Matrix similarity transformation

DryRun
Gold Member
Messages
837
Reaction score
4

Homework Statement


For a 3x3 matrix A, i know the eigenvalues and their corresponding 3 eigenvectors.

Define a matrix P such that ##PAP^{-1}## is a diagonal matrix.

Homework Equations


Similarity transformation formula: ##D=P^{-1}AP## where D is the diagonal matrix containing the eigenvalues as its trace elements.

The Attempt at a Solution


The 3 eigenvectors form a 3x3 matrix P.
The problem is that I've learned the formula to be: ##D=P^{-1}AP##
But what is being asked here is: ##PAP^{-1}##

I've re-arranged the matrix, ##D=P^{-1}AP##, into: ##PDP^{-1}=IAI## which gives ##PDP^{-1}=A## but i have no idea what to do next.
 
Physics news on Phys.org
Well, you can transponse left and right sides of D=P^{-1}AP, notice that D = D^T
 
Hi Karamata
Karamata said:
Well, you can transponse left and right sides of D=P^{-1}AP, notice that D = D^T
I don't understand what you mean.
D^T=PAP^{-1}?What would P be then?
 
sharks said:

Homework Statement


For a 3x3 matrix A, i know the eigenvalues and their corresponding 3 eigenvectors.

Define a matrix P such that ##PAP^{-1}## is a diagonal matrix.

Homework Equations


Similarity transformation formula: ##D=P^{-1}AP## where D is the diagonal matrix containing the eigenvalues as its trace elements.

The Attempt at a Solution


The 3 eigenvectors form a 3x3 matrix P.
The problem is that I've learned the formula to be: ##D=P^{-1}AP##
But what is being asked here is: ##PAP^{-1}##

I've re-arranged the matrix, ##D=P^{-1}AP##, into: ##PDP^{-1}=IAI## which gives ##PDP^{-1}=A## but i have no idea what to do next.

If you have ##D = P^{-1} A P##, then let ##Q = P^{-1}##. Do you see why this gives you the result you want?
 
spamiam said:
If you have ##D = P^{-1} A P##, then let ##Q = P^{-1}##. Do you see why this gives you the result you want?

If ##Q = P^{-1}## then does ##Q^{-1} = P##? If yes, then ##D=QAQ^{-1}##? Is there any proof for it? I'm not convinced. In that case, what would be the matrix Q?
 
Proving ##(Q^{-1})^{-1} = Q## is easy. What's the definition of the inverse of a matrix?

Well Q would be the inverse of the matrix P! :P Seriously though, if you have ##D = P^{-1} A P## or equivalently ##PD = AP## then what must the columns of P be? Think about eigenvectors.
 
OK, i think i understand the idea here.

Let ##Q = P^{-1}##, then ##Q^{-1} = P## and the problem changes to ##D=Q^{-1}AQ##, which corresponds to the standard form for the diagonal, D. Here, Q is the matrix of all 3 eigenvectors, and D is the diagonal matrix with the eigenvalues as its trace elements. Now, i need to find P, where ##P=Q^{-1}##. Is there an easier way to find ##Q^{-1}## other than finding the adjoint matrix of Q and its determinant?

Instead of using this formula to find ##Q^{-1}##: ##\frac{1}{|Q|}adj(Q)##

I'm thinking it might be easier to derive it from: ##D=Q^{-1}AQ## but i have no idea how.
 
Last edited:
sharks said:
OK, i think i understand the idea here.

Let ##Q = P^{-1}##, then ##Q^{-1} = P## and the problem changes to ##D=Q^{-1}AQ##
Don't you mean ##D = Q A Q^{-1}##?
, which corresponds to the standard form for the diagonal, D. Here, Q is the matrix of all 3 eigenvectors, and D is the diagonal matrix with the eigenvalues as its trace elements.

Right, except I think ##Q^{-1}## would be the matrix whose columns are the eigenvectors of A. Try writing out what PD = AP tells you in terms of the columns of P:
$$
PD =
\begin{pmatrix}
| & | & | \\
P_1 & P_2 & P_3 \\
| & | & |
\end{pmatrix}
\begin{pmatrix}
\lambda_1 & 0 & 0 \\
0 & \lambda_2 & 0 \\
0 & 0 & \lambda_3
\end{pmatrix}= \cdots
$$
$$
AP = A
\begin{pmatrix}
| & | & | \\
P_1 & P_2 & P_3 \\
| & | & |
\end{pmatrix} = \cdots
$$
 
I think you misunderstood the problem:
sharks said:
Define a matrix P such that ##PAP^{-1}## is a diagonal matrix.
So, ##D=Q^{-1}AQ## and Q is the matrix of eigenvectors of A.
 
  • #10
Ah, sorry I hadn't noticed you had edited your post!

Since you already know the eigenvalues and eigenvectors, just put them in a matrix and invert it using whatever method you prefer. You can use Cramer's rule, the equation for which you've listed below, or write the augemented matrix P|I and then row-reduce the left-hand side to the identity.

I don't see any shortcut using the equation ##D = Q A Q^{-1}##. Certainly you know what ##Q^{-1}## must be, but I think you just have to calculate its inverse as usual.

Edit: I think we're confusing each other about what P is and what Q is. I took P to be as in the similarity formula you listed in the OP, ##D = P^{-1} A P##, which would then make ##Q = P^{-1}## the solution you're looking for.
 
  • #11
spamiam said:
Edit: I think we're confusing each other about what P is and what Q is. I took P to be as in the similarity formula you listed in the OP, ##D = P^{-1} A P##, which would then make ##Q = P^{-1}## the solution you're looking for.
I know. :smile: So, if i find ##Q^{-1}## then i'll know the required matrix P. According to my calculations:
Q^{-1}= \begin{bmatrix}-1 & 0 & -2 \\ 1 & 1 & 3 \\ -1 & -1 & -2\end{bmatrix}
I tested the answer and got ##QQ^{-1}=I##

Thanks for the help.
 
  • #12
sharks said:
I know. :smile: So, if i find ##Q^{-1}## then i'll know the required matrix P. According to my calculations:
Q^{-1}= \begin{bmatrix}-1 & 0 & -2 \\ 1 & 1 & 3 \\ -1 & -1 & -2\end{bmatrix}
I tested the answer and got ##QQ^{-1}=I##

Thanks for the help.

Great! And just to make sure, you might want to calculate ##P A P^{-1}## and verify that you get the diagonal matrix D. And if it doesn't give you D, then I've succeeded in confusing you and you should see if ##P^{-1} A P## works! :-p
 
Back
Top