Show that stable spriral has index of 1.

901 Views Asked by At

How do you go about generally showing that the index at the origin for different linear systems is equal to +1? For instance, the center of a stable spiral.


For a specific system, I could show how the vector directions change as we traverse the unit circle around the point. But how do I show that its true in general for a type of linear system? Is there a theorem I can use about how the index won't change for a closed curve for systems of the same type?

1

There are 1 best solutions below

0
On BEST ANSWER

There is a useful and well-known result which applies to this situation: the index of a vector field $V$ at a point $p$ where $V(p) = 0$ and $J_V(p)$, the Jacobean of $V$ at $p$, is non-singular, is equal to the sign of $\det J_V(p)$. A nice explanation of this, with proof, may be found here. For a non-degenerate stable spiral, the eigenvalues of $J_V(p)$ are a complex conjugate pair $\sigma \pm i\omega$, with $\sigma < 0, \omega \ne 0$; thus

$\det J_V(p) = (\sigma + i\omega)(\sigma - i\omega) = \sigma^2 + \omega^2 > 0, \tag{1}$

whence

$\text{index}V(p) = +1. \tag{2}$

It may be seen from (1) that the index of $V(p)$ at any zero $p$ with a complex conjugate pair of eigenvalues is also $+1$; thus unstable spirals (with $\sigma > 0, \omega \ne 0$) and elliptic points ($\sigma = 0, \omega \ne 0$) are also of index $+1$; if the eigenvalues of $J_V(p)$ are both real, say $0 \ne \lambda, \mu \in \Bbb R$, then we have

$\det J_V(p) = \lambda \mu, \tag{3}$

which shows that both stable ($\lambda, \mu < 0$) and unstable ($\lambda, \mu > 0$) nodes are of index $1$ as well; the only case (for two-dimensional vector fields at least) with $\det J_V(p) < 0$ occurs when $\lambda < 0 < \mu$ or $\mu < 0 < \lambda$; only saddles have index $-1$.

This technique is especially easy to apply to linear systems of the form $V(x) = Ax$, so that $V(0) = 0$, since then $J_V(0) = A$, as may be readily seen.

It is usually a lot easier to calculate $\det J_V(p)$ than to try and perform line integrals on circles around $p$ or otherwise analyze the way $V/\Vert V \Vert$ rotates as we traverse a path enclosing $p$. Of course at a degenerate $0$ of $V$, where $\det J_V(p) = 0$, this method breaks down and other devices must be employed to arrive at $\text{index}V(p)$.

Hope this helps! Cheerio,

and as always,

Fiat Lux!!!