# Chapter 7 – Functions of Bounded Variation.

Subject: Real Analysis Level: M.Sc. Source: Syed Gul Shah (Chairman, Department of Mathematics, US Sargodha) Collected & Composed by: Atiq ur Rehman (atiq@mathcity.org), http://www.mathcity.org

**We shall now discuss the concept of functions of bounded variation which is
**

closely associated to the concept of monotonic functions and has wide application in mathematics. These functions are used in Riemann-Stieltjes integrals and Fourier series. Let a function f be defined on an interval [a, b] and P = {a = x0 , x1 ,........, xn = b} be a partition of [a, b] . Consider the sum

å

i =1

n

f ( xi ) - f ( xi-1 ) . The set of these sums

is infinite. It changes when we make a refinement in a partition. If this set of sums is bounded above then the function f is said to be a bounded variation and the supremum of the set is called the total variation of the function f on [a, b] , and is denoted by V ( f ; a, b ) or V f ( a, b ) and it is also affiliated as V ( f ) or V f . Thus V ( f ; a, b ) = sup å f ( xi ) - f ( xi -1 )

i =1 n

The supremum being taken over all the partition of [a, b] . Hence the function f is said to be of bounded variation on [a, b] if, and only, if its total variation is finite i.e. V ( f ; a, b ) < ¥ . Ð Note Since for x £ c £ y , we have f ( y ) - f ( x ) £ f ( y ) - f (c ) + f (c ) - f ( x ) Therefore the sum å f ( xi ) - f ( xi-1 ) can not be decrease (it can, in fact only increase) by the refinement of the partition. Ð Theorem A bounded monotonic function is a function of bounded variation. Proof Suppose a function f is monotonically increasing on [a, b] and P is any partition of [a, b] then

å

i =1

n

f ( xi ) - f ( xi-1 )

= å ( f ( xi ) - f ( xi -1 ) ) = f (b) - f (a )

i =1

n

\ V ( f ; a, b ) = sup å f ( xi ) - f ( xi -1 ) = f (b) - f (a ) (finite) Hence the function f is of bounded variation on [a, b] . Similarly a monotonically decreasing bounded function is of bounded variation with total variation = f (a ) - f (b ) . Thus for a bounded monotonic function f V ( f ) = f (b) - f (a) q

{{{{{{{{{{{{{{{{{{{{{{{{{{{

2

Chap 7 – Functions of bounded variation.

Ð Example A continuous function may not be a function of bounded variation. e.g. Consider a function f , where p ì ; when 0 < x £ 1 ï x sin f ( x) í x ; when x = 0 ï 0 î It is clear that f is continuous on [0,1] . 2 2 2 2 ü ì Let us choose the partition P = í0, , ,........, , ,1ý 5 3 þ î 2n + 1 2n - 1 Then 2 2 2 2 ç ÷ ç ÷ ç ÷ ç ÷ å f ( xi ) - f ( xi-1 ) = f (1) - f æ 3 ö + f æ 3 ö - f æ 5 ö + ....... + f æ 2n + 1 ö - f ( 0 ) è ø è ø è ø è ø 2 æ 3p ö 2 æ 3p ö 2 æ 5p ö = sin p - sin ç ÷ + sin ç ÷ - sin ç ÷ + .......... 3 è 2 ø 3 è 2 ø 5 è 2 ø 2 æ (2n + 1)p ö ..... + sin ç ÷-0 2n + 1 è 2 ø 2 æ2 2ö æ2 2ö 2 ö 2 æ 2 = + ç + ÷ + ç + ÷ + ............ + ç + ÷+ 3 è3 5ø è5 7ø è 2n - 1 2n + 1 ø 2 n + 1 æ æ2ö æ2ö æ2ö æ 2 öö = ç 2 ç ÷ + 2 ç ÷ + 2 ç ÷ + ............. + 2 ç ÷÷ è5ø è7ø è 2n + 1 ø ø è è3ø 1 ö æ1 1 1 = 4 ç + + + ............. + ÷ 2n + 1 ø è3 5 7 1 1 1 Since the infinite series + + + ........... is divergent, therefore its partial sums 3 5 7 1 1 1 1 sequence {Sn } , where S n = + + + ........ + , is not bounded above. 3 5 7 2n + 1 Thus å f ( xi ) - f ( xi-1 ) can be made arbitrarily large by taking n sufficiently large. Þ V ( f ;0,1) ® ¥ and so f is not of bounded variation. q Ð Remarks A function of bounded variation is not necessarily continuous. e.g. the step-function f ( x ) = [ x] , where [ x] denotes the greatest integer not greater than x, is a function of bounded variation on [0, 2] but is not continuous. Ð Theorem If the derivative of the function f bounded variation on [a, b] . Proof Q f ¢ is bounded on [a, b] \ $ k such that f ¢( x ) £ k Let P be any partition of the interval å f ( xi ) - f ( xi-1 ) = å xi exists and is bounded on [a, b] , then f is of

" x Î [ a, b] . [a, b] then - xi -1 f ¢(c) ,

c Î [ a, b]

(by M.V.T) q

Þ V ( f ; a, b )

£ k b-a is finite. Þ f is of bounded variation.

Chap 7 – Functions of bounded variation.

3

Note

Boundedness of f ¢ is a sufficient condition for V ( f ) to be finite and is not necessary. Ð Theorem A function of bounded variation is necessarily bounded. Proof Suppose f is of bounded variation on [a, b] . For any x Î [a, b] , consider the partition {a, x, b} , consisting of just three points then f ( x ) - f (a) + f (b) - f ( x ) £ V ( f ; a, b)

Þ

Again

f ( x ) - f ( a ) £ V ( f ; a, b)

f ( x ) = f (a) + f ( x ) - f (a) £ f (a ) + f ( x ) - f ( a ) £ f (a ) + V ( f ; a , b ) < ¥ Þ f is bounded on [a, b] .

q

Ð Properties of functions of bounded variation 1) The sum (difference) of two functions of bounded variation is also of bounded variation. Proof Let f and g be two functions of bounded variation on [a, b] . Then for any partition P of [a, b] we have

å ( f + g )(x ) - ( f + g )(x

i

i -1

) =å

{ f ( xi ) + g ( xi )} - { f ( xi -1 ) + g ( xi -1 )}

£ å f ( xi ) - f ( xi-1 ) + å g ( xi ) - g ( xi-1 ) £ V ( f ; a, b) + V ( g ; a, b) Þ V ( f + g ; a, b ) £ V ( f ; a, b ) + V ( g ; a, b) This show that the function f + g is of bounded variation. Similarly it can be shown that f - g is also of bounded variation. i.e. V ( f - g ) £ V ( f ) + V ( g )

= å f ( xi ) - f ( xi -1 ) + g ( xi ) - g ( xi-1 )

q

Note (i) If f and g are monotonic increasing on [a, b] then ( f - g ) is of bounded variation on [a, b] . (ii) If c is constant, the sums å f ( xi ) - f ( xi -1 ) and therefore the total variation function, V ( f ) is same for f and f - c . 2) The product of two functions of bounded variation is also of bounded variation. Proof Let f and g be two functions of bounded variation on [a, b] . Þ f and g are bounded and $ a number k such that f ( x ) £ k & g ( x) £ k " x Î [ a, b] . For any partition P of [a, b] we have

4

Chap 7 – Functions of bounded variation.

£ k V (g ) + k V ( f ) Þ fg is of bounded variation on [a, b] .

å ( fg ) ( x ) - ( fg ) ( x ) = å f (x )g (x ) - f (x )g (x ) = å f (x )g(x ) - f (x )g (x ) + f (x )g(x ) - f (x )g (x = å f ( x ){ g ( x ) - g ( x )} + g ( x ){ f ( x ) - f ( x )} £ å f (x ) g (x ) - g (x ) + å g (x ) f (x ) - f (x ) £ k å g(x ) - g(x ) + k å f (x ) - f (x )

i i -1 i i i -1 i -1 i i i i i -1 i i -1 i -1 i i -1 i -1 i i -1 i i i -1 i -1 i i -1 i i -1 i i -1

i -1

)

q

Ð Note Theorems like the above, could not be applied to quotients of functions because the reciprocal of a function of bounded variation need not be of bounded variation. 1 will not be bounded and therefore can not e.g. if f ( x ) ® 0 as x ® x0 , then f ( x) be of bounded variation on any interval which contains x0 . Therefore to consider quotient, we avoid functions whose values becomes arbitrarily close to zero. 3) If f is a function of bounded variation on [a, b] and if $ a positive number k 1 such that f ( x ) ³ k " x Î [a, b] then is also of bounded variation on [a, b] . f Proof For any partition P of [a, b] , we have

å

1 1 ( xi ) - ( xi-1 ) f f

=å =å £

1 1 f ( xi ) f ( xi-1 ) f ( xi -1 ) - f ( xi ) f ( xi ) f ( xi-1 ) £ 1 V ( f ; a, b ) k2 q

1 å f ( xi-1 ) - f ( xi ) k2

Þ

1 is of bounded variation on [a, b] . f

4) If f is of bounded variation on [a, b] , then it is also of bounded variation on [a, c ] and [c, b] , where c is a point of [a, b] , and conversely. Also V ( f ; a , b ) = V ( f ; a , c ) + V ( f ; c, b ) . Proof a) Let, first, f be of bounded variation on [a, b] . Take P = {a = x0 , x1 ,....., xm = c} & P2 = {c = y0 , y1 ,....., yn = b} any two 1 partitions of [a, c ] and [c, b] respectively. Evidently, P = P È P2 = {a = x0 ,...., xm , y0 ,...., yn = b} is a partition of [a, b] . 1 We have n ìm ü f ( xi ) - f ( xi -1 ) + å f ( yi ) - f ( yi-1 ) ý £ V ( f ; a, b ) íå î i =1 i =1 þ

Chap 7 – Functions of bounded variation.

5

Þ and

å å

i =1 i =1 n

m

f ( xi ) - f ( xi -1 ) £ V ( f ; a, b ) f ( yi ) - f ( yi -1 ) £ V ( f ; a, b )

Þ f is of bounded variation on [a, c ] and [c, b] both. b) Let, now, f be of bounded variation on [a, c ] and [c, b] both. Let P = {a = z0 , z1 ,....., zn = b} be a partition of [a, b] . If it does not contain the point c , let us consider the partition P* = P È {c} c Let c Î [ zr -1 , zr ] i.e. zr -1 £ c £ zr , r < n zr – 1 a z1 zr Then

å

i =1

n

f ( zi ) - f ( zi -1 )

= å f ( zi ) - f ( zi -1 ) + f ( zr ) - f ( zr -1 ) + £ å f ( zi ) - f ( zi -1 ) + f (c) - f ( zr -1 )

i =1 i =1 r -1

r -1

i = r +1

å

n

n

f ( zi ) - f ( zi-1 )

+ f ( zr ) - f ( c ) +

£ V ( f ; a, c ) + V ( f ; c, b ) Þ f is of bounded variation on [a, b] if it is of bounded variation on [ a, c ] & [c, b] both, then V ( f ; a, b ) £ V ( f ; a, c ) + V ( f ; c, b ) …………. (i) Now let e > 0 be any arbitrary number. Since V ( f ; a, c ) and V ( f ; c, b ) are the total variation of f on [ a, c ] & [c, b] respectively therefore $ partition P = {a = x0 , x1 , x2 ,....., xm = c} and 1 P2 = {c = y0 , y1 , y3 ,......, yn = b} of [ a, c ] & [c, b] respectively such that

i = r +1

å

f ( zi ) - f ( zi -1 )

e ……………. (ii) 2 i =1 n e & å f ( yi ) - f ( yi-1 ) > V ( f ; c, b ) - 2 …………… (iii) i =1 Adding (ii) and (iii) we get

å

m

f ( xi ) - f ( xi-1 ) > V ( f ; a, c ) -

å

i =1

m

f ( xi ) - f ( xi -1 ) + å f ( yi ) - f ( yi-1 ) > V ( f ; a, c ) + V ( f ; c, b ) - e

i =1

n

**Þ V ( f ; a, b ) > V ( f ; a, c ) + V ( f ; c, b ) - e But e is arbitrary positive number therefore we get V ( f ; a, b ) ³ V ( f ; a, c ) + V ( f ; c, b ) …………….. (iv) From (i) and (iv), we get V ( f ; a, b ) = V ( f ; a, c ) + V ( f ; c , b )
**

||||||||||||||||||||||||||||

q

6

Chap 7 – Functions of bounded variation.

Ð Variation Function Let f be a function of bounded variation on [a, b] and x is a point of [a, b] . Then the total variation of f is V ( f ; a, x ) on [a, x] , which is clearly a function of x, is called the total variation function or simply the variation function of f and is denoted by V f ( x) , and when there is no scope for confusion, it is simply written as V ( x) . Thus V f ( x) = V ( f ; a, x ) ; (a £ x £ b) If x1 , x2 are two points of the interval [a, b] such that x2 > x1 , then 0 £ f ( x2 ) - f ( x1 ) £ V ( f ; x1 , x2 ) = V ( f ; a, x1 ) - V ( f ; a, x2 ) = V f ( x2 ) - V f ( x1 ) Þ V f ( x2 ) ³ V f ( x1 ) implies that the variation function is monotonically increasing function on [a, b] .

CHARACTERIZATION OF FUNCTIONS OF BOUNDED VARIATION

**Ð Theorem A function of bounded variation is expressible as the difference of two monotonically increasing function.
**

Proof

We have 1 1 (V ( x) + f ( x) ) - (V ( x) - f ( x) ) 2 2 = G ( x) - H ( x ) (say) We shall prove that these two functions G ( x) and H ( x ) are monotonically increasing on [a, b] . Now, if x2 > x1 , we have 1 G ( x2 ) - G ( x1 ) = [V ( x2 ) - V ( x1 ) + f ( x2 ) - f ( x1 )] 2 1 = éV ( f ; x1 , x2 ) - ( f ( x1 ) - f ( x2 ) ) ù û 2ë Since V ( f ; x1 , x2 ) ³ f ( x1 ) - f ( x2 ) Þ G ( x2 ) - G ( x1 ) ³ 0 i.e. G ( x2 ) ³ G ( x1 ) so that the function G ( x) is monotonically increasing on [a, b] . Again, we have 1 H ( x2 ) - H ( x1 ) = é(V ( x2 ) - V ( x1 ) ) - ( f ( x2 ) - f ( x1 ) ) ù û 2ë 1 = éV ( f ; x1 , x2 ) - ( f ( x2 ) - f ( x1 ) ) ù û 2ë so that as before H ( x2 ) - H ( x1 ) ³ 0 i.e. H ( x2 ) ³ H ( x1 ) . i.e. H ( x ) is also monotonically increasing function. Hence the result. f ( x) = Ð Note A function f ( x) is of bounded variation over the interval [a, b] iff it can be expressed as the difference of two monotonically functions.

q

Chap 7 – Functions of bounded variation.

7

**Ð Theorem Let f be of bounded variation on [a, b] . Let V be defined on [a, b] as follows: V ( x ) = V f ( x) = V ( f ; a, x) ) if a < x £ b , V (a ) = 0 . Then i) V is an increasing function on [a, b] . ii) (V - f ) is an increasing function on [a, b] .
**

Proof

If a < x < y £ b , we can write a x V ( f ; a, y ) = V ( f ; a, x) - V ( f ; x, y ) Þ V ( y ) - V ( x ) = V ( f ; x, y ) Q V ( f ; x, y ) ³ 0 \ V ( y ) - V ( x ) ³ 0 Þ V ( x) £ V ( y ) and (i) holds. To prove (ii), let D ( x) = V ( x) - f ( x) if x Î [a, b] . Then, if a £ x < y £ b , we have D( y ) - D( x) = [V ( y ) - V ( x) ] - [ f ( y ) - f ( x )]

y

b

= V ( f ; x, y ) - [ f ( y ) - f ( x) ] But from the definition of V ( f ; x, y ) , it follows that f ( y ) - f ( x ) £ V ( f ; x, y ) This means that D ( y ) - D ( x) ³ 0 and (ii) holds.

q

Ð Theorem If c be any point of [a, b] , then V ( x ) is continuous at c if and only if f ( x) is continuous at c . i.e. A point of continuity of f ( x) is also a point of continuity of V ( x ) and conversely.

Proof

Firstly suppose that V ( x ) is continuous at c . Let e > 0 be given, then $ d > 0 such that V ( x ) - V (c ) < e for x - c < d ………… (i) Also, we have f ( x ) - f (c ) £ V ( x ) - V (c ) if x > c …………… (ii) And f ( x ) - f (c ) £ V (c ) - V ( x ) if x < c …………… (iii) From (i), (ii) and (iii), we deduce that f ( x ) - f (c ) £ V ( x ) - V (c ) < e for x - c < d Which shows that f ( x) is continuous at c . Now suppose that c is a point of continuity of f ( x) and let e > 0 be given, then $ d > 0 such that e f ( x ) - f (c ) < for x - c < d 2 Also $ a partition P = {c = y0 , y1 ,....., yq -1 , yq ,....., yn = b} of [c, b] such that 1 f ( yq ) - f ( yq -1 ) > V ( f ; c, b) - e …………… (iv) 2 q =1 Since as a result of introducing addition points to the partition P , the corresponding sum of the moduli of the differences of the function values at end points will not be decreased, therefore we may assume that

å

n

8

Chap 7 – Functions of bounded variation.

0 < y1 - c < d so that f ( y1 ) - f (c) <

e ……………. (v) 2

**Thus (iv) becomes n 1 1 1 V ( f ; c, b) - e < e + å f ( yq ) - f ( yq -1 ) < e + V ( f ; y1 , b) 2 2 2 q =2 Þ V ( f ; c, b ) - V ( f ; y1 , b) < e Þ V ( y1 ) - V (c) < e Thus for 0 < y1 - c < d , we have 0 < V ( y1 ) - V (c) < e \ lim V ( x ) = V (c)
**

x ®c + 0

**Similarly, we can have lim V ( x) = V (c )
**

x ®c -0

Which shows that V ( x ) is continuous at c . Ð Note V ( x ) is continuous in [a, b] iff f ( x) is continuous in [a, b] .

q

Ð Corollary A function f is of bounded variation on [a, b] iff there is a bounded increasing function g on [a, b] such that for any two points x¢ and x¢¢ in [a, b] , x¢ < x¢¢ , we have f ( x¢¢) - f ( x¢) £ g ( x¢¢) - g ( x¢) Moreover, if g is continuous at x¢ , so is f .

Proof

ìV x , a < x £ b Take g ( x) = í a î 0 , x=a Then g is increasing and bounded on [a, b] . Also, f ( x¢) - f ( x¢¢) £ Vxx¢¢ ( f ) = g ( x¢¢) - g ( x¢) ¢ Which also yields that if g is continuous at x¢ , so is f . Ð Question Show that the function f defined by 1 ì 2 ï x sin f ( x) = í x ï 0 î is of bounded variation on [ 0,1] .

Solution

q

; x¹0 ; x=0

f is differentiable on [ 0,1] and f ¢( x ) = 2 x sin Also 1 + sin x £ 2 + 1 = 3 x f ¢( x ) is bounded on [ 0,1] f ¢( x ) £ 2 x sin

1 - sin x for 0 £ x £ 1 . x

i.e.

Hence f is of bounded variation on [ 0,1] .

q

Chap 7 – Functions of bounded variation.

9

**Ð Question Show that
**

Solution

px ì ï x cos g ( x) = í 2 ï 0 î

, 0 < x £1 , x=0

is not of bounded variation on [ 0,1]

1 1 1 ü ì 1 Let P = í0, , ,....., , ,1ý be a partition of [ 0,1] . 3 2 þ î 2n 2n - 1 Then å f ( xi ) - f ( xi-1) æ1ö æ1ö æ1ö æ1ö æ1ö æ 1 ö = f (1) - f ç ÷ + f ç ÷ - f ç ÷ + f ç ÷ - f ç ÷ + ...... + f ç ÷ - f ( 0 ) è2ø è2ø è3ø è3ø è4ø è 2n ø p 1 p 1 p 1 p 1 p 1 p 1 p = cos - cos + cos - cos + cos - cos + ..... + cos - 0 2 2 4 2 4 3 6 3 6 4 8 2n 4n pö p ö æ1 pö p ö æ1 æ1 æ 1 = 2 ç cos ÷ + 2 ç cos ÷ + 2 ç cos ÷ + ..... + 2 ç cos ÷ 4ø 6ø 8ø 4n ø è2 è3 è4 è 2n p 1 p 1 p 1 p ö æ1 = 2 ç cos + cos + cos + ...... + cos ÷ 4 3 6 4 8 2n 4n ø è2 which is not bounded. Hence f ( x) is not of bounded variation on [ 0,1] . q

Alternative

We have g ( xk +1 ) - g ( xk ) + g ( xk ) - g ( xk -1 ) = 1 (k + 1)p 1 kp 1 kp 1 (k - 1) p cos - cos + cos cos k +1 2 k 2 k 2 k -1 2 ; ; if k is even if k is odd

2 ì ï ï k =í ï 1 + 1 ïk +1 k -1 î Þ Vab ( g ) £ Q 1 åk £ k =1

n

åk

k =1

¥

1

åk

k =1

¥

1

is divergent \ Vab ( g ) is not finite. q

Hence g is not of bounded variation. References: (1) Lectures & Notes (Year 2003-04) Prof. Syyed Gul Shah

Chairman, Department of Mathematics. University of Sargodha, Sargodha.

Made by: Atiq ur Rehman (atiq@mathcity.org) Available online at http://www.mathcity.org in PDF Format. Page Setup: Legal ( 8¢¢ 1 2 ´ 14¢¢ ) Printed: 30 April 2004 (Revised: March 19, 2006.) Submit error or mistake at http://www.mathcity.org/error