18.01 Section, September 16, 2015 Section leader: Eva Belmont (ebelmont@mit.edu, E18-401B) . Fixed point iteration and Newton’s method 1. Activity about fixed points: (a) Start with the line y = 12 x + 1. Using high school algebra, find the fixed point. (b) Now try approximating the fixed point using the fixed point iteration method with an initial guess of x0 = 1; calculate x1 , x2 , and x3 . (c) Make a plot of “error vs. time”: i.e. plot points (n, errorn ) where errorn = (true fixed point found in (a)) − xn . (d) Do the above three parts again, for the line y = 2x − 1 and initial guess x0 = 0. Does the initial guess matter for the general shape of the plot? 1 (e) In each case, apply the criterion from class that tells you when a fixed point is attracting or not. (f) Start with a linear equation of the form f (x) = α + m(x − α). (i) What’s the fixed point? (ii) Write error1 in terms of error0 . (iii) Write error2 in terms of error1 , and then write error2 in terms of error0 . (iv) Write error3 in terms of error0 . (v) Write errorn in terms of error0 . (vi) How does this explain all the behavior you observed above? 2. f (x) = 2x − 23 has a fixed point at x = −1. Start with an initial guess of x0 = 2. Does the sequence x0 , x1 , . . . approach x = −1? (You don’t have to calculate exactly; just convince yourself one way or the other.) It turns out that f 0 (−1) ≈ 0.347. Why is this not a violation of things said in class about attracting points? 2 3. Use Newton’s method to approximate √ 3. Start with x0 = 1 and calculate x1 , x2 , and x3 . 4. Bonus question: Show that Newton’s method is a special case of fixed point iteration. Use this to check that Newton’s method converges near the roots of a quadratic equation (x − a)(x − b). Review • Derivative of a polynomial: d n dx ax = anxn−1 • Linear approximation of f near a: f (x) ≈ f (a) + f 0 (a)(x − a) • Newton’s method: xn+1 = −f (xn ) + xn f 0 (xn ) • Fixed point iteration method: initial guess x0 is given; then x1 = f (x0 ), x2 = f (x1 ), x3 = f (x2 ), etc. • Attracting point: fixed point a such that |f 0 (a)| < 1 • Repelling point: fixed point a such that |f 0 (a)| > 1 3