2007-08 Second Term MAT2060A Supplementary Notes 2: Riemann Integral

An Integrability Criterion
The setting of Riemann integral is a bounded function f dened on a bounded,

closed interval [a, b]. A partition of [a, b], P , is a nite collection of points a = x0 < x1 < ... < xn = b, which divides [a, b] into n many subintervals Ij = [xj1 , xj ], j = 1, ...n. The length of a partition is given by ||P || = max(xj xj1 ). A tagged partition is the pair (P, z1 , ..., zn ) where zi Ij . We shall use P to denote a tagged partition. Given any tagged partition P , we dene the Riemann sum of f with respect to P by S(f, P ) =
1 n j

f (zj ) xj ,

xj = xj xj1 .

Geometrically, S(f, P ) is an approximate area of the region bounded by x = a, x = b, y = 0 and y = f (x) (assuming f is positive). We call f Riemann integrable if there exists L R so that , s.t. |S(f, P ) L| < , P, ||P || < . For any partition P , we dene its Darboux upper and lower sums respectively by
n1

S(f, P ) =
j=0

M j xj ,
n1

and S(f, P ) =

mj xj ,
j=0

where Mj = sup{f (x) : x [xj , xj+1 ]} and mj = inf {f (x) : x [xj , xj+1 ]}. A partition P2 is a renement of P1 if every partition point of P1 is a partition point of P2 . We have

2007-08 Second Term MAT2060A Proposition 1. Let P2 be a renement of P1 . Then S(f, P1 ) S(f, P2 ), and S(f, P1 ) S(f, P2 ).

(1)

(2)

Proof. Let [xj , xj+1 ] be a subinterval of P1 . It can be decomposed into the union of subinterval of P2 : [yk , yk+1 ] ... [yl , yl+1 ] where yk = xj and yl+1 = xj+1 Then Mj Mk , ..., Ml , where Mj = supf over [xj , xj+1 ] and Mk = supf over [yk , yk+1 ], etc. From this we conclude that S(f, P1 ) = Mj xj Mj yj

= S(f, P2 ). So (1) holds. Similarly, one can prove (2).

The following properties are now clear. Proposition 2. For any partition P and Q, S(f, P ) S(f, Q). (3)

Proof. By putting the partition points of P and Q together we obtain a partition R which renes both P and Q. By Proposition 1, S(f, P ) S(f, R) S(f, R) S(f, Q). The following proposition is a straightforward consequence from the denition of the Darboux sums.

2007-08 Second Term MAT2060A Proposition 3. For any partition P , S(f, P ) S(f, P ) S(f, P ). for any tags. Moreover, given > 0, there exists a tag such that S(f, P ) + S(f, P ), and another tag S(f, P ) such that S(f, P ) S(f, P ). We dene the Riemann upper and lower integrals respectively to be S(f ) = inf S(f, P ),
P

and S(f ) = sup S(f, P ).

Proposition 4. S(f ) = lim S(f, Pj ),

and S(f ) = lim S(f, Pj ),

for any Pj , ||Pj || 0. Proof. Given > 0, there exists Q such that S(f ) + /2 > S(f, Q). Let N be the number of partition points of Q. Consider any P and R be the partition by putting together P and Q. The partition points of Q are contained in at most 2(N 2) + 2 many subinterval of R. Denote the indices of the collection of these subintervals by J. We have |S(f, P ) S(f, R)|
jJ

4M xj 2M 2(N 2),

2007-08 Second Term MAT2060A

where M = sup |f |, because the contributions of S(f, P ) and S(f, Q) from the subintervals not in J cancel out. Hence S(f ) + /2 > S(f, R) > S(f, P ) 8M (N 1), i.e., 0 S(f, P ) S(f ) < /2 + 8M (N 1). Now, as ||Pj || 0, we can nd j0 such that ||Pj || < then 0 S(f, Pj ) S(f ) < , j j0 . Similarly, one can prove the second assertion. , j j0 , 16M (N 1)

The following is our main result a criterion for Riemann integrability in terms of the Darboux sums and upper/lower integrals. Theorem 5. Let f be bounded in [a, b]. The following are equivalent: (a) f is Riemann integrable in [a, b], (b) For > 0, there exists P such that S(f, P ) S(f, P ) < , (c) S(f ) = S(f ), Under (a),(b) or (c),
b a

f = S(f ) = S(f ).

Notice that when (b) holds, for any renement of P, Q, we also have S(f, Q) S(f, Q) < . Proof. (a) (b). Given > 0, P such that |S(f, P ) L| < ,

2007-08 Second Term MAT2060A for any tags. By Proposition 3, |S(f, P ) S(f, P )| |S(f, P ) L| + |L S(f, P )| |S(f, P ) S(f, P )| + |S(f, P ) L| + |S(f, P ) S(f, P )| + |S(f, P ) L| 3, whence (b) holds. (b) (c). By denition, S(f ) S(f ) S(f, P ) S(f, P ) < . Since is arbitrary, (c) holds.

(c) (a). Since f is not integrable. Then for the number S(f ), there exists 0 and Pj , ||Pj || 0, such that S(f, Pj ) S(f ) 0 , for some tags. Then S(f, Pj ) S(f ) S(f, Pj ) S(f ) 0 . By Proposition 4 and (c) S(f, Pj ) S(f ) = S(f, Pj ) S(f ) 0. as j 0, contradiction holds. Theorem 6. Let f and g be integrable in [a, b] and , R. We have (a) f + g is integrable in [a, b] and
b b b

(f + g) =
a a

f +
a

g.

(b) If f is non-negative, then

f 0.

2007-08 Second Term MAT2060A

Proof of (a) We only consider the case = = 1 in (a). From theorem 5 for given > 0, we can nd a partition P such that S(f, P ) S(f, P ) < /2, S(g, P ) g(f, P ) < /2. Using the inequalities sup(f + g) sup f + sup g, inf(f + g) inf f + inf g, we deduce S(f + g, P ) S(f + g, P ) < /2 + /2 = . Using theorem 5 again f + g is integrable and from S(f + g, P ) S(f, P ) + S(g, P ) = S(f, P ) + S(g, P ) S(f + g, P ) we deduce the identity in (a). We leave the proof of (b) to the reader.

Integrable Functions

Theorem 7. Any continuous function in [a,b] is integrable. Proof. That f is continuous in [a,b] implies that it is uniformly continuous in [a,b]. For > 0, s.t. |f (x) f (y)| < , x, y, |x y| < . 2(b a)

2007-08 Second Term MAT2060A Consider any P, ||P || < , we have S(f, P ) = S(f, P ) = f (zj ) xj , f (wj ) xj ,

where f (zj ) = Mj and f (wj ) = mj by continuity. Therefore, 0 S(f, P ) S(f, P ) = (f (zj ) f (wj )) xj , (b a) 2(b a)

< . By Theorem 5 (b) f is integrable. Theorem 8. Let f and g be integrable in [a, b] and [b, c] respectively. Then f (x), x [a, b], F (x) = g(x), x (b, c], is integrable in [a,c] and
a c b c

F =
a

f+
b

g.

Proof. Given > 0, P of [a,b] and Q of [b,c] s.t.

f S(f, P ) < /3, g S(f, Q) < /3.

Putting P and Q together to form a partition R for [a,c], with the inherited tags, S(F, R) = S(f, P ) + S(g, Q) + error. The error arises from the fact that f may not equal to g at b. When one takes the tags for R, P or Q, the contributions from the subintervals touching b may not equal. However, letting the subinterval of P touching b be [, b] and tag be z, the subinterval of Q touching b be [b, ] and tag be w, and the tags of R over [, b] and [b, ] be z and w respectively, the error terms are F (z )(b ) + F (w )( b) f (z)(b ) g(w)( b).

2007-08 Second Term MAT2060A It can be estimated by 4M , where = ||R||. So S(F, R)

f
b

g < S(f, P ) < 2 + 4M , 3

f + S(f, Q)
a

g + |error|

< , after we choose < 1/12M. Since the tags are arbitrary, it means, S(F, R)
a b b c

f
b c

g < , g < ,
b

S(F, R)
a

or |S(F, R) S(F, R)| < 2. Now the theorem follows from Theorem 5 (b).

In the exercise you are asked to show that the function obtained from modifying an integrable one at nitely many points is still integrable and has the same integral as the old one. Together with Theorem 8 it shows that any piecewise continuous function in [a, b] is integrable. Recall that a function is piecewise continuous if there exists a partition x0 < ... < xN of [a, b] so that f |(xj1 ,xj ) is continuous and limxx f, limxx+ f exist. j j1 Theorem 9. Any monotone function in [a,b] is integrable. Proof. Let P be the partition which divides [a,b] equally. Observing that Mj mj = f (xj+1 ) f (xj ) (here f is assumed to be increasing),
n1

S(f, P ) S(f, P ) =
j=0

(Mj mj ) xj ba n
n1

= =

(f (xj+1 ) f (xj ))
j=0

(b a)(f (b) f (a)) n

< ,

2007-08 Second Term MAT2060A

if we choose n so large that (b a)(f (b) f (a))/ < n. By Theorem 5 (b) f is integrable.

Lebesgue Theorem

For any bounded f in [a,b] and x [a, b], its oscillation at x is dened by (f, x) = inf {(sup f (y) inf f (y)) : y (x , x + ) [a, b]}

= lim {(sup f (y) inf f (y)) : y (x , x + ) [a, b]}. +

It is clear that (f, x) = 0 if and only if f is continuous at x. The set of discontinuity of f , D, can be written as D = O(k), k=1 where O(k) = {x [a, b] : (f, x) 1/k}. A subset E of R is measure zero if > 0, intervals {Ij } such that E Ij , j=1 and

(4)

|Ij | < .
j=1

It is not hard to show that Proposition 10. (i) Any countable set is of measure zero. (ii) Any countable union of measure zero sets is again of measure zero. Proof. Let E = {x1 , x2 , ...} be a countable set. Given > 0, the intervals Ij = (xj
, xj 2j+2

) 2j+2

satisfy E Ij , j=1

and

|Ij | =
j=1 j=1

2 = < , j+2 2 2

2007-08 Second Term MAT2060A

10

so E is of measure zero. (i) is proved. (ii) can be proved by a similar argument. We leave it as an exercise. There are uncountable sets of measure zero. The famous Cantor set is one of them. See Chapter 7 of the text. Now, we state the necessary and sucient condition for Riemann integrability due to Lebesgue. Theorem 11. A bounded function f in [a, b] is Riemann integrable if and only if its discontinuity set is measure zero. We shall use the compactness of a closed, bounded interval in the proof of this theorem. Recall that compactness is equivalent the following property: Let K be a compact set in R. Suppose {Ij } is a sequence of open intervals satisfying K Ij , the we can choose nitely many intervals Ij1 , ..., IjN so that K Ij1 ... IjN . Proof. Recall the formula D = O(k). k=1 (5)

By Proposition 10(ii) it suces to show that each O(k) is of measure zero. Given > 0, we x a partition P such that S(f, P ) S(f, P ) < /2k. Let J be the index set of those subintervals of P which contains some elements of O(k) in their interiors. Then 1 k |Ij | =
jJ jJ n1

(sup f inf f ) xj
Ij Ij

j=0

(sup f inf f ) xj
Ij Ij

= S(f, P ) S(f, P ) < /2k. Therefore |Ij | < /2.

2007-08 Second Term MAT2060A

11

Now, the only possibility that an element of O(k) is not contained by one of these Ij is it being a partition point. Since there are nitely many partition points, we can nd I1 , ..., IN containing these partition points which satisfy |Ij | < /2. So {Ij } and {Ii } together form a covering of O(k) and its total length< . We conclude that O(k) is of measure zero. Conversely, given > 0, x a large k,
1 k

< . Now the set O(k) is of measure

zero, we can nd open intervals {Ij } satisfying O(k) Ij , j=1

|Ij | < .
j=1

One can show that O(k) is closed and bounded, hence it is compact. As a result, we can nd Ii1 , ..., IiN from {Ij } so that O(k) Ii1 ... IiN ,
N

|Ij | < .
j=1

Observe that [a, b] \ Ii1 ...IiN is a nite disjoint union of closed intervals, call them Vi s, i A. At every point x of these closed intervals we have (f, x) < 1 . k

We can nd a partition on each Vi so ne that the oscillation on each subinterval is less than 1/k. All these partitions and the endpoints of Ii1 , ..., IiN form a partition P of [a,b]. We have S(f, P ) S(f, P ) =
Iij N

(Mj mj ) xj + 1 k

(Mj mj ) xj

= 2M
j=1

|Iij | +

xj

= 2M + (b a) = [2M + (b a)],

2007-08 Second Term MAT2060A

12

where M = sup[a,b] |f | and the second summation is over all subinterval in Vi , i A. By Theorem 5 (b) f is integrable.