I was asked to linearize the first order ODE for an oscillating spring $$(x,v)^{'}=(v,-av^3-b(x^2-1))$$ where $a,b>0$.
Is it correct to simply find the jacobian of $(v, -av^3-b(x^2-1))$?
For the jacobian I got $$J=\begin{bmatrix}0&1\\-2bx&-3av^2\end{bmatrix}$$ After plugging in the critical points $(1,0)$ and $(-1,0)$ my matrix became $\begin{bmatrix}0&1\\-2b&0\end{bmatrix}$ and $\begin{bmatrix}0&1\\2b&0\end{bmatrix}$
After this, I used tried to find the eigenvalues by using the equations:
$\lambda^2-2b=0$ for $(1,0)$ and $\lambda^2+2b=0$ for $(-1,0)$.
By my calculations, $\lambda=\pm{\sqrt{2b}}$ for the point $(1,0)$ and $\lambda=\pm{\sqrt{-2b}}$ for the point $(-1,0)$.
This should mean $(1,0)$ is a saddle point while $(-1,0)$ is a an ellipse. This result does not seem intuitive though. Is there any point where I made a mistake? Should there not be a stable equilibrium for a spring?