- Inicio
- Interpolation
- Newton's method
- Ejercicios
Newton's method
Considering the table of the function $f(x)=e^x$:
| $x$ | $0.0$ | $0.2$ | $0.4$ |
| $f(x)$ | $1.0000$ | $1.2214$ | $1.4918$ |
- Find the value of $\sqrt[3]{e}$ by interpolation.
- Give a bound for the error.
- Find the value of $\sqrt[3]{e}$ by interpolation if we add a new point: $(0.6, 1.8221)$.
- We are going to interpolate the function from the given information and later we will find an approximation for $\sqrt[3]{e}$. We write the table:
| $0.0$ | $1.0000$ | ||
| $\dfrac{1.2214-1.0000}{0.2-0.0}=\dfrac{0.2214}{0.2}=1.107$ | |||
| $0.2$ | $1.2214$ | $\dfrac{1.352-1.107}{0.4-0.0}=\dfrac{0.245}{0.4}=0.6125$ | |
| $\dfrac{1.4918-1.2214}{0.4-0.2}=\dfrac{0.2704}{0.2}=1.352$ | |||
| $0.4$ | $1.4918$ |
Thus the interpolating polynomial is
$$ \begin{array}{rl} P_2(x)=& 1.000+1.107\cdot(x-0)+0.6125\cdot(x-0)\cdot(x-0.2) \\ =& 1.000+1.107x-0.1225x+0.6125x^2\\ =& 1.000+0.9845x+0.6125x^2 \end{array}$$
Then, $\sqrt[3]{e}=e^{\frac{1}{3}}\approx P_2\Big( \dfrac{1}{3}\Big) = 1+0.9845\cdot\dfrac{1}{3}+0.6125\cdot\dfrac{1}{9}=1.3962$
- Since we already know the calculation that we have done is an approximation, we are going to try to give a bound for the error. To do so, we will have to know the third derivative of the function and a bound for its module knowing that $x\in(0,0.4)$:
$$f^{(3)}(x)=e^x \Rightarrow |f^{(3)}(x)|\leqslant e^{0.4} $$
since the function is increasing. Therefore:
$$\begin{array}{rl} |\text{error}|=& \Big| f\Big( \dfrac{1}{3} \Big) - P_2 \Big( \dfrac{1}{3} \Big) \Big| = \Big| \dfrac{f^{(3)}(\xi(x))}{3!}\cdot\Big( \dfrac{1}{3}-0\Big)\cdot \Big( \dfrac{1}{3}-0.2\Big)\cdot\Big( \dfrac{1}{3}-0.4\Big) \Big| \\ \leqslant & \Big| \dfrac{e^{0.4}}{3!}\cdot \dfrac{1}{3} \cdot \cdot \Big( \dfrac{1}{3}-0.2\Big)\cdot\Big( \dfrac{1}{3}-0.4\Big) \Big| = 0.36\cdot 10^{-4} \end{array} $$
- Bearing in mind that we have calculated the interpolating polynomial using Newton's method, when adding another point we will use the calculations we already did. We add a new line with the new point in the previous table:
| $0.0$ | $1.0000$ | |||
| $\dfrac{1.2214-1.0000}{0.2-0.0}=\dfrac{0.2214}{0.2}=1.107$ | ||||
| $0.2$ | $1.2214$ | $\dfrac{1.352-1.107}{0.4-0.0}=\dfrac{0.245}{0.4}=0.6125$ | ||
| $\dfrac{1.4918-1.2214}{0.4-0.2}=\dfrac{0.2704}{0.2}=1.352$ | $\dfrac{0.74875-0.6125}{0.6-0.0}=\dfrac{0.13625}{0.6}=0.22708$ | |||
| $0.4$ | $1.4918$ | $\dfrac{1.6515-1.352}{0.6-0.2}=\dfrac{0.2995}{0.4}=0.74875$ | ||
| $\dfrac{1.8221-1.4918}{0.6-0.4}=\dfrac{0.3303}{0.2}=1.6515$ | ||||
| $0.6$ | $1.8221$ |
With the new calculations. Now, the interpolating polynomial (now of degree $3$) is:
$$ \begin{array}{rl} P_3(x)=& 1+1.107\cdot(x-0)+0.6125\cdot(x-0)\cdot(x-0.2)\\ & +0.22708\cdot(x-0)\cdot(x-0.2)\cdot(x-0.4)\\ =& 1+1.107x-0.1225x+0.6125x^2+0.22708x^3\\ &-0.0908x^2-0.0454x^2+0.0182x\\ =& 1+1.0027x+0.4853x^2 +0.2271x^3 \end{array}$$
So, the value of $\sqrt[3]{e}$ will be: $$\sqrt[3]{e}=e^{\frac{1}{3}}\approx P_3\Big( \dfrac{1}{3}\Big) = 1+1.0027\cdot\dfrac{1}{3}+0.4853\cdot\dfrac{1}{9}+0.2271\dfrac{1}{27}=1.3967$$
- $1.3962$
- $7.36\cdot 10^{-4}$
- $1.3967$