4.

14 Gaussian Integration
Gaussian Integration Formulas

b
Newton-Cotes formulas for approximating ∫ f (x) dx  work best if f(x) is a smooth function, such as a polynomial. This is also true for Gaussian quadrature. However, Gaussian formulas are also good at estimating
a

integrals of the form

b
∫ w (x) f (x) dx
a

(6.15)

1 ∞
where w(x), called the weighting function, can contain singularities, as long as they are integrable. An example of such integral is ∫ 2
(1 + x ) ln x dx . Sometimes infinite limits, as in ∫ e
−x
sin x dx , can also be
0 0

accommodated.

Gaussian integration formulas have the same form as Newton-Cotes rules

(6.16)

where, as before, I represents the approximation to the integral in Eq. (6.15). The difference lies in the way that the weights Ai and nodal abscissas xi are determined. In Newton-Cotes integration, the nodes were evenly
spaced in (a, b) (i.e., their locations were predetermined). In Gaussian quadrature the nodes and weights are chosen so that Eq. (6.16) yields the exact integral if f(x) is a polynomial of degree 2n + 1 or less; that is,

(6.17)

One way of determining the weights and abscissas is to substitute P0(x) = 1, P1(x) = x, . . . , P2n+1(x) = x2n+1 in Eq. (6.17) and solve the resulting 2n + 2 equations
for the unknowns Ai and xi.

As an illustration , let w(x) = e−x, a = 0, b = ∞, and n = 1. The four equations determining x0, x1, A0, and A1 are as follows:

After evaluating the integrals, we get these results:

The solution is

So that the integration formula becomes

Because of the nonlinearity of the equations, this approach does not work well for large n. Practical methods of finding xi and Ai require some knowledge of orthogonal polynomials and their relationship to Gaussian
quadrature. There are, however, several “classical” Gaussian integration formulas for which the abscissas and weights have been computed with great precision and then tabulated. These formulas can be used without
knowing the theory behind them because all one needs for Gaussian integration are the values of xi and Ai.

Abscissas and Weights for Classical Gaussian Quadratures

Here we list some classical Gaussian integration formulas. The tables of nodal abscissas and weights, covering n = 1 to 5, have been rounded off to six decimal places. These tables should be adequate for hand
computation, but in programming, you may need more precision or a larger number of nodes. In that case, you should consult other references or use a subroutine to compute the abscissas and weights within the
integration program.

The truncation error in Gaussian quadrature

has the form E = K(n) f(2n+2)(c), where a < c < b (the value of c is unknown; only its bounds are given). The expression for K(n) depends on the particular quadrature being used. If the derivatives of f(x) can be evaluated,
the error formulas are useful in estimating the error bounds.

Gauss-Legendre Quadrature

(6.26)

Gauss-Legendre quadrature is the most often used Gaussian integration formula. As seen in Table 6.3, the nodes are arranged symmetrically about ξ = 0, and the weights associated with a symmetric pair of nodes are
equal. For example, for n = 1 we have ξ0 = −ξ1 and A0 = A1. The truncation error in Eq. (6.26) is

(6.27)
 Table 6.3. Nodes and weights for Gauss–Legendre quadrature.

b
To apply the Gauss-Legendre quadrature to the integral ∫ f (x) dx , we must first map the integration range (a, b) into the “standard” range (−1, 1˙). We can accomplish this by the transformation

(6.28)

Now dx = dξ(b − a)/2, and the quadrature becomes

 (6.29)

where the abscissas xi must be computed from Eq. (6.28). The truncation error here is

(6.30)

Gauss-Chebyshev Quadrature

(6.31)

Note that all the weights are equal: Ai = π/ (n + 1). The abscissas of the nodes, which are symmetric about x = 0, are given by

(6.32)

The truncation error is

(6.33)

 
Gauss-Laguerre Quadrature

(6.34)
 Table 6.4. Nodes and weights for Gauss–Laguerre quadrature (Multiply

numbers by 10k, where k is given in parentheses.)

(6.35)

Gauss-Hermite Quadrature

(6.36)
 Table 6.5. Nodes and weights for Gauss–Hermite quadrature (Multiply

numbers by 10k, where k is given in parentheses.)

The nodes are placed symmetrically about x = 0.

(6.37)

 
Gauss Quadrature with Logarithmic Singularity

 Table 6.6. Nodes and weights for quadrature with a logarithmic singularity (Multiply

numbers by 10k, where k is given in parentheses.)

(6.39)

where k(1) = 0.00 285, k(2) = 0.000 17, and k(3) = 0.000 01.

Illustrative Example:
3
1
Evaluate ∫ 2
(1 − x )
2
dx  as accurately as possible with Gaussian integration.

−1

Solution. Because the integrand is smooth and free of singularities, we could use the Gauss-Legendre quadrature. However, the exact integral can be obtained with the Gauss-Chebyshev formula. We write

The numerator f(x) = (1 − x2)2 is a polynomial of degree four so that Gauss-Chebyshev quadrature is exact with three nodes.

The abscissas of the nodes are obtained from Eq. (6.32). Substituting n = 2, we get

Therefore,
and Eq. (6.31) yields

Illustrative Example:

0.5
Use Gaussian integration to evaluate ∫ cos πx ln x dx .

Solution. We split the integral into two parts:

The first integral on the right-hand-side, which contains a logarithmic singularity at x = 0, can be computed with the special Gaussian quadrature in Eq. (6.38). Choosing n = 3, we have

The sum is evaluated in the following table:

Thus

The second integral is free of singularities so that it can be evaluated with the Gauss-Legendre quadrature. Choosing n = 3, we have

where the nodal abscissas are—see Eq. (6.28)—

Looking up ξi and Ai in Table 6.3 leads to the following computations:

from which

Therefore

which is correct to six decimal places.

 Illustrative Example:
Evaluate as accurately as possible

Solution. In its present form, the integral is not suited to any of the Gaussian quadratures listed in this section. But using the transformation

x = t2

dx = 2t dt

the integral becomes

which can be evaluated exactly with the Gauss-Hermite formula using only two nodes (n = 1). Thus

F = A0(t20 + 3) + A1(t21 + 3)

= 0.886 227 [(0.707 107)2 + 3] + 0.886 227 [(−0.707 107)2 + 3]

= 6. 203 59

Illustrative Example:

3
Evaluate numerically ∫ , where f(x) is represented by the following unevenly spaced data:
f (x) dx
1.5

x 1.2 1.7 2.0 2.4 2.9 3.3

Knowing that the data points lie on the curve f(x) = − cos x, evaluate the accuracy of the solution.

3 3
Solution. We approximate f(x) by the polynomial P5(x) that intersects all the data points and then evaluate  with ∫ f (x) dx ≈ ∫ P5 (x) dx with the Gauss-Legendre formula. Because the polynomial is of degree
1.5 1.5

five, only three nodes (n = 2) are required in the quadrature.

From Eq. (6.28) and Table 6.6, we obtain the abscissas of the nodes 
We now compute the values of the interpolant P5(x) at the nodes. The results are

P5(x0) = 0.098 08

P5(x1) = 0.628 16

P5(x2) = 0.952 16

From Gauss-Legendre quadrature

we g e t

I = 0.75 [0.555 556(0.098 08) + 0.888 889(0.628 16) + 0.555 556(0.952 16)]

= 0.856 37

3
Comparison with −∫ cos (x) dx = 0.85638   shows that the discrepancy is within the

1.5

roundoff error. 

Reference:

Kiusalaas, J. (2013). Numerical Methods in Engineering with Python 3. USA: Cambridge University Press.