0
$\begingroup$

Here are the preliminaries of my question:

Let $\Omega$ be a compact metric space, and suppose a flow $(\Omega,\mathbb{R})$ is given. Let $A\colon\Omega\to M^2$ be continuous. Here $M^2$ is the set of the real $2\times 2$-matrices. If $\omega\in\Omega$ and $x_0\in\mathbb{R}^2$, define $(\omega,x_0)\cdot t=(\omega\cdot t,x(t))$, where $x(t)$ is the solution to $$ E(\omega):~~~\dot{x}=A(\omega\cdot t)x, $$ satisfying $x(0)=x_0$. (We read $E(\omega)$ as "the equation corresponding to $\omega$.") Now write $A\colon\Omega\to M^2$ $$ A(\omega)=\begin{bmatrix}a(\omega) & -b(\omega)\\b(\omega) & a(\omega)\end{bmatrix}+\begin{bmatrix}\delta(\omega) & \epsilon(\omega)\\\epsilon(\omega) & -\delta(\omega).\end{bmatrix} $$ In polar coordinates $r,\theta$, equation $E(\omega)$ becomes $$ \frac{d}{dt}\theta(t)=b(\omega t)+\epsilon(\omega t)\cos 2\theta(t)-\delta(\omega t)\sin 2\theta(t),~~~~~(3)\\ \frac{d}{dt}\ln r(t)=a(\omega t)+\delta(\omega t)\cos 2\theta(t)+\epsilon(\omega t)\sin 2\theta(t).~~~(4) $$ Moreover define $$ \sigma\colon\Omega\times S^1\to\Sigma, $$ whereat $S^1=\left\{x\in\mathbb{R}^2 : \lVert x\rVert =1\right\}$ and $\Sigma=\Omega\times P^1$. Let $P^1$ be obtained from $S^1$ by identifying antipodal points. Let $\Psi$ be a flow on $\Sigma$; letting $\phi=\exp(i\cdot 2\theta)$ be a coordinate on $P^1$, we will write $(\omega,\phi)\cdot t=(\omega t,\phi(t))$ when referring to this flow.

Now consider the following:

Let $x(t)$ be a solution to equation $E(\omega)$. For fixed $\tau$, let $(\omega,\Phi_0)=\sigma(\omega,x(\tau)/\lVert x(\tau)\rVert)$. Assume that the trace of $A(\omega)$ is $0$, i.e. $\text{trace}(A(\omega))=0$ $(\omega\in\Omega$).

Now the text says:

By the assumption $\text{trace}(A(\omega))=0$ $(\omega\in\Omega$) and equation (4), we have $$ \frac{1}{t-\tau}\ln\frac{\lVert x(t)\rVert}{\lVert x(\tau)\rVert}=\frac{1}{t-\tau}\int_{\tau}^t f(\omega s,\phi_0(s))\, ds, $$ where $$ f(\omega,\phi)=\delta(\omega)\cos 2\theta + \epsilon(\omega)\sin 2\theta,~~\phi=\exp(i\cdot 2\theta). $$

My question is how we get the equation $$ \frac{1}{t-\tau}\ln\frac{\lVert x(t)\rVert}{\lVert x(\tau)\rVert}=\frac{1}{t-\tau}\int_{\tau}^t f(\omega s,\phi_0(s))\, ds. $$

Could you pls explain that to me?

$\endgroup$
1
$\begingroup$

You have $$ \dot z = f(t) $$ where $z(t)=\ln(r(t)$ and $f(t)=δ(ωt)\cos(2θ(t))+ϵ(ωt)\sin(2θ(t))$ So obviously $$ z(t)=z(\tau)+\int_\tau^t f(s) ds $$ Namely $$ z(t)-z(\tau)=\ln(r(t))-\ln(r(\tau)=\ln(\frac{r(t)}{r(\tau)})=\int_\tau^t f(s) ds $$ but $r(t)=\sqrt{x_1^2(t)+x_2^2(t)}=||x(t)||$

The fact that $f(s)=f(\omega(s),\phi(s))$ follows from the condition $a(\omega)=0$ (trace of $A$ is equal to zero)

$\endgroup$
  • $\begingroup$ And where does the factor $\frac{1}{t-\tau}$ come from and which role does $\sigma$ play? $\endgroup$ – mathfemi Oct 22 '14 at 19:52
  • $\begingroup$ This the same factor on the both sides so it is not relevant. What $\sigma$ you are talking about? $\endgroup$ – Alexander Vigodner Oct 22 '14 at 19:54
  • $\begingroup$ The $\sigma$ I mentioned in the preliminaries, i.e. the fact that $f(\omega s,\Phi_0(s))=f((\sigma(\omega,x(\tau)/\lVert x(\tau)\rVert)\cdot s)=f(\Psi(\sigma(\omega,x(\tau)/\lVert x(\tau)\rVert),s)$ $\endgroup$ – mathfemi Oct 22 '14 at 19:58
  • $\begingroup$ The result is not connected to $\sigma$. $\sigma$ is your internal way to describe the system. $\endgroup$ – Alexander Vigodner Oct 22 '14 at 20:09
  • $\begingroup$ Another and last question: What exactly is $P^1$? I cannot imagaine it. $\endgroup$ – mathfemi Oct 22 '14 at 20:28

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.