Consider the dimensionless boundary-value problem on with and

PHYSICS · Multi-part · Fermi University Physics · Mathematical Methods for Physics · Numerical and computational mathematics · Numerical and computational mathematics

Error, Chance and the Machine, Ch Ordinary differential equations: order, stability and stiffness

Problem

Consider the dimensionless boundary-value problem y′′=32y2y'' = \tfrac32y^2 on 0≤x≤10 \le x \le 1 with y(0)=4y(0) = 4 and y(1)=1y(1) = 1. Solve it by shooting on the trial slope s=y′(0)s = y'(0), with miss F(s)=y(1;s)−1F(s) = y(1; s) - 1. Integrate with an adaptive Runge-Kutta code (the Dormand-Prince pair, for example) with relative tolerance 10−1010^{-10} and absolute tolerance 10−1210^{-12}. ((a)) Verify that y=4/(1+x)2y = 4/(1 + x)^2 satisfies the boundary-value problem, and give the value of ss it corresponds to. ((b)) Compute F(s)F(s) for s=−40,−30,−20,−10s = -40, -30, -20, -10 and −6-6, and sketch FF on −40≤s≤0-40 \le s \le 0. How many roots does the sketch suggest? ((c)) Run the secant iteration from (s0,s1)=(−10,−6)(s_0, s_1) = (-10, -6). Give the first iterate s2s_2 and the root to which the iteration converges. ((d)) Run it again from (−40,−30)(-40, -30). Give the root reached and the minimum value of yy on 0≤x≤10 \le x \le 1 for the solution it defines. Is this a solution of the same boundary-value problem as in (a)? ((e)) Start the secant iteration from (−20,−15)(-20, -15) instead. Find the first iterate and explain why this pair is a poor start.

Browse the problem library