Interpolation (Numerical Analysis)

431 Views Asked by At

Suppose you have a table of the logarithm function $\ln x$ for positive integer values of $x$, and you compute $\ln 11.1$ by quadratic interpolation at $x_0 = 10, x_1 = 11, x_2 = 12$.

Estimate the relative error incurred.

Is this problem is about calculation of Error bound?

1

There are 1 best solutions below

0
On

The goal is to interpolate three data points using quadratic interpolation for the natural log function using the three data points $x = 10, 11, 12$, yielding the data set:

$$(x, y) = (x, \ln x) = (10 , \ln 10), (11, \ln 11), (12 \ln 12)$$

We will use the following 10-digits of precision for these calculations:

$$(x, y) = (10 , 2.302585093), (11, 2.397895273), (12 , 2.484906650)$$

We want to interpolate these three data points with the quadratic:

$$y(x) = a_2 x^2 + a_1 x + a_0$$

In order to do this, we want to solve the linear system for $a_0, a_1, a_2$: $$a_2 x_1^2 + a_1x_1 + a_0 = y_1 \\ a_2 x_2^2 + a_1x_2 + a_0 = y_2 \\ a_2 x_3^2 + a_1x_3 + a_0 = y_3$$

In matrix form, we can show this as:

$$ \begin{bmatrix} 1 & x_1 & x_1^2 \\ 1 & x_2 & x_2^2 \\ 1 & x_3 & x_3^2 \end{bmatrix} \begin{bmatrix} a_0 \\ a_1 \\ a_2 \end{bmatrix} = \begin{bmatrix} y_1 \\ y_2 \\ y_3 \end{bmatrix}$$

We can now apply Gaussian Elimination (or whatever floats your boat) to solve for $a_0, a_1, a_2$, so we have:

$$ \begin{bmatrix} 1 & 10 & 100 \\ 1 & 11 & 121 \\ 1 & 12 & 144 \end{bmatrix} \begin{bmatrix} a_0 \\ a_1 \\ a_2 \end{bmatrix} = \begin{bmatrix} 2.302585093 \\ 2.397895273 \\ 2.484906650 \end{bmatrix}$$

This yields:

$$a_0 = 0.893049 , a_1 = 0.182448 , a_2 = -0.0041494 $$

This gives us the quadratic interpolating polynomial of:

$$y(x) = -0.0041494 x^2 + 0.182448 x + 0.893049 $$

Using the data to compare, we arrive at:

$$\begin{array}{c|c|c|c} \text {x} & y(x) & \ln x & |y(x) - \ln x)| \\ \hline 10 & 2.30259 & 2.302585093 & 3.907005953873721 \times 10^{-6} \\ \hline 11 & 2.3979 & 2.397895273 & 4.327201629017452 \times 10^{-6} \\ \hline 11.1 & 2.40697 & 2.40695 & 0.0000291177 \\ \hline 12 & 2.48491 & 2.484906650 & 4.750211999748899 \times 10^{-6} \end{array}$$

To calculate the relative error at $x = 11.1$, we have:

$$ \mbox{Relative Error} = \left|\dfrac{\mbox{calculated - actual}}{\mbox{actual}}\right| = \left|\dfrac{2.40697 - 2.40695}{2.40695}\right| = 8.309271069065387 \times 10^{-6}$$