Values of a matrix for which a system represents stable or unstable systems

1.2k Views Asked by At

For $\alpha$, $\beta$, $\gamma$ $\in\mathbb{R}$, consider the system $$\frac{d}{dt}x=\underbrace{\begin{bmatrix}0 & \alpha & \gamma & 0 \\ -\alpha & 0 & 0 & 0 \\ 0 & 0 & 0 & \beta \\ 0 & 0 & -\beta & 0 \end{bmatrix}}_{A}x$$

I want to determine for which values of $\alpha,\beta,\gamma\in\mathbb{R}$ the system represents a stable, asymptotically stable, or unstable system.

Then $$\begin{aligned}\det(I\xi-A)&=\begin{vmatrix}\xi & -\alpha & -\gamma & 0 \\ \alpha & \xi & 0 & 0 \\ 0 & 0 & \xi & -\beta \\ 0 & 0 & \beta & \xi\end{vmatrix}\\&=\xi\begin{vmatrix}\xi & 0 & 0 \\ 0 & \xi & -\beta \\ 0 & \beta & \xi \end{vmatrix}-\alpha\begin{vmatrix}-\alpha & -\gamma & 0 \\ 0 & \xi & -\beta \\ 0 & \beta & \xi \end{vmatrix} \\&=\xi^{2}\begin{vmatrix}\xi & -\beta \\ \beta & \xi \end{vmatrix}+\alpha^{2}\begin{vmatrix}\xi & -\beta \\ \beta & \xi\end{vmatrix} \\ &=\xi^{2}(\xi^{2}+\beta^{2})+\alpha^{2}(\xi^{2}+\beta^{2}) \\ &=\xi^{4}+\beta^{2}\xi^{2}+\alpha^{2}\xi^{2}+\alpha^{2}\beta^{2} \\&=(\xi^{2}+\alpha)(\xi^{2}+\beta)\end{aligned}$$

Hence the eigenvalues are given by $$\lambda_{1,2}=\pm\sqrt{-\alpha}=\pm i\sqrt{\alpha}$$

$$\lambda_{3,4}=\pm\sqrt{-\beta}=\pm i\sqrt{\beta}$$

Thus $\Re(\lambda_{i})=0$ for $i\in[1,4]$ and are nonsemisimple, thus the system is unstable for all $\alpha,\beta,\gamma\in\mathbb{R}$. Is that correct?

1

There are 1 best solutions below

2
On BEST ANSWER

Let's say $\alpha \neq \beta$. Then there is no repetitive eigenvalues exist, so the system is marginally stable, but not asymptotically stable.

Now assume $\alpha = \beta$ and $\gamma = 0$. In this case even if the eigenvalues have algebraic multiplicity 2, they also have geometric multiplicity 2, so the system is again marginally stable.

İf $\alpha = \beta$ and $\gamma \neq 0$, then the eigenvalues have geometric multiplicity 1 while having algebraic multiplicity 2, hence the system is unstable.

Note: Characteristic polynomial isn't enough to conclude the stability in case of repeated eigenvalues. For example let's look at the following systems: $$\dot{x} = \begin{bmatrix}0 & 0 \\ 0 & 0\end{bmatrix} x$$ $$\dot{x} = \begin{bmatrix}0 & 1 \\ 0 & 0\end{bmatrix} x$$ Both matrices have the same characteristic equation, but the first system is stable and the second system is not. Because in the first system there are two linearly independent eigenvectors for the eigenvalue $0$, i.e. geometric multiplicity 2, but in the second system there is only one linearly independent eigenvector, i.e. geometric multiplicity 1.