Find the eigenvalues of a 3x3 matrix

AI Thread Summary
The discussion revolves around finding the eigenvalues of a 3x3 matrix related to a mass-spring system. The original poster struggles with the determinant calculation and seeks alternative methods. Another participant suggests setting variables differently to simplify the determinant calculation, ultimately leading to the correct eigenvalues. After some back-and-forth, the original poster realizes their misunderstanding and acknowledges the clarity of the explanation provided. The conversation highlights the importance of correctly interpreting the matrix and the variables involved in the eigenvalue calculation.
happyparticle
Messages
490
Reaction score
24
Homework Statement
Find the eigenvalues of a 3x3 matrix
Relevant Equations
##\begin{pmatrix} - \omega^2 +\frac{k}{m} &-\frac{k}{m} &0\\ -\frac{k}{M}& - \omega^2 + 2 \frac{k}{M} &-\frac{k}{M} \\
0&- \frac{k}{m} & -\omega^2 + \frac{k}{m}
\end{pmatrix} \begin{pmatrix}A_1\\A_2\\A_3 \end{pmatrix} =0##
Hi,
I have a 3 mass system. ##M \neq m##

I found the forces and I get the following matrix.

I have to find ##\omega_1 , \omega_2, \omega_3## I know I have to find the values of ##\omega## where det(A) = 0, but with a 3x3 matrix it is a nightmare. I can't find the values.

I'm wondering if there's another way to have the eigenvalues.
WxfG6.png

##\begin{pmatrix} - \omega^2 +\frac{k}{m} &-\frac{k}{m} &0\\ -\frac{k}{M}& - \omega^2 + 2 \frac{k}{M} &-\frac{k}{M} \\
0&- \frac{k}{m} & -\omega^2 + \frac{k}{m}
\end{pmatrix} \begin{pmatrix}A_1\\A_2\\A_3 \end{pmatrix} =0
##
 
Physics news on Phys.org
It is not a nightmare. Set ##W=-\omega^2-q-x\, , \,V=W-q\, , \,q=-k/M## and develop the determinant along the first row. It took me eight lines to find the eigenvalues.
 
fresh_42 said:
It is not a nightmare.
It's perhaps the sort of thing some people have nightmares about!
 
fresh_42 said:
It is not a nightmare. Set ##W=-\omega^2-q-x\, , \,V=W-q\, , \,q=-k/M## and develop the determinant along the first row. It took me eight lines to find the eigenvalues.

I'm not sure to understand the way you set the variables. Why you have ##x## and why not ##W = \omega^2 + \frac{k}{m}##
 
The eigenvalues are the roots of the polynomial ##\det (A-x\cdot I).## Hence we have ##a_{ii}-x## on the diagonal. You can also work with ##W-x## instead. I simply found it more convenient to calculate ##\det\left(\begin{bmatrix}
W&q&0\\q&V&q\\0&q&W
\end{bmatrix}\right) ## and replace the actual values afterwards rather than
##\det\left(\begin{bmatrix}
W-x&q&0\\q&W-q-x&q\\0&q&W-x
\end{bmatrix}\right) ##. It's a matter of taste.
 
You typed ##q = \frac{k}{M}##, but you also replaced ##\frac{k}{m}## by q even if ##M \neq m##. I don't know if it matters.

At the end I have ##VW^2 - \frac{2k^2W}{mM}## = 0. I didn't replace ##\frac{k}{m}##

Using quadratic formula I have ##W = 0 ## and ##W = -4k^2/mM##

then
##-\omega^2 -\frac{k}{m} - x = \frac{-4k^2}{mM}##

It Doesn't give me the right eigenvalues which are, ##\omega^2= 0, k/m , k/m + 2k/M##

As usual I probably just don't understood correctly the explanation.
 
You have to solve for the determinant ##\det(A-xI)## not just ##\det A##.
EpselonZero said:
You typed ##q = \frac{k}{M}##, but you also replaced ##\frac{k}{m}## by q even if ##M \neq m##. I don't know if it matters.

At the end I have ##VW^2 - \frac{2k^2W}{mM}## = 0. I didn't replace ##\frac{k}{m}##

Using quadratic formula I have ##W = 0 ## and ##W = -4k^2/mM##

then
##-\omega^2 -\frac{k}{m} - x = \frac{-4k^2}{mM}##

It Doesn't give me the right eigenvalues which are, ##\omega^2= 0, k/m , k/m + 2k/M##

As usual I probably just don't understood correctly the explanation.
You are right. I made a mistake as I used only one mass ##M=m##. Sorry.

My corrected result is ##\det (A-xI)= \ldots##
$$
0=W(VW-2qQ)=\left(-\omega^2+\dfrac{k}{m}-x\right)\cdot\left(\left(-\omega^2+\dfrac{k}{m}-x\right)\cdot \left(-\omega^2+2\dfrac{k}{M}-x\right)-2\dfrac{k^2}{mM}\right)
$$
which results in ##x=-\omega^2+\dfrac{k}{m}## and the quadratic equation
$$
0=x^2+2x\left(\omega^2-\dfrac{1}{2}\cdot\dfrac{k}{m}-\dfrac{k}{M}\right)+\left(\omega^4-\omega^2\dfrac{k}{m}-2\omega^2\dfrac{k}{M}\right)
$$
which is easy to solve (if I haven't made a mistake again).

In case ##\omega ## was your variable, then I do not understand where the square comes from. But then your calculation was correct. I also get ##0\, , \,\dfrac{k}{m}\, , \,\dfrac{k}{m}+2\dfrac{k}{M}.##
 
Last edited:
  • Like
Likes happyparticle
I may have misunderstood the problem. You said "I need to find the eigenvalues of a ##3\times 3## matrix" and then came the matrix ... or did you mean I need to solve ##\det (A-\omega I)=0## with eigenvalues ##\omega ##? I thought it was the first question and then we need a variable, say ##x## for the polynomial. The second possibility has ##\omega ## as variable, or ##\omega^2## in your case. But why ##\omega^2?##

So what is the matrix you want to find the eigenvalues to?
Or, what is the cubic polynomial you want to find the roots to?
 
fresh_42 said:
I may have misunderstood the problem. You said "I need to find the eigenvalues of a ##3\times 3## matrix" and then came the matrix ... or did you mean I need to solve ##\det (A-\omega I)=0## with eigenvalues ##\omega ##? I thought it was the first question and then we need a variable, say ##x## for the polynomial. The second possibility has ##\omega ## as variable, or ##\omega^2## in your case. But why ##\omega^2?##

So what is the matrix you want to find the eigenvalues to?
Or, what is the cubic polynomial you want to find the roots to?
Sorry I found my errors. You explanation was perfect. I just didn't realize I don't need to go that far to find the roots for ##\omega##

Thanks!
 
Back
Top