Diagonalization of matrix using change of variables

738 Views Asked by At

In linear algebra, we know that a system of equations $AX=b$ can be easily solved if $A$ is found to be of diagonal nature. If however $A$ is not diagonal but can be changed into a diagonal form by using the change of variables how can it be shown that the original system $AX=b$ and the new system $DY=z$ has indeed the same set of solutions?

Here, $X=Py$ and $Y=P^{-1}X$. Also $b=Pz$. So, $P^{-1}APY=z$ or $DY=z$

For example, if $AX=b$ is consistent in nature, then the transformed system $DY=z$ is also found to be consistent in nature. Likewise, if the original system is inconsistent, then $DY=z$ is also found to be inconsistent.

My Question is this : Is there any validity or proof for these assumptions ?

Thanks in advance guys! If this is a very basic question consider closing it but do post the link to the relevant solution.

1

There are 1 best solutions below

1
On BEST ANSWER

In the square case, everything comes down to eigenvalues. In this case, suppose $A$ is similar to $D$ and $D$ is diagonal. Then:

  • $AX=b$ is consistent for all $b$ if and only if ...
  • $A$ is invertible if and only if ...
  • None of $A$'s eigenvalues are zero if and only if ...
  • None of the diagonal entries of $D$ are zero if and only if ...
  • The system $DY=z$ is consistent for all $z$.

When the systems are consistent, you get the correspondence of the solutions by undoing the change of variable: if $z=P^{-1}b$ and $DY=z$ then $A(PY)=b$.

In the case where $A$ is $m \times n$ with $m>n$, you can do the same thing by considering the least squares problem, because any solution to the original problem is also a solution to the least squares problem.

I'm not sure how I would handle the case $m<n$.