Root finding and numerical integration
Introduzido| English | Significado |
|---|---|
| iteration/ˌɪtəˈreɪʃn/ | iteração |
Newton's own cubic, 1690
- In 1690, the English mathematician Joseph Raphson published a method for finding roots that improves a guess again and again. Isaac Newton had used the same idea on the equation $x^3 - 2x - 5 = 0$, which has no neat solution. Starting from $x = 2$, two steps give the root to four decimal places. The method — now called Newton–Raphson — runs inside every calculator's equation solver.
- When an equation or integral cannot be solved exactly, numerical methods give approximations, and A-level questions ask how reliable they are.
Locating a root
- If $f$ is continuous and $f(a)$, $f(b)$ have opposite signs, there is a root between $a$ and $b$.
- For $f(x) = x^3 - 2x - 5$: $f(2) = -1$ and $f(3) = 16$, so a root lies in $(2, 3)$.
$f(x) = x^3 - 3x + 1$ has a root between 0 and 1.
Dificuldade 1/5 · Moedas 100 + pequeno bônus na primeira conclusão hoje
$f(0) = 1$ e $f(1) = -1$; $f$ is continuous.
Fixed-point iteration
- Rearrange $f(x) = 0$ as $x = g(x)$, then use the iteration 迭代 $x_{n+1} = g(x_n)$.
- $x = \sqrt[3]{2x + 5}$ with $x_0 = 2$ gives $2.0801, 2.0924, 2.0942, \dots$ — a staircase that converges to the root.
Root finding and numerical integration
A continuous function with opposite signs at two endpoints has a root between them
Compare the model with the worked case and explain one change.
Use $x_{n+1} = \sqrt{x_n + 3}$ with $x_0 = 2$. Find $x_1$ to 4 d.p.
Dificuldade 2/5 · Moedas 140 + pequeno bônus na primeira conclusão hoje
$\sqrt{5} = 2.23607$.
Worked example: Newton–Raphson
Use $x_{n+1} = x_n - \dfrac{f(x_n)}{f'(x_n)}$ with $x_0 = 2$ for $f(x) = x^3 - 2x - 5$.
- Derivative: $f'(x) = 3x^2 - 2$, so $f'(2) = 10$ and $x_1 = 2 - \dfrac{-1}{10} = 2.1$.
- Next: $f(2.1) = 0.061$ and $f'(2.1) = 11.23$, so $x_2 = 2.1 - \dfrac{0.061}{11.23} = 2.09457$.
- Check: $f(2.09455) < 0$ and $f(2.09465) > 0$, so the root is $2.0946$ to 4 d.p.
Use one step of Newton–Raphson on $f(x) = x^2 - 5$ from $x_0 = 2$.
Dificuldade 2/5 · Moedas 140 + pequeno bônus na primeira conclusão hoje
$2 - \dfrac{-1}{4} = 2.25$.
The trapezium rule
- $\displaystyle\int_a^b y\,dx \approx \tfrac{h}{2}\left[y_0 + y_n + 2(y_1 + \dots + y_{n-1})\right]$ with strip width $h$.
- $\displaystyle\int_1^3 \frac{1}{x}\,dx$ with 4 strips: $0.25[1 + 0.333 + 2(0.667 + 0.5 + 0.4)] = 1.117$, against the exact $\ln 3 = 1.099$.
Use the trapezium rule with 2 strips to estimate $\displaystyle\int_0^2 x^2\,dx$.
Dificuldade 2/5 · Moedas 140 + pequeno bônus na primeira conclusão hoje
$h = 1$: $\tfrac{1}{2}[0 + 4 + 2(1)] = 3$; the exact value is $\tfrac{8}{3}$.
How reliable?
- The trapezium rule overestimates when the curve is convex (bends upwards), as $\tfrac{1}{x}$ does; more strips reduce the error.
- Newton–Raphson fails if $f'(x_n) = 0$, and can jump to a different root if the start is near a turning point.
Why does the trapezium rule overestimate $\displaystyle\int_0^2 x^2\,dx$?
Dificuldade 1/5 · Moedas 100 + pequeno bônus na primeira conclusão hoje
Chords lie above a convex curve.
Newton–Raphson fails at $x_0$ when
Dificuldade 1/5 · Moedas 100 + pequeno bônus na primeira conclusão hoje
The formula divides by $f'(x_0)$; the tangent is horizontal.
A sign change does not always mean a root. $f(x) = \dfrac{1}{x}$ changes sign between $-1$ and 1, but it has no root — it has an asymptote. The function must be continuous on the interval.
A sign change of a continuous function locates a root. Iteration $x_{n+1} = g(x_n)$ and Newton–Raphson $x_{n+1} = x_n - \tfrac{f(x_n)}{f'(x_n)}$ refine it; the trapezium rule estimates integrals. Always judge reliability: convexity, strip number and starting points matter.
Show that $e^x = 3 - x$ has a root between 0 and 1, then use Newton–Raphson from $x_0 = 1$ to find it to 3 d.p.
Locate, then refine.