Stability of non-hyperbolic critical point in two dimensions, where the linearization has one zero egenvalue

443 Views Asked by At

I am having a two dimensional autonomous system

$S' = 2S^3+2S^2+\frac{1}{2}SA-\frac{3}{2}S-\frac{3}{4}A$

$A' = 4AS^2+A^2+4AS$

which exhibits a critical point at the origin $(S,A)=(0,0)$, and others. The eigenvalues of the linearization of the system at the origin yields

an eigenvalue $\lambda_1 = -\frac{3}{2}$ corresponding to the eigenvector $v_1=\left[\begin{matrix}1 \\ 0\end{matrix}\right]$,

and an eigenvalue $\lambda_2=0$ corresponding to the eigenvector $v_2=\left[\begin{matrix}-1/2 \\ 1\end{matrix}\right]$.

Hence there is a center manifold tangent to the second vector. I am new to center manifold theory, but found some theorems which look easy to apply for simple cases in two dimensions in the book of Perko – Differential Equations and Dynamical systems (3rd ED), section 2.11. The theorem however requires the system to first be put in a specific form. I quote from the book on p. 150:

We first consider the case when the matrix $A$ [that is the linearization matrix at the fixed point] has one zero eigenvalue, i.e. when $\det A=0$, but $\mathrm{tr}\,A\neq0$. In this case, as in Chapter 1 and as is shown in [A-I] on p. 338, the system (1) can be put into the form

$\dot x=p_2(x,y)$

$\dot y=y+q_2(x,y)$

where $p_2$ and $q_2$ are analytic in a neighborhood of the origin and have expansions that begin with second-degree terms in $x$ and $y$.

So I am certainly looking at such a case with one zero eigenvalue. However I do not see, how to bring it into the required form, and I need it in this form to apply the theorem which follows in the book right after.

The source [A-I] he quotes is A. A. Andronov, E. A. Leontovich, I. I. Gordon and A. G. Maier, Qualitative Theory of Second-Order Dynamical Systems, John Wiley and Sons, New York 1973. My library doesn't have it, it seems out of print, and I also couldn't find it anywhere on the internet.

The reference to Chapter 1 in his own book didn't help me either so far, but I will check that in greater detail.

Does anyone here have an idea how to bring the system into the required form?

Cheers!

1

There are 1 best solutions below

3
On BEST ANSWER

You use the eigenvalues and eigenvectors to diagonalise the system. So, you choose your coordinate axes along the eigenvectors, and write $$ \begin{pmatrix} S \\ A \end{pmatrix} = y\,v_1 + x\,v_2 = \begin{pmatrix} y-\frac{1}{2}x \\ x \end{pmatrix}, $$ which gives you the linear coordinate transformation $x = A$, $y = S + \frac{1}{2} A$. Hence, you obtain the system \begin{align} \dot{x} &= \text{second degree terms},\\ \dot{y} &= -\frac{3}{2} y + \text{second degree terms}. \end{align} Now, rescaling time by introducing $\tau = \lambda_1 t = -\frac{3}{2} t$, you transform the system into the required standard form.