Chapter 4

Numerical Differentiation and Integration

Numerical Differentiation

Forward and Backward Differences

Inspired by the definition of derivative:

𝑓(𝑥0 + ℎ) − 𝑓(𝑥0 )
𝑓 ′ (𝑥0 ) = lim ,
ℎ→0 ℎ
choose a small ℎ and approximate

𝑓(𝑥0 + ℎ) − 𝑓(𝑥0 )
𝑓 ′ (𝑥0 ) ≈
ℎ
The error term for the linear Lagrange polynomial gives:

𝑓(𝑥0 + ℎ) − 𝑓(𝑥0 ) ℎ ″
𝑓 ′ (𝑥0 ) = − 𝑓 (𝜉)
ℎ 2
Also known as the forward-difference formula if ℎ > 0 and the
backward-difference formula if ℎ < 0
General Derivative Approximations

Differentiation of Lagrange Polynomials

Differentiate
𝑛
(𝑥 − 𝑥0 ) ⋯ (𝑥 − 𝑥𝑛 ) (𝑛+1)
𝑓(𝑥) = ∑ 𝑓(𝑥𝑘 )𝐿𝑘 (𝑥) + 𝑓 (𝜉(𝑥))
𝑘=0
(𝑛 + 1)!

to get
𝑛
′
𝑓 (𝑛+1) (𝜉(𝑥𝑗 ))
𝑓 (𝑥𝑗 ) = ∑ 𝑓(𝑥𝑘 )𝐿′𝑘 (𝑥𝑗 ) + ∏(𝑥 − 𝑥𝑘 )
𝑘=0
(𝑛 + 1)! 𝑘≠𝑗 𝑗

This is the (𝑛 + 1)-point formula for approximating 𝑓 ′ (𝑥𝑗 ).

Commonly Used Formulas

Using equally spaced points with ℎ = 𝑥𝑗+1 − 𝑥𝑗 , we have the

three-point formulas
1 ℎ2
𝑓 ′ (𝑥0 ) = [−3𝑓(𝑥0 ) + 4𝑓(𝑥0 + ℎ) − 𝑓(𝑥0 + 2ℎ)] + 𝑓 (3) (𝜉0 )
2ℎ 3
2
1 ℎ
𝑓 ′ (𝑥0 ) = [−𝑓(𝑥0 − ℎ) + 𝑓(𝑥0 + ℎ)] − 𝑓 (3) (𝜉1 )
2ℎ 6
1 ℎ2
𝑓 ′ (𝑥0 ) = [𝑓(𝑥0 − 2ℎ) − 4𝑓(𝑥0 − ℎ) + 3𝑓(𝑥0 )] + 𝑓 (3) (𝜉2 )
2ℎ 3
2
1 ℎ
𝑓 ″ (𝑥0 ) = 2 [𝑓(𝑥0 − ℎ) − 2𝑓(𝑥0 ) + 𝑓(𝑥0 + ℎ)] − 𝑓 (4) (𝜉)
ℎ 12
and the five-point formula
1
𝑓 ′ (𝑥0 ) = [𝑓(𝑥0 − 2ℎ) − 8𝑓(𝑥0 − ℎ) + 8𝑓(𝑥0 + ℎ) − 𝑓(𝑥0 + 2ℎ)]
12ℎ
ℎ4
+ 𝑓 (5) (𝜉)
30
Optimal ℎ

Consider the three-point central difference formula:

1 ℎ2
𝑓 ′ (𝑥0 ) = [𝑓(𝑥0 + ℎ) − 𝑓(𝑥0 − ℎ)] − 𝑓 (3) (𝜉1 )
2ℎ 6
Suppose that round-off errors 𝜀 are introduced when computing
𝑓. Then the approximation error is

̃ + ℎ) − 𝑓(𝑥
𝑓(𝑥 ̃ − ℎ) 𝜀 ℎ2
0 0
∣𝑓 ′ (𝑥0 ) − ∣ ≤ + 𝑀 = 𝑒(ℎ)
2ℎ ℎ 6

where 𝑓 ̃ is the computed function and |𝑓 (3) (𝑥)| ≤ 𝑀

Sum of truncation error ℎ2 𝑀 /6 and round-off error 𝜀/ℎ
Minimize 𝑒(ℎ) to find the optimal ℎ = √ 3
3𝜀/𝑀
Richardson’s Extrapolation

Suppose 𝑁 (ℎ) approximates an unknown 𝑀 with error

𝑀 − 𝑁 (ℎ) = 𝐾1 ℎ + 𝐾2 ℎ2 + 𝐾3 ℎ3 + ⋯

then an 𝑂(ℎ𝑗 ) approximation is given for 𝑗 = 2, 3, … by

ℎ 𝑁𝑗−1 (ℎ/2) − 𝑁𝑗−1 (ℎ)

𝑁𝑗 (ℎ) = 𝑁𝑗−1 ( ) +
2 2𝑗−1 − 1
The results can be written in a table:

𝑂(ℎ) 𝑂(ℎ2 ) 𝑂(ℎ3 ) 𝑂(ℎ4 )

1:𝑁1 (ℎ) ≡ 𝑁 (ℎ)
2:𝑁1 ( ℎ2 ) ≡ 𝑁 ( ℎ2 ) 3:𝑁2 (ℎ)
4:𝑁1 ( ℎ4 ) ≡ 𝑁 ( ℎ4 ) 5:𝑁2 ( ℎ2 ) 6:𝑁3 (ℎ)
7:𝑁1 ( ℎ8 ) ≡ 𝑁 ( ℎ8 ) 8:𝑁2 ( ℎ4 ) 9:𝑁3 ( ℎ2 ) 10:𝑁4 (ℎ)
Richardson’s Extrapolation

If some error terms are zero, different and more efficient

formulas can be derived
Example: If

𝑀 − 𝑁 (ℎ) = 𝐾2 ℎ2 + 𝐾4 ℎ4 + ⋯

then an 𝑂(ℎ2𝑗 ) approximation is given for 𝑗 = 2, 3, … by

ℎ 𝑁𝑗−1 (ℎ/2) − 𝑁𝑗−1 (ℎ)

𝑁𝑗 (ℎ) = 𝑁𝑗−1 ( ) +
2 4𝑗−1 − 1
Numerical Quadrature

Integration of Lagrange Interpolating Polynomials

Select {𝑥0 , … , 𝑥𝑛 } in [𝑎, 𝑏] and integrate the Lagrange polynomial
𝑃𝑛 (𝑥) = ∑𝑛𝑖=0 𝑓(𝑥𝑖 )𝐿𝑖 (𝑥) and its truncation error term over [𝑎, 𝑏]
to obtain
𝑏 𝑛
∫ 𝑓(𝑥) 𝑑𝑥 = ∑ 𝑎𝑖 𝑓(𝑥𝑖 ) + 𝐸(𝑓)
𝑎 𝑖=0
with
𝑏
𝑎𝑖 = ∫ 𝐿𝑖 (𝑥) 𝑑𝑥
𝑎
and
𝑏 𝑛
1
𝐸(𝑓) = ∫ ∏(𝑥 − 𝑥𝑖 )𝑓 (𝑛+1) (𝜉(𝑥)) 𝑑𝑥
(𝑛 + 1)! 𝑎 𝑖=0
Trapezoidal and Simpson’s Rules
The Trapezoidal Rule
Linear Lagrange polynomial with 𝑥0 = 𝑎, 𝑥1 = 𝑏, ℎ = 𝑏 − 𝑎, gives
𝑏
ℎ ℎ3
∫ 𝑓(𝑥) 𝑑𝑥 = [𝑓(𝑥0 ) + 𝑓(𝑥1 )] − 𝑓 ″ (𝜉)
𝑎
2 12

Simpson’s Rule
Second Lagrange polynomial with 𝑥0 = 𝑎, 𝑥2 = 𝑏, 𝑥1 = 𝑎 + ℎ,
ℎ = (𝑏 − 𝑎)/2 gives
𝑥2
ℎ ℎ5
∫ 𝑑𝑥 = [𝑓(𝑥0 ) + 4𝑓(𝑥1 ) + 𝑓(𝑥2 )] − 𝑓 (4) (𝜉)
𝑥0
3 90

Definition
The degree of accuracy, or precision, of a quadrature formula is the
largest positive integer 𝑛 such that the formula is exact for 𝑥𝑘 , for
each 𝑘 = 0, 1, … , 𝑛.
The Newton-Cotes Formulas
The Closed Newton-Cotes Formulas
Use nodes 𝑥𝑖 = 𝑥0 + 𝑖ℎ, 𝑥0 = 𝑎, 𝑥𝑛 = 𝑏, ℎ = (𝑏 − 𝑎)/𝑛:
𝑏 𝑛
∫ 𝑓(𝑥) 𝑑𝑥 ≈ ∑ 𝑎𝑖 𝑓(𝑥𝑖 )
𝑎 𝑖=0
𝑥𝑛 𝑥𝑛 (𝑥 − 𝑥𝑗 )
𝑎𝑖 = ∫ 𝐿𝑖 (𝑥) 𝑑𝑥 = ∫ ∏ 𝑑𝑥
𝑥0 𝑥0 𝑗≠𝑖
(𝑥𝑖 − 𝑥𝑗 )

𝑛 = 1 gives the Trapezoidal rule, 𝑛 = 2 gives Simpson’s rule.

The Open Newton-Cotes Formulas

Use nodes 𝑥𝑖 = 𝑥0 + 𝑖ℎ, 𝑥0 = 𝑎 + ℎ, 𝑥𝑛 = 𝑏 − ℎ,
ℎ = (𝑏 − 𝑎)/(𝑛 + 2). Setting 𝑛 = 0 gives the Midpoint rule:
𝑥1
ℎ3 ″
∫ 𝑓(𝑥) 𝑑𝑥 = 2ℎ𝑓(𝑥0 ) + 𝑓 (𝜉)
𝑥−1
3
Composite Rules
Theorem
Let 𝑓 ∈ 𝐶 2 [𝑎, 𝑏], ℎ = (𝑏 − 𝑎)/𝑛, 𝑥𝑗 = 𝑎 + 𝑗ℎ, 𝜇 ∈ (𝑎, 𝑏). The
Composite Trapezoidal rule for 𝑛 subintervals is
𝑏 𝑛−1
ℎ 𝑏−𝑎 2 ″
∫ 𝑓(𝑥) 𝑑𝑥 = [𝑓(𝑎) + 2 ∑ 𝑓(𝑥𝑗 ) + 𝑓(𝑏)] − ℎ 𝑓 (𝜇)
𝑎
2 𝑗=1
12

Theorem
Let 𝑓 ∈ 𝐶 4 [𝑎, 𝑏], 𝑛 even, ℎ = (𝑏 − 𝑎)/𝑛, 𝑥𝑗 = 𝑎 + 𝑗ℎ, 𝜇 ∈ (𝑎, 𝑏).
The Composite Simpson’s rule for 𝑛 subintervals is

𝑏 (𝑛/2)−1 𝑛/2
ℎ
∫ 𝑓(𝑥) 𝑑𝑥 = [𝑓(𝑎) + 2 ∑ 𝑓(𝑥2𝑗 ) + 4 ∑ 𝑓(𝑥2𝑗−1 ) + 𝑓(𝑏)]
𝑎
3 𝑗=1 𝑗=1
𝑏 − 𝑎 4 (4)
− ℎ 𝑓 (𝜇)
180
Romberg Integration

Compute a sequence of 𝑛 integrals using the Composite

Trapezoidal rule, where 𝑚1 = 1, 𝑚2 = 2, 𝑚3 = 4, … and
𝑚𝑛 = 2𝑛−1 .
The step sizes are then ℎ𝑘 = (𝑏 − 𝑎)/𝑚𝑘 = (𝑏 − 𝑎)/2𝑘−1
The Trapezoidal rule becomes
𝑘−1
𝑏 2 −1
ℎ𝑘
∫ 𝑓(𝑥) 𝑑𝑥 = [𝑓(𝑎) + 𝑓(𝑏) + 2 ( ∑ 𝑓(𝑎 + 𝑖ℎ𝑘 ))]
𝑎
2 𝑖=1
(𝑏 − 𝑎) 2 ″
− ℎ𝑘 𝑓 (𝜇𝑘 )
12
Romberg Integration

Let 𝑅𝑘,1 denote the trapezoidal approximation, then

ℎ1 (𝑏 − 𝑎)
𝑅1,1 = [𝑓(𝑎) + 𝑓(𝑏)] = [𝑓(𝑎) + 𝑓(𝑏)]
2 2
1
𝑅2,1 = [𝑅1,1 + ℎ1 𝑓(𝑎 + ℎ2 )]
2
1
𝑅3,1 = {𝑅2,1 + ℎ2 [𝑓(𝑎 + ℎ3 ) + 𝑓(𝑎 + 3ℎ3 )]}
2
2𝑘−2
1
𝑅𝑘,1 = [𝑅𝑘−1,1 + ℎ𝑘−1 ∑ 𝑓(𝑎 + (2𝑖 − 1)ℎ𝑘 )]
2 𝑖=1

Apply Richardson extrapolation to these values:

𝑅𝑘,𝑗−1 − 𝑅𝑘−1,𝑗−1
𝑅𝑘,𝑗 = 𝑅𝑘,𝑗−1 +
4𝑗−1 − 1
Romberg Integration

MATLAB Implementation
function R = romberg(f, a, b, n)
% Compute integral of f(x) from a to b using Romberg integration.

h = b-a;
R = zeros(n,n);
R(1,1) = h/2 * (f(a) + f(b));
for i = 2:n
R(i,1) = 1/2 * (R(i-1,1) + h*sum(f(a + ((1:2^(i-2))-0.5)*h)));
for j = 2:i
R(i,j) = R(i,j-1) + (R(i,j-1)-R(i-1,j-1)) / (4^(j-1)-1);
end
h = h/2;
end
Error Estimation

The error term in Simpson’s rule requires knowledge of 𝑓 (4) :

𝑏
ℎ5 (4)
∫ 𝑓(𝑥) 𝑑𝑥 = 𝑆(𝑎, 𝑏) − 𝑓 (𝜇)
𝑎
90

Instead, apply it again with step size ℎ/2:

𝑏
𝑎+𝑏 𝑎+𝑏 1 ℎ5
∫ 𝑓(𝑥) 𝑑𝑥 = 𝑆 (𝑎, ) + 𝑆( , 𝑏) − ( ) 𝑓 (4) (𝜇)̃
𝑎
2 2 16 90

The assumption 𝑓 (4) (𝜇) ≈ 𝑓 (4) (𝜇)̃ gives the error estimate
𝑏
𝑎+𝑏 𝑎+𝑏
∣∫ 𝑓(𝑥) 𝑑𝑥 − 𝑆 (𝑎, ) − 𝑆( , 𝑏)∣
𝑎
2 2
1 𝑎+𝑏 𝑎+𝑏
≈ ∣𝑆(𝑎, 𝑏) − 𝑆 (𝑎, ) − 𝑆( , 𝑏)∣
15 2 2
Adaptive Quadrature

To compute ∫ 𝑓(𝑥) 𝑑𝑥 within a tolerance 𝜀 > 0, first apply

𝑏
𝑎
Simpson’s rule with ℎ = (𝑏 − 𝑎)/2 and with ℎ/2
If
𝑎+𝑏 𝑎+𝑏
∣𝑆(𝑎, 𝑏) − 𝑆 (𝑎, ) − 𝑆( , 𝑏)∣ < 15𝜀
2 2
then the integral is sufficiently accurate
If not, apply the technique to [𝑎, (𝑎 + 𝑏)/2] and [(𝑎 + 𝑏)/2, 𝑏],
and compute the integral within a tolerance of 𝜀/2
Repeat until each portion is within the required tolerance
Gaussian Quadrature

Basic idea: Calculate both nodes 𝑥1 , … , 𝑥𝑛 and coefficients

𝑐1 , … , 𝑐𝑛 such that
𝑏 𝑛
∫ 𝑓(𝑥) 𝑑𝑥 ≈ ∑ 𝑐𝑖 𝑓(𝑥𝑖 )
𝑎 𝑖=1

Since there are 2𝑛 parameters, we might expect a degree of

precision of 2𝑛 − 1
Example: 𝑛 = 2 gives the rule
1
√ √
− 3 3
∫ 𝑓(𝑥) 𝑑𝑥 ≈ 𝑓 ( ) + 𝑓( )
−1
3 3

with degree of precision 3

Legendre Polynomials

The Legendre polynomials 𝑃𝑛 (𝑥) have the properties

1. For each 𝑛, 𝑃𝑛 (𝑥) is a monic polynomial of degree 𝑛 (leading
coefficient 1)
2. ∫ 𝑃 (𝑥)𝑃𝑛 (𝑥) 𝑑𝑥 = 0 when 𝑃 (𝑥) is a polynomial of degree
1
−1
less than 𝑛
The roots of 𝑃𝑛 (𝑥) are distinct, in the interval (−1, 1), and
symmetric with respect to the origin.
Examples:

𝑃0 (𝑥) = 1, 𝑃1 (𝑥) = 𝑥
1 3
𝑃2 (𝑥) = 𝑥2 − 𝑃3 (𝑥) = 𝑥3 − 𝑥
3 5
6 3
𝑃4 (𝑥) = 𝑥4 − 𝑥2 +
7 35
Gaussian Quadrature

Theorem
Suppose 𝑥1 , … , 𝑥𝑛 are roots of 𝑃𝑛 (𝑥) and
1 𝑛 𝑥 − 𝑥𝑗
𝑐𝑖 = ∫ ∏ 𝑑𝑥
−1 𝑗≠𝑖
𝑥𝑖 − 𝑥𝑗

If 𝑃 (𝑥) is any polynomial of degree less than 2𝑛, then

1 𝑛
∫ 𝑃 (𝑥) 𝑑𝑥 = ∑ 𝑐𝑖 𝑃 (𝑥𝑖 )
−1 𝑖=1
Computing Gaussian Quadrature Coefficients

MATLAB Implementation
function [x, c] = gaussquad(n)
% Compute Gaussian quadrature points and coefficients.

P = zeros(n+1,n+1);
P([1,2],1) = 1;
for k = 1:n-1
P(k+2,1:k+2) = ((2*k+1)*[P(k+1,1:k+1) 0] - ...
k*[0 0 P(k,1:k)]) / (k+1);
end
x = sort(roots(P(n+1,1:n+1)));

A = zeros(n,n);
for i = 1:n
A(i,:) = polyval(P(i,1:i),x)';
end
c = A \ [2; zeros(n-1,1)];
Arbitrary Intervals

Transform integrals ∫ 𝑓(𝑥) 𝑑𝑥 into integrals over [−1, 1] by a

𝑏
𝑎
change of variables:

2𝑥 − 𝑎 − 𝑏 1
𝑡= ⇔ 𝑥 = [(𝑏 − 𝑎)𝑡 + 𝑎 + 𝑏]
𝑏−𝑎 2
Gaussian quadrature then gives
𝑏 1
(𝑏 − 𝑎)𝑡 + (𝑏 + 𝑎) (𝑏 − 𝑎)
∫ 𝑓(𝑥) 𝑑𝑥 = ∫ 𝑓 ( ) 𝑑𝑡
𝑎 −1
2 2
Double Integrals

Consider the double integral

∬ 𝑓(𝑥, 𝑦) 𝑑𝐴, 𝑅 = {(𝑥, 𝑦) | 𝑎 ≤ 𝑥 ≤ 𝑏, 𝑐 ≤ 𝑦 ≤ 𝑑}

𝑅

Partition [𝑎, 𝑏] and [𝑐, 𝑑] into even number of subintervals 𝑛, 𝑚

Step sizes ℎ = (𝑏 − 𝑎)/𝑛 and 𝑘 = (𝑑 − 𝑐)/𝑚
Write the integral as an iterated integral
𝑏 𝑑
∬ 𝑓(𝑥, 𝑦) 𝑑𝐴 = ∫ (∫ 𝑓(𝑥, 𝑦) 𝑑𝑦) 𝑑𝑥
𝑅 𝑎 𝑐

and use any quadrature rule in an iterated manner.

Composite Simpson’s Rule Double Integration
The Composite Simpson’s rule gives
𝑏 𝑑
ℎ𝑘 𝑛 𝑚
∫ (∫ 𝑓(𝑥, 𝑦) 𝑑𝑦) 𝑑𝑥 = ∑ ∑ 𝑤 𝑓(𝑥𝑖 , 𝑦𝑗 ) + 𝐸
𝑎 𝑐
9 𝑖=0 𝑗=0 𝑖,𝑗

where 𝑥𝑖 = 𝑎 + 𝑖ℎ, 𝑦𝑗 = 𝑐 + 𝑗𝑘, 𝑤𝑖,𝑗 are the products of the nested

Composite Simpson’s rule coefficients (see below), and the error is

(𝑑 − 𝑐)(𝑏 − 𝑎) 4 𝜕 4 𝑓 4
4𝜕 𝑓
𝐸=− [ℎ ( 𝜂,
̄ 𝜇)
̄ + 𝑘 (𝜂,̂ 𝜇)]
̂
180 𝜕𝑥4 𝜕𝑦4

𝑑 •1 •4 •2 •4 •1

•4 •16 •8 •16 •4

𝑐 •1 •4 •2 •4 •1
𝑎 𝑏
Non-Rectangular Regions

The same technique can be applied to double integrals of the form

𝑏 𝑑(𝑥)
∫ ∫ 𝑓(𝑥, 𝑦) 𝑑𝑦 𝑑𝑥
𝑎 𝑐(𝑥)

The step size for 𝑥 is still ℎ = (𝑏 − 𝑎)/𝑛, but for 𝑦 it varies with 𝑥:

𝑑(𝑥) − 𝑐(𝑥)
𝑘(𝑥) =
𝑚
Gaussian Double Integration

For Guassian integration, first transform the roots 𝑟𝑛,𝑗 from

[−1, 1] to [𝑎, 𝑏] and [𝑐, 𝑑], respectively
The integral is then
𝑏 𝑑
(𝑏 − 𝑎)(𝑑 − 𝑐) 𝑛 𝑛
∫ ∫ 𝑓(𝑥, 𝑦) 𝑑𝑦 𝑑𝑥 ≈ ∑ ∑ 𝑐𝑛,𝑖 𝑐𝑛,𝑗 𝑓(𝑥𝑖 , 𝑦𝑗 )
𝑎 𝑐
4 𝑖=1 𝑗=1

Similar techniques can be used for non-rectangular regions

Improper Integrals with a Singularity

The improper integral below, with a singularity at the left endpoint,

converges if and only if 0 < 𝑝 < 1 and then
𝑏 𝑏
1 (𝑥 − 𝑎)1−𝑝 (𝑏 − 𝑎)1−𝑝
∫ 𝑑𝑥 = ∣ =
𝑎
(𝑥 − 𝑎)𝑝 1−𝑝 1−𝑝
𝑎

More generally, if

𝑔(𝑥)
𝑓(𝑥) = , 0 < 𝑝 < 1, 𝑔 continuous on [𝑎, 𝑏],
(𝑥 − 𝑎)𝑝

construct the fourth Taylor polynomial 𝑃4 (𝑥) for 𝑔 about 𝑎:

𝑔″ (𝑎)
𝑃4 (𝑥) = 𝑔(𝑎) + 𝑔′ (𝑎)(𝑥 − 𝑎) + (𝑥 − 𝑎)2
2!
𝑔‴ (𝑎) 𝑔(4) (𝑎)
+ (𝑥 − 𝑎)3 + (𝑥 − 𝑎)4
3! 4!
Improper Integrals with a Singularity

and write
𝑏 𝑏 𝑏
𝑔(𝑥) − 𝑃4 (𝑥) 𝑃4 (𝑥)
∫ 𝑓(𝑥) 𝑑𝑥 = ∫ 𝑝
𝑑𝑥 + ∫ 𝑑𝑥
𝑎 𝑎
(𝑥 − 𝑎) 𝑎
(𝑥 − 𝑎)𝑝

The second integral can be computed exactly:

𝑏 4 𝑏 (𝑘)
𝑃4 (𝑥) 𝑔 (𝑎)
∫ 𝑝
𝑑𝑥 = ∑ ∫ (𝑥 − 𝑎)𝑘−𝑝 𝑑𝑥
𝑎
(𝑥 − 𝑎) 𝑘=0 𝑎
𝑘!
4
𝑔(𝑘) (𝑎)
=∑ (𝑏 − 𝑎)𝑘+1−𝑝
𝑘=0
𝑘!(𝑘 + 1 − 𝑝)
Improper Integrals with a Singularity

For the first integral, use the Composite Simpson’s rule to compute
the integral of 𝐺 on [𝑎, 𝑏], where
𝑔(𝑥)−𝑃4 (𝑥)
(𝑥−𝑎)𝑝 , if 𝑎 < 𝑥 ≤ 𝑏
𝐺(𝑥) = {
0, if 𝑥 = 𝑎

Note that 0 < 𝑝 < 1 and 𝑃4 (𝑎) agrees with 𝑔(𝑘) (𝑎) for each
(𝑘)

𝑘 = 0, 1, 2, 3, 4, so 𝐺 ∈ 𝐶 4 [𝑎, 𝑏] and Simpson’s rule can be applied.

Singularity at the Right Endpoint

For an improper integral with a singularity at the right endpoint

𝑏, make the substitution 𝑧 = −𝑥, 𝑑𝑧 = −𝑑𝑥 to obtain
𝑏 −𝑎
∫ 𝑓(𝑥) 𝑑𝑥 = ∫ 𝑓(−𝑧) 𝑑𝑧
𝑎 −𝑏

which has its singularity at the left endpoint

For an improper integral with a singularity at 𝑐, where
𝑎 < 𝑐 < 𝑏, split into two improper integrals
𝑏 𝑐 𝑏
∫ 𝑓(𝑥) 𝑑𝑥 = ∫ 𝑓(𝑥) 𝑑𝑥 + ∫ 𝑓(𝑥) 𝑑𝑥
𝑎 𝑎 𝑐
Infinite Limits of Integration

An integral of the form ∫ 𝑥1𝑝 𝑑𝑥, with 𝑝 > 1, can be converted to

∞
𝑎
an integral with left endpoint singularity at 0 by the substitution

𝑡 = 𝑥−1 , 𝑑𝑡 = −𝑥−2 𝑑𝑥, so 𝑑𝑥 = −𝑥2 𝑑𝑡 = −𝑡−2 𝑑𝑡

which gives
∞ 0 1/𝑎
1 𝑡𝑝 1
∫ 𝑝
𝑑𝑥 = ∫ − 2
𝑑𝑡 = ∫ 2−𝑝
𝑑𝑡
𝑎
𝑥 1/𝑎
𝑡 0
𝑡

More generally, this variable change converts ∫

∞
𝑓(𝑥) 𝑑𝑥 into
𝑎

∞ 1/𝑎
1
∫ 𝑓(𝑥) 𝑑𝑥 = ∫ 𝑡−2 𝑓 ( ) 𝑑𝑡
𝑎 0
𝑡