MathDB
iterating function f(x)=sqrt4(1-x)

Source: France 1986 P5

May 19, 2021
functionalgebra

Problem Statement

The functions f,g:[0,1]Rf,g:[0,1]\to\mathbb R are given with the formulas f(x)=1x4,g(x)=f(f(x)),f(x)=\sqrt[4]{1-x},\enspace g(x)=f(f(x)), and cc denotes any solution of x=f(x)x=f(x).
(a) i. Analyze the function f(x)f(x) and draw its graph. Prove that the equation f(x)=xf(x)=x has the unique root cc satisfying c[0.72,0.73]c\in[0.72,0.73]. ii. Analyze the function f(x)f'(x). Let M1M_1 and M2M_2 be the points of the graph of f(x)f(x) with different xx coordinates. What is the position of the arc M1M2M_1M_2 of the graph with respect to the segment M1M2M_1M_2? iii. Analyze the function g(x)g(x) and draw its graph. What is the position of that graph with respect to the line y=xy=x? Find the tangents to the graph at points with xx coordinates 00 and 11. iv. Prove that every sequence {an}\{a_n\} with the conditions a1(0,1)a_1\in(0,1) and an+1=f(an)a_{n+1}=f(a_n) for nNn\in\mathbb N converges. [hide=Official Hint]Consider the sequences {a2n1},{a2n} (nN)\{a_{2n-1}\},\{a_{2n}\}~(n\in\mathbb N) and the function g(x)g(x) associated with the graph. (b) On the graph of the function f(x)f(x) consider the points MM and MM' with xx coordinates xx and f(x)f(x), where xcx\ne c. i. Prove that the line MMMM' intersects with the line y=xy=x at the point with xx coordinate h(x)=x(f(x)x)2g(x)+x2f(x).h(x)=x-\frac{(f(x)-x)^2}{g(x)+x-2f(x)}. ii. Prove that if x(0,c)x\in(0,c) then h(x)(x,c)h(x)\in(x,c). iii. Analyze whether the sequence {an}\{a_n\} satisfying a1(0,c),an+1=h(an)a_1\in(0,c),a_{n+1}=h(a_n) for nNn\in\mathbb N converges. Prove that the sequence {an+1canc}\{\tfrac{a_{n+1}-c}{a_n-c}\} converges and find its limit. (c) Assume that the calculator approximates every number b[2,2]b\in[-2,2] by number b\overline b having pp decimal digits after the decimal point. We are performing the following sequence of operations on that calculator:
1) Set a=0.72a=0.72; 2) Calculate δ(a)=f(a)a\delta(a)=\overline{f(a)}-a; 3) If δ(a)>0.510p|\delta(a)|>0.5\cdot10^{-p}, then calculate h(a)\overline{h(a)} and go to the operation 2)2) using h(a)\overline{h(a)} instead of aa; 4) If δ(a)0.510p|\delta(a)|\le0.5\cdot10^{-p}, finish the calculation.
Let c\overleftrightarrow c be the last of calculated values for h(a)\overline{h(a)}. Assuming that for each x[0.72,0.73]x\in[0.72,0.73] we have f(x)f(x)<ϵ\left|\overline{f(x)}-f(x)\right|<\epsilon, determine δ(c)\delta(\overleftrightarrow c), the accuracy (depending on ϵ\epsilon) of the approximation of cc with c\overleftrightarrow c. (d) Assume that the sequence {an}\{a_n\} satisfies a1=0.72a_1=0.72 and an+1=f(an)a_{n+1}=f(a_n) for nNn\in\mathbb N. Find the smallest n0Nn_0\in\mathbb N, such that for every nn0n\ge n_0 we have anc<106|a_n-c|<10^{-6}.