Functions of Bounded Variation

Our main theorem concerning the existence of RiemannStietjes integrals assures us that the integral

b
a
f(x) d(x) exists when f is continuous and is monotonic. Our linearity theorem then guarantees that
the integral

b
a
f(x) d(x) exists when f is continuous and is the dierence of two monotonic functions.
In these notes, we prove that is the dierence of two monotonic functions if and only if it is of bounded
variation, where
Denition 1
(a) The function : [a, b] IR is said to be of bounded variation on [a, b] if and only if there is a constant
M > 0 such that
n

i=1

(x
i
) (x
i1
)

M
for all partitions P = {x
0
, x
1
, , x
n
} of [a, b].
(b) If : [a, b] IR is of bounded variation on [a, b], then the total variation of on [a, b] is dened to be
V

(a, b) = sup

i=1

(x
i
) (x
i1
)

P = {x
0
, x
1
, , x
n
} is a partition of [a, b]

Example 2 If : [a, b] IR is monotonically increasing, then, for any partition P = {x

0
, x
1
, , x
n
} of
[a, b]
n

i=1

(x
i
) (x
i1
)

=
n

i=1

(x
i
) (x
i1
)

= (x
n
) (x
0
) = (b) (a)
Thus is of bounded variation and V
f
(a, b) = (b) (a).
Example 3 If : [a, b] IR is continuous on [a, b] and dierentiable on (a, b) with sup
a<x<b
|

(x)| M,
then, for any partition P = {x
0
, x
1
, , x
n
} of [a, b], we have, by the Mean Value Theorem,
n

i=1

(x
i
) (x
i1
)

=
n

i=1

(t
i
)[x
i
x
i1
]

i=1
M[x
i
x
i1
] = M(b a)
Thus is of bounded variation and V
f
(a, b) M(b a).
Example 4 Dene the function : [0, 1] IR by
(x) =

0 if x = 0
xcos

x
if x = 0
This function is continuous, but is not of bounded variation because it wobbles too much near x = 0. To see
this, consider, for each m IN, the partition P
m
= {0,
1
2m
,
1
2m1
, ,
1
3
,
1
2
, 1}. The values of at the points
of this partition are (P
m
) = {0,
1
2m
,
1
2m1
,
1
2m2
, ,
1
3
,
1
2
, 1}.
January 28, 2008 Functions of Bounded Variation 1
x
n
= 1
x
n1
=
1
2
x
n2
For this partition,
n

i=1

(x
i
) (x
i1
)

1
2m
0

1
2m1

1
2m

1
(2m2)
+
1
2m1

+ +

1
3

1
4

1
2
+
1
3

1
1
2

=
1
2m
+
1
2m1
+
1
2m
+
1
(2m2)
+
1
2m1
+ +
1
3
+
1
4
+
1
2
+
1
3
+ 1 +
1
2
= 2

1
2m
+
1
2m1
+ +
1
2

+ 1
The series

k=2
1
k
diverges. So given any M, there is a partition P
m
for which
n

i=1

(x
i
) (x
i1
)

> M
Theorem 5
(a) If , : [a, b] IR are of bounded variation and c, d IR, then c + d is of bounded variation and
V
c+d
(a, b) |c|V

(a, b) +|d|V

(a, b)
(b) If : [a, b] IR is of bounded variation on [a, b] and [c, d] [a, b], then is of bounded variation on
[c, d] and
V

(c, d) V

(a, b)
(c) If : [a, b] IR is of bounded variation and c (a, b), then
V

(a, b) = V

(a, c) + V

(c, b)
(d) If : [a, b] IR is of bounded variation then the functions V (x) = V

(a, x) and V (x) (x) are both

increasing on [a, b].
(e) The function : [a, b] IR is of bounded variation if and only if it is the dierence of two increasing
functions.
Proof: We shall use the shorthand notation

for
n

i=1

(x
i
) (x
i1
)

where the partition P = {x

0
, x
1
, , x
n
}.
(a) follows from the observation that, for any P partition of [a, b],

i
(c + d)

|c|

+|d|

|c|V

(a, b) +|d|V

(a, b)
January 28, 2008 Functions of Bounded Variation 2
(b) follows from the observation that, for any partition P of [c, d],

P{a,b}

(a, b)
(c) Since

(x
i
) (x
i1
)

(x
i
) (c)

(c) (x
i1
)

, we have

P{c}

(P{c})[a,c]

(P{c})[c,b]

(a, c) + V

(c, b)
which implies that V

(a, b) V

(a, c) + V

(c, b). To prove the other inequality, we let > 0 and select a
partition P
1
of [a, c] for which

P1

(a, c) and a partition P

2
of [c, b] for which

P2

(c, b) . Then

P1P2

P1

P2

(a, c) + V

(c, b) 2
This assures that V

(a, b) V

(a, c) +V

(c, b) 2 for all > 0 and hence that V

(a, b) V

(a, c) +V

(c, b).
(d) Let a x
1
x
2
b. That V (x
1
) = V

(a, x
1
) V

(a, x
2
) = V (x
2
) follows immediately from part (b).
By part (c),
[V (x
2
) (x
2
)] [V (x
1
) (x
1
)] = V

(x
1
, x
2
) [(x
2
) (x
1
)] V

(x
1
, x
2
)

(x
2
) (x
1
)

So the inequality [V (x
2
) (x
2
)] [V (x
1
) (x
1
)] follows from

(x
2
) (x
1
)

{x1,x2}

(x
1
, x
2
)
(e) If is of bounded variation then (x) = V

(a, x)

(a, x) (x)

expresses as the dierence of two

increasing functions. On the other hand if is the dierence of two increasing functions, then and
are of bounded variation by Example 2 and is of bounded variation by part (a).
Example 6 We know that if f is continuous and is of bounded variation on [a, b], then f R() on
[a, b]. If f is of bounded variation and is continuous on [a, b], then we have f R() on [a, b] with

b
a
f d = f(b)(b) f(a)(a)

b
a
df
by our integration by parts theorem. It is possible to have f R() on [a, b] even if neither f nor are of
bounded variation on [a, b]. For example, we have seen, in Example 4, that
(x) =

0 if x = 0
xcos

x
if x = 0
is continuous but not of bounded variation on [0, 1], because of excessive oscillation near x = 0. So f(x) =
(1 x) (still with the of Example 4) is continuous but not of bounded variation on [0, 1], because of
excessive oscillation near x = 1. But f R() on [0,
1
2
], by integration by parts, because f is of bounded
variation on [0,
1
2
]. And f R() on [
1
2
, 1], because is of bounded variation on [0,
1
2
]. So f R() on
[0, 1], by our linearity theorem.
January 28, 2008 Functions of Bounded Variation 3