This preview shows pages 1–3. 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: L. Vandenberghe 12/2/10 EE103 Homework 8 solutions 1. Exercise 13.4 We define x = bracketleftBigg bracketrightBigg , A = 1 u 1 1 u 2 . . . . . . 1 u 42 , b = v 1 v 2 . . . v 42 . (a) We solve minimize bardbl Ax b bardbl 2 . The Matlab code is [u,v] = ch13ex4; m = length(u); A = [ones(m,1) u]; b = v; xa = A\b; The solution is = 4 . 1841, = 0 . 5755. (b) To make it easier to find the derivatives we express g as g ( x ) = h ( Ax b ) where h ( z ) = 42 summationdisplay i =1 radicalBig z 2 i + , and = 25. The first and second derivatives of h are given by h ( z ) z i = z i ( z 2 i + ) 1 / 2 , 2 h ( z ) z 2 i = ( z 2 i + ) 3 / 2 , 2 h ( z ) z i z j = 0 ( i negationslash = j ) . In other words, 2 h ( z ) is diagonal. To form the gradient and Hessian of g we can then use the formulas g ( x ) = A T h ( Ax b ) , 2 g ( x ) = A T h ( Ax b ) A. The Matlab code is as follows. rho = 25; x = xa; % use LS solution as starting point for k = 1:50 z = A*x - b; 1 grad = A * (z./sqrt(z.^2+rho)); if (norm(grad) < 1e-6); break; end; hess = A * diag( rho./(z.^2+rho).^(3/2) ) * A; x = x - hess\grad; end; xb = x; The solution is = 4 . 8164, = 0 . 8975....
View Full Document
This note was uploaded on 02/03/2011 for the course EE 103 taught by Professor Vandenberghe,lieven during the Spring '08 term at UCLA.
- Spring '08