main 2007/3/6 18:25 page 39 #1

Bull Braz Math Soc, New Series 38(1), 39-50

2007, Sociedade Brasileira de Matemtica

Quadrature formulas for the generalized

Riemann-Stieltjes integral
Marilena Munteanu*
Abstract. In this paper we propose a technique of approximation for the generalized
Riemann-Stieltjes integral and we found an analogue for Newton-Cotes formulas in the
case n = 2 and n = 3.
Keywords: Riemann Stieltjes integral, Weyl derivative, quadrature formula.
Mathematical subject classification: 65D32.
1 Introduction
The study of nonlinear and integral equations leads to consider the fixed point
problem
u = Tu
(1)
with T a completely continuous operator acting on a Hilbert space H . I. Moret
& P. Omari in [3] were concerned with the numerical solution of (1) by iterative
techniques based on linearization. These procedures consist of approximating
(1) by linear equations of the form
(I Am )(u u m ) = u m + T u m

(2)

where u m is the current iterative and Am is a suitable linear model of T at u m .

The next iterate u m+1 is defined as the unique solution of (2). In the financial
mathematics, for examples, there are studied the equations of the form
X (t, ) = X 0 () +

Zt
0

(s, X (s, ))d B(s, ) +

Zt

b(s, X (s, ))ds

(3)

Received 12 August 2005.

*Beneficiary of a Socrates fellowship at the Department of Mathematics, University of Study of
Cagliari, Via Ospedale, n. 72, Cagliari, 09124, Italy, in the period February July 2002.

main 2007/3/6 18:25 page 40 #2

40

MARILENA MUNTEANU

Rt
where 0 (s, X (s, ))d B(s, ) is the stochastic integral which can be defined
Rint various manner depending on the exponent of the Brownian motion and
0 b(s, X (s, ))ds is the classical integral. If the exponent of the Brownian
motion is bigger than 21 (situation requested by practical necessities) the stochastic integral can be defined by trajectory, i.e. for all fixed, we have
I =

Zt

(s, X (s))d B(s).

(4)

In this situation, the Brownian motion is a process -Hlder continuous, with

1
< < 1. In order to apply the algorithm described by Moret & Omari, given
2
the iteration at the step m, we should be able to compute the right hand side
of (3), more precisely to approximate the stochastic integral.
The hypothesis concerning the function are: is Hlder continuous
with respect to the first argument and is Lipschitz with respect to the second
argument,
by consequence, the integral (4) is a Riemann-Stieltjes integral (RS)
Rb
a f dg where f is Hlder continuous and g is Hlder continuous. D.
Nualart & A. Rascanu proved a global existence and uniqueness of the solutions
for integral equations containing integrals of type (RS) (see for details [4]). In
the following we use the definition of the Riemann-Stieltjes integral given by M.
Zhle [5] and we propose a technique of the approximation. More precisely, we
give the following results:
Theorem A. Let f and g be two Hlder continuous functions on [a, b], namely
| f (x) f (y)| M f |x y| and |g(x)g(y)| Mg |x y| for all x, y [a, b]
where M f , Mg are positive constants and + > 1. Let us consider the nodes
a = x1 < x2 = b and denote h = b a. Then
Zb

f (x)dg(x) =

1
(g(x2 ) g(x1 )) ( f (x1 ) + f (x1 ) + f (x2 )) + R2
2

where |R2 | cM f Mg h + with c a positive constant.

Bull Braz Math Soc, Vol. 38, N. 1, 2007

(5)

main 2007/3/6 18:25 page 41 #3

QUADRATURE FORMULAS FOR THE GENERALIZED RIEMANN-STIELTJES INTEGRAL

41

Theorem B. Consider as above f, g and the nodes a = x1 < x2 < x3 = b

and denote h = ba
. We have
2
Zb

f (x)dg(x) =

1
[ f (x1 ) + 4 f (x2 ) + f (x3 )] [g(x3 ) g(x1 )]
6

[g(x1 ) + 4g(x2 ) + g(x3 )] [ f (x3 ) f (x1 )]
+

(6)

1
[g(x3 ) + g(x1 )] [ f (x3 ) f (x1 )] + R3
2

where |R3 | cM f Mg h + with c a positive constant.

2 Preliminaries
In this section we present the definition of Riemann Stieltjes integral by using
Weyl derivatives and the method of approximation. Let (0, 1) be an arbitrary
number and denote
f a+ (x) = ( f (x) f (a+))1(a,b) (x) ,

gb (x) = (b(x) g(b))1(a,b) (x)

(1) = ei

(throughout of this paper we denote f (a+) = lim f (a +), g(b) = lim g(b
0

) supposing that the limits exist).

We define the integral by
Zb

f (x)dg(x) = (1)

Zb

Da+ f a+ (x)Db gb (x)d x

+ f (a+)(g(b) g(a+))
where

Zx
1
f
(x)
f
(x)

f
(y)

Da+ f (x) =
+
dy 1(a,b) (x)
0(1 ) (x a)
(x y) +1
a

and

Zb

f
(x)
f
(x)

f
(y)
(1)

+
dy 1(a,b) (x)
Db f (x) =
0(1 ) (b x)
(y x) +1
x

Bull Braz Math Soc, Vol. 38, N. 1, 2007

(7)

main 2007/3/6 18:25 page 42 #4

42

MARILENA MUNTEANU

are Weyl representations for the fractional

R derivatives in the sense of Riemann
and Liouville (e.g. [5]). Here 0(t) = 0 x t1 ex d x is Eulers function. Remark
that the definition is good, namely it does not depend on the choice of the parameter (0, 1). See [5] for a detailed analysis of integrals defined by (7). Let f
and g be such that | f (x) f (y)| M f |x y| and |g(x) g(y)| Mg |x y| ,
with , (1/2, 1). Consider the nodes a x1 < x2 < < xn b. We
approximate the functions f and g with the Lagrange interpolating polynomials (namely we have f (x) = L n f (x) + R f (x) and g(x) = L n g(x) + Rg (x),
respectively). By using additivity properties of the Riemann Stieltjes integral
we get
Zb

f (x)dg(x) =

Zb

Zb

(L n f (x) + R f (x))d(L n g(x) + Rg (x))

L n f (x)d L n g(x) +

Zb

R f (x)d L n g(x)

Zb

L n f (x)d Rg (x) +

Zb

R f (x)d Rg (x)

= I1 + I2 + I3 + I4 .

We denote by wn (x) = (x x1 ) (x xn ). Recall that the Lagrange

interpolating polynomial for the function f is
L n f (x) =

n
X
i=1

f (xi )

wn (x)
,
(x xi )wn0 (xi )

and similarly for g. Consequently,

I1 =

Zb

L n f (x)d L n g(x) =

n X
n
X
i=1 k=1

Zb

L n f (x)L 0n g(x)d x

Zb

wn (x) wn0 (x)(x xk ) wn (x)

d x.
x xi
(x xk )2

f (xi )g(xk )
wn0 (xi )wn0 (xk )

Denoting by
I1ik

1
= 0
wn (xi )wn0 (xk )

Bull Braz Math Soc, Vol. 38, N. 1, 2007

Zb
a

wn (x) wn0 (x)(x xk ) wn (x)

d x,
x xi
(x xk )2

(8)

main 2007/3/6 18:25 page 43 #5

QUADRATURE FORMULAS FOR THE GENERALIZED RIEMANN-STIELTJES INTEGRAL

43

we can approximate the integral by the sum

I1 =

n
n X
X

I1ik f (xi )g(xk )

(9)

i=1 k=1

and the rest of the quadrature formula is Rn = I2 + I3 + I4 .

3 Quadrature formulas proofs of theorems
In this section we give an analogue for Newton-Cotes formulas in the cases n = 2
and n = 3.
3.1

Case n=2 (trapezoidal method)

We have x1 = a, x2 = b, w2 (x) = (x a)(x b) and w20 (x) = 2x a b.

The coefficients of the quadrature formula I1ik , i, k {1, 2} are:
1
I111 = ,
2

I112 =

1
,
2

1
I121 = ,
2

I122 =

1
.
2

Consequently,
I1 =
=

1
1
1
1
f (x1 )g(x1 ) + f (x1 )g(x2 ) f (x2 )g(x1 ) + f (x2 )g(x2 )
2
2
2
2

1
(g(x2 ) g(x1 ))( f (x1 ) + f (x2 )).
2

In order to estimate the rest R2 we will evaluate R f and Rg (the rests of the
Lagrange interpolation of the two functions). By taking into account that g is
Hlder continuous, we have


 Rg (x) = |[x1 , x2 , x]g (x x1 )(x x2 )|





g(x) g(x2 )
g(x2 ) g(x1 )

= 

(x x1 )(x x2 )
(x x2 )(x x1 ) (x2 x1 )(x x1 )


Mg |x x2 |
Mg |x2 x1 |
+
(x x1 )(x x2 )

(x2 x)(x x1 ) (x2 x1 )(x x1 )

= Mg (x2 x) + (x2 x1 ) 2Mg h
In the same manner we obtain |R f (x)| 2M f h .
Bull Braz Math Soc, Vol. 38, N. 1, 2007

main 2007/3/6 18:25 page 44 #6

44

MARILENA MUNTEANU

The integral I2 and I3 can be evaluated in the same way. We will write down
only the estimation for I2 . We have
I2 =

Zb

R f (x)d L 2 g(x) =

L 2 g(x) = g(x1 )
One gets
b

R f (x)L 02 g(x)d x

where

Zb

x x2
x x1
+ g(x2 )
.
x1 x2
x2 x1
Z

1
|g(x2 ) g(x1 )|d x
h
a
a
= 2M f h |g(x2 ) g(x1 )| 2M f h Mg |x2 x1 | = 2M f Mg h + .

|I2 |

|R f (x)L 02 g(x)|d x

2M f h

Similarly |I3 | 2M f Mg h + .
The main point of this subsection is to estimate I4 . By using the definition of
the Riemann Stieltjes integral one has:
Zb

R f d Rg = (1)

Zb

Da+ R fa+ (x)Db Rgb (x)d x

where

Zx
R
(x)
R
(x)

R
(y)
1
f a+
f a+

fa+
+
dy
Da+ R fa+ (x) =
(x y) +1
0(1 ) (x a)
a

:=

1
F
0(1 )

and
1

Db Rgb (x) =

(1)
0( )

:=

Rgb (x)
+ (1 )
(b x)1

(1)
G.
0( )

1
With these notations |I4 |
0(1 )0( )
Bull Braz Math Soc, Vol. 38, N. 1, 2007

Zb
a

Zb
x

Rgb Rgb (y)

dy
(y x)2

|F| |G|d x .

main 2007/3/6 18:25 page 45 #7

QUADRATURE FORMULAS FOR THE GENERALIZED RIEMANN-STIELTJES INTEGRAL

45

Since R fa+ := (R f (x) R f (a+))1(a,b) (x) = R f (x)1(a,b) (x) and R f (a) = 0

we deduce
|R fa+ (x)|
|R f (x)|
|R f (x) R f (a)|
=
=
.
(x a)
(x a)
(x a)
In our purpose we establish the following inequality
f (x2 ) f (x1 )
(x y)|
x2 x1


 f (x2 ) f (x1 ) 

 |x y|
| f (x) f (y)| + 

x2 x1

|R f (x) R f (y)| = | f (x) f (y) +

M f |x y| + M f h 1 |x y| .

By straightforward computations we have |Rg (x) Rg (y)| Mg |x y| +

Mg h 1 |x y| and consequently


|R fa+ (x)|
1
M f |x a| + M f h 1 |x a|

(x a)
(x a)
= M f |x a| + M f h 1 |x a|1 .
Due to the arbitrariness of we set < . It follows
|R fa+ (x)|
2M f h .
|x a|
So, we can write
|R fa+ (x)|
|F|
+
(x a)

Zx
a

|R fa+ (x) R fa+ (y)|

dy .
(x y) +1

The next step is to evaluate the expression

|R fa+ (x) R fa+ (y)|

. We may write
(x y) +1

|R fa+ (x) R fa+ (y)|

|R f (x) R f (y)|
=
(x y) +1
(x y) +1

M f |x y| + M f h 1 |x y|
= Mf

|R fa+ (x)|
Thus |F|
+
|x a|

Bull Braz Math Soc, Vol. 38, N. 1, 2007

1
(x y) +1

1
1
+ M f h 1
1+

|x y|
|x y|

1
1
Mf
+ M f h 1
1+

|x y|
|x y|

dy.

main 2007/3/6 18:25 page 46 #8

46

MARILENA MUNTEANU

Let us remark that the condition imposed < assures the convergence of
the (improper) integrals which appear in the above formula. At this point we can
conclude

1 
(x a)

1 (x a)
|F| 2M f h
+ Mf
+ Mfh

1



1
1
+
h .
Mf 2 +

1
In the following we will do an analogue calculus for G. Choosing > 1
(recall that is arbitrary) we obtain
|G| 2Mg h + 1 + (1 )Mg

 Zb

Zb

1
dy
(y x)2

1
dy
(y x)1
x



1
1
Mg 2 + (1 )
+
h + 1
+ 1
+ h 1

(the condition > 1 assures the convergence of the integrals).

At this moment we fix (1 , ) and consequently
|I4 |

1
M f Mg K h + ,
0(1 )0( )

where K is a constant depending on and .

Since
Z 
Z
2
1
x2
2
0( )0(1 ) =
x e
dx
0

Z

x 2 ex d x

2

=0

x 2 e 2

 2
1
2

2

dx

we have
1
1

0( )0(1 )
0( 21 )2

Bull Braz Math Soc, Vol. 38, N. 1, 2007

and |I4 |

K M f Mg +
h
.
0( 21 )2

main 2007/3/6 18:25 page 47 #9

QUADRATURE FORMULAS FOR THE GENERALIZED RIEMANN-STIELTJES INTEGRAL

At the end we can conclude the estimation of the rest

|R| |I2 | + |I3 | + |I4 | M f Mg

K
4+ 1 2
0( 2 )

h + .

47

(10)

In order to obtain a better approximation, the quadrature formula must be iterate.

With this end in view we will consider a division of the interval [a, b] of the form
a = 1 < 2 < < m = b. The fact that g is (Hlder) continuous function,
allows as to write
Zb

f (x)dg(x) =

m1
X

Z j+1

f (x)dg(x).

j=1
j

Consider m = 2 N + 1, i.e. the interval [a, b] was divided into 2 N subintervals.

Suppose that i+1 i = 21N and denote by R2,i the rest of the quadrature formula
for the interval [i , i+1 ]. The total rest is
!
2N
2N
X
X
K
+
|R2 |
|R2,i |
M f Mg 4 + 1 2 h i
0(
)
2
i=1
i=1
!
K
= M f Mg 4 + 1 2 2(1)N
0( 2 )
!
K
1
= M f Mg 4 + 1 2
(+1)N
0( 2 ) 2
!
K
= M f Mg 4 + 1 2 h +1 .
0( 2 )
The condition + > 1 yields to lim R2 = 0.
N

3.2

Case n=3 (Simpson formula)

The coefficients of the quadrature formula are:

1
2
1
I111 = , I112 = , I113 = ,
2
3
6
2
2
23
22
23
I1 = , I1 = 0 , I1 = ,
3
3
1
2
1
31
32
33
I1 = , I1 = , I1 = .
6
3
2
Bull Braz Math Soc, Vol. 38, N. 1, 2007

main 2007/3/6 18:25 page 48 #10

48

MARILENA MUNTEANU

In this case, the quadrature formula is:

I1 =

1
[ f (x1 ) + 4 f (x2 ) + f (x3 )] [g(x3 ) g(x1 )]
6

[g(x1 ) + 4g(x2 ) + g(x3 )] [ f (x3 ) f (x1 )]
1
+ [g(x3 ) + g(x1 )] [ f (x3 ) f (x1 )] .
2

(11)

We will work as in previous case. After a straightforward computation we

obtain
|R f (x)| 6M f h , |Rg (x)| 6Mg h ,

|L 3 f 0 (x)| 2M f h 1 , |L 3 g 0 (x)| 2Mg1 .

Hence
|I2 |

Zb

|R f | |L 03 g(x)|d x 24M f Mg h +

and

|I3 | 24M f Mg h + .

To finalize we have to evaluate I4 . The following estimations hold:

|R f (x) R f (y)| M f |x y| + 2M f h 1 |x y|,
|R f (x) R f (y)|
M f (|x y| 1 + 2h 1 |x y| )
|x y| +1

1
M f (|x a| + 2h 1 |x a|)

|Da+ R f a+ (x)|
0(1 )
|x a|

Zx
1
1

+
M f |x y|
+ 2h |x y| dy
a

Mf

2
(2 + 22 +
+
)h .

0(1 )

1
The integrals are convergent if we take < . Similarly,

Mg
1
|Db Rgb (x)|
(21+ + 21+ )h 1+
0( )
 
Zb 
1
2h 1
+ (1 )
+
dy
(y x)2
(y x)1
x

Bull Braz Math Soc, Vol. 38, N. 1, 2007

main 2007/3/6 18:25 page 49 #11

QUADRATURE FORMULAS FOR THE GENERALIZED RIEMANN-STIELTJES INTEGRAL

49

10

10

10

10

10

10

(,)=(3/4,5/6)
(,)=(5/6,8/9)
(,)=(7/8,11/12)
(,)=(10/11,17/18)

10

10

10

10

10

10

10

10

10

10

10

(,)=(3/4,5/6)
(,)=(5/6,8/9)
(,)=(7/8,11/12)
(,)=(10/11,17/18)

10

10

10

10

10

10

10

Figure 1: Logarithmic representation of the errors versus h for different values

of and ; the trapezoidal method on the left side and Simpson method on the
right side.
Bull Braz Math Soc, Vol. 38, N. 1, 2007

main 2007/3/6 18:25 page 50 #12

50

MARILENA MUNTEANU


Mg
1

21+ + 21+ +
2+ 1
0( )
+ 1

+1 1
+ 2

h + 1

with 1 < .
Thus we have obtained an estimation for I4
|I4 |

2M f Mg c(, , ) +
h
.
0( )0(1 )

In the end, we may conclude that

|R3 | = |I2 + I3 + I4 | |I2 | + |I3 | + |I4 | cM f Mg h + .

(12)

As in the case n = 2, for the iterate formula we have lim R3 = 0.

n

In order to illustrate the behavior of the quadratures formulas we considered

f (x) = x , g(x) = x and we present some numerical results. (see Figure 1).

Acknowledgement.
useful comments.

The author thanks the anonymous referee for his or her

References
[1]
[2]
[3]
[4]
[5]

H. Brezis, Analyse fonctionelle, Thorie et applications, Masson, Paris, 1983.

R. Kress, Numerical Analysis Graduate Texts in Mathematics, Springer-Verlag,
New York, 1998.
I. Moret & P. Omari, A quasi-Newton method for solving fixed point problems in
Hilbert spaces, Numer. Math. 59 (1991), 159177.
D. Nualart & A. Rascanu, Differential equations driven by factorial Brownian
motion, Collect. Math. 53(1) (2002), 5581.
M. Zhle, Integration with respect to fractal function and stochastic calculus. I,
Probab. Theory Relat. Fields, 111 (1998), 333374.

Marilena Munteanu
University Al. I. Cuza Iasi
Faculty of Mathematics
Bd. Carol I, no. 11
700506 Iasi
ROMNIA
E-mail: marilena_munteanu@hotmail.com

Bull Braz Math Soc, Vol. 38, N. 1, 2007