This preview shows pages 1–2. Sign up to view the full content.
This preview has intentionally blurred sections. Sign up to view the full version.View Full Document
Unformatted text preview: 18.02 Fall 2008 Problem Set 4, Part B Solutions 1. a) LS (4) translates into braceleftBigg ( x x ) a + ( x u ) b = ( x y ) ( x u ) a + ( u u ) b = ( y u ) or equivalently A z = r , with A = bracketleftbigg x x x u x u u u bracketrightbigg , z = bracketleftbigg a b bracketrightbigg , r = bracketleftbigg x y y u bracketrightbigg In Matlab notation: A=[x*x x*u; x*u u*u] , and r=[x*y; y*u] (or even shorter: A=[x;u]*[x u] , r=[x;u]*y ) c) Matlab input and output (condensed for brevity) >> x = [1:1:7]; y = [0.25 1 2.1 4.7 9.8 18.7 48.1]; u = [1 1 1 1 1 1 1]; >> A = [x*x x*u; x*u u*u]; r = [x*y; y*u]; z = inv(A)*r z = 6.6661-14.5714 (the best fit is: y = 6.666 x - 14.571) >> v = z*[x;u] v = -7.9054 -1.2393 5.4268 12.0929 18.7589 25.4250 32.0911 (predicted) >> y - v 8.1554 2.2393 -3.3268 -7.3929 -8.9589 -6.7250 16.0089 (differences) a = 6 . 666, b = 14 . 571, and the largest error is about 16 (for the last data point). d) Matlab input and output: (note A remains the same) >> lny = log(y); r1 = [x*lny; lny*u]; z1 = inv(A)*r1 z1 = 0.8277-1.8841 (the best fit is: ln(y) = 0.8277 x - 1.8841)(the best fit is: ln(y) = 0....
View Full Document
This note was uploaded on 04/28/2009 for the course MATH 18.02 taught by Professor Auroux during the Fall '08 term at MIT.
- Fall '08