Pure 9.2 - Iteration, cobweb and staircase diagrams

Pure 9.2 - Iteration, cobweb and staircase diagrams

Build successive approximations to a root, and see the same calculation as a path between a curve and a diagonal. You will rearrange equations, calculate accurately, draw iteration paths and judge whether they approach a fixed point.

Fixed points and iteration formulas

Can pressing the same calculator instruction repeatedly solve x3+x2=9x^3+x^2=9? Start with 11, apply x↦9/(x+1)x\mapsto\sqrt{9/(x+1)}, and the next value is 2.121320…2.121320\ldots. Feeding that result back gives 1.698056…1.698056\ldots. The input and output are beginning to chase a common value.

You already know

A recurrence defines each new term from previous terms. A root of F(x)=0F(x)=0 is an input making the original expression zero; rearranging an equation must preserve its domain and the required branch of a root.

We use gg for the repeated instruction, keeping it distinct from the original equation F(x)=0F(x)=0.

Fixed point

A fixed point α\alpha of gg satisfies g(α)=αg(\alpha)=\alpha. For example, 22 is fixed by g(x)=x/2+1g(x)=x/2+1, whereas 00 is not.

If the sequence settles, its next value and current value tend to the same number.

Fixed-point iteration

xn+1=g(xn)x_{n+1}=g(x_n)

If xn→αx_n\to\alpha and gg is continuous at α\alpha, then α=g(α)\alpha=g(\alpha). The rearrangement must also make this a solution of the original equation.

Not in the formula booklet: learn it.

A valid rearrangement gives a possible target, but does not guarantee that repeated substitution will reach it.

Worked example 1: selecting the positive root

Question. Rearrange x3+x2−9=0x^3+x^2-9=0 into an iteration for its positive root.

Plan. Collect the terms containing x2x^2, then take the positive square root.

x2(x+1)=9factorisex2=9x+1x≠−1x=9x+1x>0\begin{aligned} x^2(x+1)&=9 && \text{factorise} \\ x^2&=\frac9{x+1} && x\ne-1 \\ x&=\sqrt{\frac9{x+1}} && x>0 \end{aligned}

Why this step? The positive square root matches the requested positive solution; it is not a rearrangement covering negative solutions.

Answer. Use xn+1=9/(xn+1)x_{n+1}=\sqrt{9/(x_n+1)}, with a suitable positive starting value.

Check. Squaring a positive fixed point and multiplying by α+1\alpha+1 returns α3+α2=9\alpha^3+\alpha^2=9.

Change only the constant from 99 to 1010. The domain and branch decision stay the same.

Worked example 2: varying the constant

Question. Form an iteration for the positive root of x3+x2−10=0x^3+x^2-10=0.

Plan. Isolate x2x^2 in the same way.

x2(x+1)=10factorisex=10x+1x>0\begin{aligned} x^2(x+1)&=10 && \text{factorise} \\ x&=\sqrt{\frac{10}{x+1}} && x>0 \end{aligned}

Why this step? A positive input makes the denominator positive and produces a positive next input.

Answer. xn+1=10/(xn+1)x_{n+1}=\sqrt{10/(x_n+1)}.

Check. A fixed point satisfies α2(α+1)=10\alpha^2(\alpha+1)=10, precisely the original equation.

Taking a cube root has no positive/negative branch choice for real inputs. For example, x3−2x−5=0x^3-2x-5=0 gives xn+1=2xn+53x_{n+1}=\sqrt[3]{2x_n+5}. The same rearrange-and-check principle applies.

Watch out

Wrong: a correct fixed-point equation proves convergence. Right: it identifies what a convergent sequence could approach. Inspect the actual iterates or graph before claiming success.

For your turn, change only the constant from 1010 to 1111, preserving the positive-root decision.

Calculating the sequence

Use each unrounded answer as the next input. A displayed table can show rounded values, but its next row must come from the stored value. Also follow the given indexing: a start labelled x0x_0 differs from a start labelled x1x_1.

Worked example 3: repeated substitution

Question. Use xn+1=9/(xn+1)x_{n+1}=\sqrt{9/(x_n+1)}, with x0=1x_0=1, to calculate x1x_1, x2x_2 and an approximate positive root to three decimal places.

Plan. Store each output and continue until the displayed values settle.

x1=9/2=2.121320344…first substitutionx2=9/(2.121320344…+1)=1.698056292…use stored x1\begin{aligned} x_1&=\sqrt{9/2}=2.121320344\ldots && \text{first substitution} \\ x_2&=\sqrt{9/(2.121320344\ldots+1)} \\ &=1.698056292\ldots && \text{use stored }x_1 \end{aligned}

Continuing gives x3=1.826399382…x_3=1.826399382\ldots, x4=1.784450437…x_4=1.784450437\ldots, x5=1.797841946…x_5=1.797841946\ldots and x6=1.793534216…x_6=1.793534216\ldots. Later, x9=1.794615135…x_9=1.794615135\ldots and x10=1.794569372…x_{10}=1.794569372\ldots.

Why this step? Alternating values can still converge; the size of the movement matters.

Answer. The calculation suggests a root of 1.7951.795 to three decimal places.

Check. Substituting 1.794581.79458 into x3+x2−9x^3+x^2-9 gives about −0.00000655-0.00000655, close to zero. This is a sense-check, not an accuracy proof.

Keep the same instruction and change only the constant to 1010. Retaining all stored digits matters just as much.

Worked example 4: a second iteration table

Question. Use xn+1=10/(xn+1)x_{n+1}=\sqrt{10/(x_n+1)}, with x0=1x_0=1, to find x1x_1, x2x_2 to four decimal places and the apparent limit to three decimal places.

Plan. Iterate with full precision, rounding only the requested outputs.

x1=5=2.236067977…first outputx2=1.757887921…repeat with stored input\begin{aligned} x_1&=\sqrt5=2.236067977\ldots && \text{first output} \\ x_2&=1.757887921\ldots && \text{repeat with stored input} \end{aligned}

Continuing gives x9=1.867503507…x_9=1.867503507\ldots, x10=1.867445866…x_{10}=1.867445866\ldots, and eventually 1.867460…1.867460\ldots.

Why this step? Two early values need not be close to the limit, even when the recurrence ultimately converges.

Answer. x1=2.2361x_1=2.2361, x2=1.7579x_2=1.7579; the apparent limit is 1.8671.867 to three decimal places.

Check. x9x_9 and x10x_{10} round differently to three decimal places, so stopping at x9x_9 would be premature.

The word “apparent” keeps the evidence honest: a finite list is evidence of convergence rather than a general proof.

Exam tip

If asked for x2x_2 and x3x_3, give those iterates. A calculator's final root does not replace the requested intermediate values.

The cube-root rule above, with x0=1.8x_0=1.8, gives 2.048799615…2.048799615\ldots, 2.087575864…2.087575864\ldots, eventually 2.094551…2.094551\ldots. For your turn use the square-root rule, changing only its constant to 1111.

Drawing the iteration on a graph

Draw both y=g(x)y=g(x) and y=xy=x. Their intersection is (α,α)(\alpha,\alpha): input and output agree there. Starting on the xx-axis, move vertically to the curve to evaluate the function, then horizontally to the diagonal to copy that output into the next input.

[DIAGRAM: asset_slug: p1_9_2_solve_equations_approximately_using_iterative_methods_and_draw_associated_cobweb_and_staircase_diagrams__diagram_01; asset_name: Evaluating and copying an iterate; caption: Vertical moves evaluate g; horizontal moves copy the output onto y equals x; description: The curve g(x)=sqrt(9/(x+1)) and the diagonal y=x have fixed point approximately 1.79458. Four evaluate-then-copy pairs start at x0=1 on equal coordinate scales.]
Vertical moves evaluate g

Continue this lesson.

You’re previewing NovaLearn without an account. Create a free account to read the rest.

Create free account