• Support PF! Buy your school textbooks, materials and every day products Here!

Least Squares

  • Thread starter Mindscrape
  • Start date
  • #1
1,860
0
This is a pretty basic question, but I just want to make sure. The question is to find the least square solution for
[tex] \newcommand{\colv}[2] {\left(\begin{array}{c} #1 \\ #2 \end{array}\right)}
\left(
\begin{array}{cc}
1 & 2\\
2 & 4
\end{array}
\right) x = \colv{2}{2}[/tex]

I can just find the orthogonal projection of the two vectors, right? In other words, use the find [tex]w = a_1\mathbf{v_1} + a_2\mathbf{v_2}[/tex]
[tex]\newcommand{\colv}[2] {\left(\begin{array}{c} #1 \\ #2 \end{array}\right)}
a_1 {\colv{1}{2}} + a_2 {\colv{2}{4}} = {\colv{w_1}{w_2}}[/tex]
where
[tex] a_1 = \frac{<b,v_1>}{||v_1||^2}[/tex]
and
[tex] a_2 = \frac{<b,v_2>}{||v_2||^2}[/tex]

I don't really want to use the method with positive definite because the numbers turn out sticky.

Then solution can be expressed as [tex]\mathbf{z} = \mathbf{b} - \mathbf{w}[/tex], such that z is orthogonal to the range of K?
 
Last edited:

Answers and Replies

  • #2
1,860
0
Two problems I just thought of though. If I want to do an orthogonal projection the basis vectors must be orthogonal. I suppose I could convert to an orthogonal basis.

Furthermore, K^T K = Kx will still not give an answer such that b is in the range of K. No wonder this problem was assigned, it's not as basic as I thought. :(

So if I convert the second vector to an orthogonal vector with Gram-Schmitt then I can use the orthogonal projection as the least squares answer?
 

Related Threads on Least Squares

  • Last Post
Replies
8
Views
2K
  • Last Post
Replies
7
Views
727
  • Last Post
Replies
3
Views
959
  • Last Post
Replies
1
Views
784
  • Last Post
Replies
4
Views
893
  • Last Post
Replies
3
Views
5K
  • Last Post
Replies
10
Views
737
  • Last Post
Replies
0
Views
2K
  • Last Post
Replies
1
Views
1K
  • Last Post
Replies
12
Views
2K
Top