January 12th, 2018, 02:20 PM  #1 
Senior Member Joined: Jan 2017 From: Toronto Posts: 163 Thanks: 2  Least Squares Problem
This problem projects $\displaystyle b = (b_1,b_2....,b_m)$ onto the line through $\displaystyle a = (1, 1, 1, ....1)$. We solve m equations ax = b in 1 unknown (by least squares). (a) Solve $\displaystyle a_T~a~\hat{x} = a_T~b $ to show that \hat{x} is the mean (the average) of the b’s. (b) Find $\displaystyle e = b  a \hat{x}$ and the variance $\displaystyle e^2$ and the standard deviation $\displaystyle e$. (c) The horizontal line $\displaystyle \hat{b} = 3 $ is closest to b = (1, 2, 6). Check that p = (3, 3 3) is perpendicular to e and find the 3 by 3 projection matrix P. Ans(a): Becauase a = (1,1,1,....1), therefore $\displaystyle a_T a = 1 + 1 + 1 +....+ 1 = 1~*~m = m$ And $\displaystyle a_T b = b_1 + b_2 + .... + b_m $ So $\displaystyle \hat{x} = \frac{b_1 + b_2 + b_3 + .... + b_m}{m} = b_{avg} $ Ans(b): Need help.. Ans(c): Need help.. Last edited by zollen; January 12th, 2018 at 02:42 PM. 
January 12th, 2018, 02:39 PM  #2 
Global Moderator Joined: May 2007 Posts: 6,416 Thanks: 557 
b) Let $\displaystyle \epsilon_k=b_x\hat{x} $. $\displaystyle \epsilon = (\epsilon_1,...\epsilon_m) $, $\displaystyle \epsilon ^2=\frac{\epsilon_1^2+...\epsilon_m^2}{m} $. Ill let you do (c). 
January 13th, 2018, 07:45 AM  #3 
Senior Member Joined: Jan 2017 From: Toronto Posts: 163 Thanks: 2 
Ans(b): $\displaystyle Variance = \frac{\sum_{k=1}^{m} (b_k  b_{avg})^2}{m} $ Ans(c): e = (1, 2, 6)  (3, 3, 3) = (2, 1, 3) (2, 1, 3) dot_product (3, 3, 3) = 0 Since p = Pb, (3, 3, 3) = P (1, 2, 6) Project Matrix (P) = \begin{bmatrix} 3&0&0\\ 0&3/2&0\\ 0&0&1/2\\ \end{bmatrix} 
January 14th, 2018, 07:23 AM  #4 
Senior Member Joined: Jan 2017 From: Toronto Posts: 163 Thanks: 2 
Ans(b): Given $\displaystyle \hat{x} = b_{avg} $ Therefore $\displaystyle e = b  a \hat{x} = \begin{bmatrix} b_1\\ b_2\\ ..\\ b_m \end{bmatrix} ~~ \begin{bmatrix} 1\\ 1\\ ..\\ 1 \end{bmatrix} ~b_{avg} ~=~ \begin{bmatrix} e_1\\ e_2\\ ..\\ e_m \end{bmatrix} $ $\displaystyle e^2~=~\frac{{e_1}^2 + {e_2}^2 + ... + {e_m}^2}{m} $ $\displaystyle e~=~\sqrt{ \frac{{e_1}^2 + {e_2}^2 + ... + {e_m}^2}{m} } $ Ans(c): $\displaystyle p=Pb~~and~~P= \frac{a * a^t}{a^t * a} $ $\displaystyle a = \begin{bmatrix} 1\\ 1\\ 1 \end{bmatrix} $ But... $\displaystyle a^t * a ~=~ \begin{bmatrix} 1&1&1\\ 1&1&1\\ 1&1&1 \end{bmatrix} $ has no inverse.... Would anyone show me how to calculate the projection matrix? Last edited by zollen; January 14th, 2018 at 07:25 AM. 

Tags 
problem, squares 
Thread Tools  
Display Modes  

Similar Threads  
Thread  Thread Starter  Forum  Replies  Last Post 
Least Squares Problem  ALLIRIX  Linear Algebra  2  January 27th, 2015 01:15 AM 
A difficult problem in geometry  A Triangle and Two Squares  metavdr  Geometry  7  February 8th, 2014 08:59 AM 
squares  gaussrelatz  Number Theory  6  September 1st, 2012 04:56 AM 
A tough problem about perfect squares  elim  Number Theory  0  April 19th, 2010 08:15 AM 
Problem of Squares  hemi  Algebra  3  February 19th, 2009 01:00 AM 