;::: such that R .'kj C1 'kj / < 2 2j for each j D 1; 2; : : :. For
R

any ˛ > 1 observe that

S  [j1D1 Sj ; where Sj D {x 2 R W 'kj C1 .x/ 'kj .x/  2 j

˛ };

because x … Sj for each j implies 'kN .x/ 'k1 .x/ D jND1 'kj C1 .x/ 'kj .x/ < ˛ jND1 2 j < ˛ for
P P

each N  1. On the other hand for each j we can apply 1.6 with D 'kj C1 'kj and with 2 j ˛ in
place of ˛ , thus giving a partition Pj of R and a subcollection Fj  Pj with 'kj C1 'kj < 2 j ˛ on
[I 2Pj nFj IM and 2 j ˛ I 2Fj jI j  R .'kj C1 'kj / < 2 2j , hence ˛ j1D1 I 2Fj jI j < j1D1 2 j D
P R P P P

1. Since by construction {IM W I 2 Fj } covers all of Sj n .[I 2Pj @I /, and [j1D1 [I 2Pj @I has measure
zero by Lemma 2, we then have that [j1D1 {IM W I 2 Fj } is a collection of open intervals covering
almost all of [j1D1 Sj . S /, and hence we can construct a countable collection I1 ; I2 ; : : : of open
intervals with S  [j1D1 Sj  [j1D1 Ij and j1D1 jIj j  2=˛ . Since ˛ is arbitrarily large this shows
P

that S has measure zero as claimed.

 R IEMANN & L EBESGUE I NTEGRALS 7

We conclude this section with the very elegant theorem of Lebesgue which completely characterizes
those functions which are Riemann integrable.
Theorem (Lebesgue’s Theorem on the Riemann Integral.) Let f W R ! R be a bounded function.
f is Riemann integrable on R ” there is a set A  R of Lebesgue measure zero such that f W R ! R
is continuous at x for each point x 2 R n A.
(i.e. A bounded function f W R ! R is Riemann integrable if and only if f is continuous a.e. on R.)
Caution: “f W R ! R is continuous at each x 2 R n A” is a much stronger condition than “f jR n A
is a continuous function,” and indeed f jR n A continuous is in general not sufficient to ensure that
f is Riemann integrable even if A has measure zero. For example if we take R D Œ0; 1, A D the set
of rationals in Œ0; 1, then A has measure zero but the function f which is 1 on A and 0 on R n A is
R R
not Riemann integrable because evidently R f D 0 and R f D 1.
Proof of ): Observe, by the definition of continuity, that f discontinuous at y 2 RM ” 9 "0 > 0
such that supI f infI f > "0 8 open interval I with y 2 I  R, which is the same as saying there
is a positive integer j such that supI f infI f > 1=j 8 open interval I with y 2 I  R. Thus the
set of discontinuities of f jRM can be written [j1D1 Sj , where

Sj D {y 2 RM W supI f infI f > 1=j for every open interval I with y 2 I  R}:

By Lemma 2 we know that [j1D1 Sj has Lebesgue measure zero if each Sj has Lebesgue measure
zero, so it is thus enough to prove that Sj has Lebesgue measure zero for each j .
Let " > 0, j 2 {1; 2; : : :}. By Lemma 1 of 2 (with "=j in place of "), we can pick a partition P and
corresponding '; 2 S.R/ with '  f  and
X Z
.supIM f infIM f /jI j  . '/ < "=j:
I 2P R

Since supIM f infIM f  1=j whenever Sj \ IM ¤ ∅ (by definition of Sj ), the above evidently implies
P
{I 2PWSj \IM ¤∅} .1=j /jI j < "=j ; that is,
X
./ jI j < ":
{I 2PWSj \IM ¤∅}

But the intervals IM 2 P cover the entire interval R except for the set E D [I 2P @I of measure zero,
hence R n E D [I 2P IM and trivially therefore Sj n E  [{I 2PWSj \IM¤∅} IM. Thus ./ proves that Sj can
be covered by a finite union of intervals of total volume < " and hence Sj has Lebesgue measure
zero as required.
Proof of (: Let " > 0 and cover the set of discontinuities of f by a union [j1D1 Ij of open intervals
such that j1D1 jIj j < ". Then K  R n [j1D1 Ij is a compact set and f (as a function on R) is
P

continuous at each point of this compact set. We can therefore assert that there is ı > 0 such that

./ jf .x/ f .y/j < " whenever x 2 K; y 2 R; and jx yj < ı:

Notice that the statement is stronger than the standard fact that a continuous function on a compact
set is uniformly continuous, because only the point x , and not necessarily the point y , is required
8 M ATH . 171 L ECTURE S UPPLEMENT 2011

to be in the compact set K —on the other hand, the proof using the Bolzano-Weierstrass theorem
is almost identical to the usual Bolzano-Weierstrass proof of this standard fact, as follows: If there
is " > 0 such that ./ fails for each ı > 0 then it fails with ı D k1 , k D 1; 2; : : :, and hence
there are points xk 2 K; yk 2 R such that jxk yk j < k1 but jf .xk / f .yk /j  ". Then by the
Bolzano-Weierstrass theorem we can find a convergent subsequence xkj with x D lim xkj 2 K .
Since jxkj ykj j < k1j  j1 we also have lim ykj D x , and so by continuity of f at x we have
f .xkj / f .ykj / ! f .x/ f .x/ D 0, contradicting the fact that jf .xkj / f .ykj /j  " for each j .
Now, with ı as in ./, we select any partition P of R with each edge of each I 2 P having length
p
< ı= n (so diameter of each I 2 P is < ı ). For any I 2 P such that I \ K ¤ ∅ we have by ./
that supI f infI f D supz1 ;z2 2I .f .z1 / f .z2 // D supz1 ;z2 2I ..f .z1 / f .yI // .f .z2 / f .yI /// 
" C " D 2", where yI is any point in I \ K , while of course the sum of the volumes jI j over the
remaining I 2 P is  " (because the remaining I have the property I \ K D ∅ and hence all such
I  R n K  [j Ij , and j jIj j < "). Thus we have
P

X
.sup f inf f /jI j  2"jRj C .sup f inf f /"  2".jRj C M /; M D sup jf j:
I 2P I R
I R R

Thus if we define '; 2 S.R/ by 'jIM D infI f; I 2 P , and ' D infR f on R n [I 2P IM, and
jIM D supI f; I 2 P , and D supR f on R n [I 2P IM then R .
R
'/  2".jRj C M /, so f is
Riemann integrable by Lemma 1 of 2.

4 Definition and Properties of the Lebesgue Integral

L0 denotes the set of functions f W R ! R with the property that there exists an increasing sequence
R
{'k }kD1;2;:::  S.R/ with { R 'k }kD1;2;::: bounded and f .x/ D lim 'k .x/ a.e. x 2 R.
Similarly U 0 denotes the set of functions f W R ! R with the property that there exists a decreasing
R
sequence {'k }kD1;2;:::  S.R/ with { R 'k }kD1;2;::: bounded and f .x/ D lim 'k .x/ a.e. x 2 R.
(Observe that then actually U 0 D { f W f 2 L0 }.)
We are going to define the Lebesgue integral on the vector space L1 .R/ D L0 C U 0 .D {f W f D
g Ch with g 2 L0 and h 2 U 0 }/; notice that L1 .R/ can alternatively be written as L0 L0 .D {g h W
g; h 2 L0 }/. Observe that L1 .R/ is a vector space since it is a non-empty subset of the vector space
of real-valued functions on R which is closed under addition and multiplication by scalars.
In order to define the Lebesgue integral on L1 .R/, we first need to define it on L0 :
R R
4.1 Definition: For f 2 L0 we define the Lebesgue integral R f D lim R 'k , where 'k 2 S.R/ are
as in the definition of L0 .
R R
Notice that of course lim R 'k exists because { R 'k } is bounded by assumption, and it is mono-
R
tone increasing by 1.5. However before the definition makes sense we have to show that lim R 'ek D
R R
lim R 'k for any other sequence {'ek }kD1;2;:::  S.R/ with { R 'ek }kD1;2;::: bounded and f .x/ D
lim 'ek .x/ a.e. x 2 R. This is in fact an easy consequence of Lemma 3 of 3, as follows: For each
fixed k D 1; 2; : : : we have that {.'k 'e` /C }`D1;2;::: is a decreasing sequence of non-negative step
functions which converges to zero at a.e. x 2 R, and hence by Lemma 3 of 3 (the “(” direction of
 R IEMANN & L EBESGUE I NTEGRALS 9

R R R R
that lemma) we have lim`!1 R .'k 'e` /C D 0 for each k , and hence R 'k R 'e` D R .'k 'e` / 
R R R
R .'k 'e` /C ! 0 as ` ! 1. Thus lim R 'k  lim R 'ek , and the reverse inequality is proved by
interchanging 'k and 'ek .
Thus the definition of the Lebesgue integral makes sense on L0 and evidently has the additivity
property that
Z Z Z


4:2 g; h 2 L0 ) g C h 2 L0 and f Cg D gC h;
R R R

because if 'k ; k are increasing sequences of step functions converging a.e. to f; g respectively with
R R
R ' k and R k bounded, then 'k C k is an increasing sequence of step functions with 'k C k !
R R R
g C h a.e. and by 1.4 R .'k C k / D R 'k C R k for each k .

We can now immediately extend the definition of the Lebesgue integral to the vector space L1 .R/
as follows:
4.3 Definition: If f 2 L1 .R/ then we define
R R R
R f D R g R h, where g; h 2 L0 are such that
f D g h.
We of course have to check that this definition is independent of which g; h 2 L0 we choose to
represent f . So suppose f; g; fe; ge 2 L0 and f g D fe ge. Then f C ge D fe C g and by the
R R R R
additivity 4.2 we then have R f Rg D Rf
e e.
Rg

We now check some of the basic properties of the Lebesgue integral on L1 .R/:
First we have the linearity
Z Z Z
1 1

4:4 f; g 2 L .R/; ;  2 R ) f C g 2 L .R/ and f C g D  f C g:
R R R

This is clear because L0 is closed under addition and multiplication by non-negative scalars, and
L1 .R/ is by definition closed under multiplication by 1, and by the Definition 4.3 (in the special
R R R R
case when g D 0) we have R . h/ D R h for each h 2 L0 , and hence R . f / D R f for each
f 2 L1 .R/.
Next we show that changing an L1 .R/ function on a set of measure zero neither changes the fact
that the function is in L1 .R/ nor does it change the value of the integral:
Z Z
4:5 f 2 L1 .R/; fe W R ! R; and fe D f a.e. ) fe 2 L1 .R/ and f D fe:
R R

This is clear because if g 2 L0 with and if ge W R ! R is such that ge D g a.e. then ge 2 L0 and
R R
Rg D Rg e (indeed according to Definition 4.1 the same increasing sequence of step functions 'k
R R
can be used to define both R g and R ge because that definition only requires 'k to converge to the
relevant function a.e.).
If f 2 L0 with f  0 then by definition we have g; h 2 L0 with f D g h hence g  h, and it
R R
easily follows from Definition 4.1 and from 1.5 that R g  R h. Hence, by the Definition 4.3,
R R R
Rf D Rg R h  0. Thus:
Z
1
4:6 f 2 L .R/; f  0 ) f  0;
R
10 M ATH . 171 L ECTURE S UPPLEMENT 2011

and of course then by applying this to differences of functions f; g 2 L1 .R/ and using the linear-
ity 4.4 we have
Z Z
1
4:7 f; g 2 L .R/ with f  g ) f  g:
R R

Next we observe that L1 .R/ is closed under the operation of taking absolute values:
ˇZ ˇ Z
1 1
4:8 f 2 L .R/ ) jf j 2 L .R/ and ˇ f ˇ 
ˇ ˇ
jf j:
R R

To check this we choose g; h 2 L0 with f D g h and observe that then jf j D max{g; h}

min{g; h} and also max{g; h}; min{g; h} 2 L0 (by definition of L0 ), and so jf j 2 L1 .R/ and by the
R R R R R R
Definition 4.3 and by 4.7 R jf j D R max{g; h} R min{g; h}  R g R h D R f , and then
the proof of 4.8 is completed by applying the same argument to f .
The next property seems intuitively obvious but the proof requires Lemma 3:
Z
4:9 f 2 L1 .R/; f  0; f D 0 ) f D 0 a.e.
R

To check 4.9 choose g; h 2 L0 with f D g h, and then observe that by definition of L0 we have
increasing sequences { ` }`D1;2;::: and {'k }kD1;2;::: of step functions with ` ! g a.e. and 'k ! h
a.e., and so . ` h/C D limk!1 . ` 'k /C a.e., and the right side here is a decreasing sequence of
step functions for each `, so by Definitions 4.1,4.3 and by 4.7,4.5 we have
Z Z Z
lim . ` 'k /C D . ` h/C  f D 0;
k!1 R R R

where at the last step we used the hypothesis f  0, and hence we can apply Lemma 3 (the “)”
direction of that lemma with . ` 'k /C in place of 'k ) to give . ` h/C D lim. ` 'k /C D 0
a.e., and hence ` h  0 a.e., so by taking the limit with respect to ` we have f D g h  0 a.e.,
and hence f D 0 a.e. as claimed.
We’ll now check that the Lebesgue integral exists and agrees with the Riemann integral whenever
the Riemann integral exists. (Thus the Lebesgue integral entirely subsumes the Riemann integral.)
Using Lemma 1 of 2 with " D 1=k we see that if f W R ! R is Riemann integrable then there are
R R R R
step functions k ; 'k with 'k  f  k and R k 1=k  R 'k  RI.f /  R k  R 'k C 1=k
R R
for k D 1; 2; : : :, where RI.f / is the Riemann integral of f , and hence R ‰k and R ˆk both
converge to RI.f /, where ‰k D min{M; 1 ; : : : ; k } and ˆk D max{ M; '1 ; : : : ; 'k }, where
M is any positive constant such that jf j < M on R. Thus we have ˆk  f  ‰k , ˆk is a
bounded increasing sequence of step functions, and ‰k is a bounded decreasing sequence of step
R R
functions. Also R .‰k ˆk /  R . k 'k / ! 0, so by Lemma 3 (the “)” direction of that
lemma) we have ‰k .x/ ˆk .x/ ! 0 a.e. x 2 R. Since ˆk  f  ‰k for all x we thus have
lim ˆk .x/ D lim ‰k .x/ D f .x/ a.e. x 2 R. But then f 2 L0 \ U 0 by definition of L0 ; U 0 ; here, as
introduced at the beginning of this section, U 0 means L0 D { f W f 2 L0 }. Also, by definition
R R
the Lebesgue integral on L0 , R f is limk!1 R ˆk , which is also the Riemann integral of f by
construction. Thus we have proved:
 R IEMANN & L EBESGUE I NTEGRALS 11

Lemma 5. The set R.R/ of Riemann integrable functions f W R ! R is a linear subspace of L0 \ U 0

and the Riemann integral and the Lebesgue integral coincide on R.R/.

Actually there is also a partial converse: If f 2 L0 \ U 0 then we can find an increasing sequence
'k and a decreasing sequence k of step functions with f .x/ D lim 'k .x/ D lim k .x/ a.e. x 2 R.
For each k , let P k be a partition of R such that both 'k and k are constant on each IM with
I 2 P k . Evidently we must then have 'k jIM  k jIM for each I 2 P k . Let M > 0 be such that
M < '1 and 1 < M everywhere on R. Then we can define 'ek .x/ D 'k .x/ for x 2 [I 2P k IM
and 'ek .x/ D M on [I 2P k @I , and similarly ek .x/ D k .x/ for x 2 [I 2P k IM and ek .x/ D M on
[I 2P k @I . Then 'ek .x/  ek .x/ at every point and they converge to f .x/ a.e. Indeed if E is a set
of measure zero such that both 'k .x/ and k .x/ converge to f .x/ at every point of R n E , then
by construction 'ek .x/ and ek .x/ both converge to f .x/ for every x 2 R n .E [ .[1 [
kD1 I 2P k
@I //.
However 'ek ; k are not monotone sequences. To remedy this we define ˆk D max{'e1 ; : : : ; 'ek }
e
and ‰k D min{ e1 ; : : : ; ek }. These are respectively increasing and decreasing sequences of step
functions with M  ˆk  ‰k  M everywhere on R and ˆk .x/; ‰k .x/ converge to f .x/ at every
point x 2 R n .E [ .[1 [
kD1 I 2P k
@I //. Thus if we let fe.x/ D limk!1 ˆk .x/ for every x 2 R, then
we have fe D f a.e. and ˆk  fe  ‰k for each k , so fe is Riemann integrable by Lemma 1 of 2.
Thus we have proved:
Lemma 6. If f 2 L0 \ U 0 then there is a Riemann integrable fe with fe D f a.e.
Remark: Note that by combining Lemma 5 and Lemma 6 we have proved that f 2 L0 \U 0 ” f
is a.e. equal to a Riemann integrable function fe, and in that case the Riemann integral of fe is equal
R
to the Lebesgue integral R f of f .

We conclude these notes with a discussion of the completeness property of the space L1 .R/; the
completeness property described in the theorem below highlights one of the key advantages of
using the Lebesgue integral rather that the Riemann integral.
First, For f 2 L1 .R/ we define Z
kf k1 D jf j;
R

and observe that k k1 has the first two properties of a norm: kf k2 D jjkf k1 and kf C gk1 
kf k1 C kgk1 , but in place of the usual positivity we have only kf k1  0 for each f 2 L1 .R/,
because kf k1 D 0 evidently does not imply f D 0 but rather that f D 0 a.e. by 4.9. We therefore
classify k k1 as a “seminorm” rather than a norm, but we keep in mind that the positivity of k k1 fails
in very mild way, namely (by 4.9)

kf k1 D 0 ) f .x/ D 0 a.e. x 2 R:

Here we establish the following completeness property of L1 .R/:

Theorem. Suppose {fk }kD1;2;::: is a Cauchy sequence with respect to the seminorm k k1 above; that is,
for each " > 0 there is N such that R jfk f` j < " whenever both k; `  N . Then there is f 2 L1 .R/
R
R
such that R jfk f j ! 0.
12 M ATH . 171 L ECTURE S UPPLEMENT 2011

Proof: We use the definition of Cauchy sequence with " D 2 j ; j D 1; 2; : : :, in order to select kj
such that R jf` fkj j < 2 j for all `  kj , and by selecting these kj successively we can at the
R

same time arrange that kj C1 > kj for all j  1, so that we have in particular that {fkj }j D1;2;::: is a
subsequence of {fk }kD1;2;::: and
j
.1/ kfkj fkj C1 k1 < 2

for each j . Since fkj 2 L1 .R/ we have fkj D gj hj with gj ; hj 2 L0 and by definition we can
j 1 , so that
R R
select 'j ; j 2 S.R/ with 'j  gj and j  hj and with R .gj 'j /; R .hj j/ < 2
with j D 'j j we have
Z Z Z
j
.2/ jfkj j j D j.gj 'j / .hj j /j  .gj 'j / C .hj j/ 2 ;
R R R

and hence by (1) and the triangle inequality we have

j C2
.3/ kj j C1 k1  2 ; j D 1; 2; : : : :

Notice that then with 0 D 0 we have ` D j`D10 .j C1 j /, and ˆ` D j`D10 .j C1
P P
j /C ; ‰` D
P` 1
j D0 .j C1 j / are increasing sequences of non-negative step functions with
Z X` 1 X` 1
j C2
.4/ .ˆ` C ‰` / D kˆ` C ‰` k1  kj C1 j k1  2 C k1 k1 < 4 C k1 k1
R j D0 j D1

by (3), and hence we have by Lemma 4 that {ˆ` .x/}`D1;2;::: and {‰` .x/}`D1;2;::: are bounded a.e.
x 2 R and we can use the definition of L0 .R/ to assert that ˆ` .x/ ! g.x/ and ‰` .x/ ! h.x/ a.e.
x 2 R, where g; h 2 L0 .R/ and R .g ˆ` /; R .h ‰` / ! 0. We therefore have f D g h 2 L1 .R/
R R
R
and R jf ` j ! 0, so by (2) kf fj k1 D kf j C j fkj C fkj fj k1  kf j k1 C kj
fkj k1 C kfkj fj k1 ! 0 as j ! 1.