3

I am new to Dynamical Systems and recently I found the book of Brin and Stuck and decided to begin with this. I have a problem to understand one example it mentions about hyperbolic toral automorphisms.

To begin with, first we take the matrix $ A= \begin{pmatrix} 2 & 1 \\ 1 & 1 \end{pmatrix} $ , which has eigenvalues $ \lambda = \frac{3+\sqrt{5}}{2} $ and $\frac{1}{\lambda} $ and corresponding eigenvectors $ v_{\lambda}=(\frac{1+\sqrt{5}}{2},1)$ and $ v_{\frac{1}{\lambda}}=(\frac{1-\sqrt{5}}{2},1) $ . The matrix has integer entries, so it induces a map from the torus $ \mathbb{T}^{2}= {\mathbb{R}^2}/{\mathbb{Z}^2} $ to the torus $ \mathbb{T}^2$ written explicitly as $$ A \begin{pmatrix}x_1 \\ x_2 \end{pmatrix} = \begin{pmatrix} (2x_1 + x_2) \mod 1 \\ (x_1 + x_2) \mod 1 \end{pmatrix} $$ Ok,my problem is to understand what the book says in the next lines. Precisely it states : ''the lines in $\mathbb{R}^2$ parallel to the eigenvector $v_{\lambda}$ project to a family $W^u$ of parallel lines in $\mathbb{T}^2$. For $x\in \mathbb{T}^2$, the line $W^u (x)$ through $x$ is called the unstable manifold of $x$. The family $W^u$ partitions $\mathbb{T}^2$ and is called the unstable foliation of A. This foliation has the invariance property,meaning that $A(W^u(x))=W^u(A(x))$. Also $A$ expands each line in $W^u$ by a factor of $\lambda$. Similarly the stable foliation $W^s$ is obtained by projecting the lines of $\mathbb{R}^2$ parallel to $v_{1/{\lambda}}$. This foliation os also invariant under $A$ and $A$ contracts each stable manifold by a factor $1/{\lambda}$. Since the slopes of both eigenvectors are irrational, it follows that each stable and unstable manifold is dense in $\mathbb{T}^2 $ . Similarly any integer entried matrix $B \in \mathbb{Z}^{n\times n}$ induces a group endomorphism of the torus $\mathbb{T}^n$. This map is invertible iff $|\det(B)|=1$. If $B$ is invertible and the eigenvalues do not lie in the unit circle, then $B : \mathbb{T}^n \to \mathbb{T}^n $ has expanding and contracting subspaces of complementary dimensions and is calles a hyperbolic toral automorphism. One can show that all eigenvalues of a two-dimensional hyperbolic toral automorphism are irrational, hence the stable and unstable manifolds are dense in $\mathbb{T}^2$.''

My first question is this : ok,I understand why all the lines parallel to $v_{\lambda}$ are being projected to parallel lines of $\mathbb{T}^2$ (Each line parallel to $v_{\lambda}$ has the form $\{(x, \frac{\sqrt{5}-1}{2}x+b ) | x\in \mathbb{R} \}$ and this is mapped to $ \{ (2x + \frac{\sqrt{5}-1}{2}x+b \mod 1, x+\frac{\sqrt{5}-1}{2}x+b \mod 1) | x\in \mathbb{R} \}$). But I do not see why each manifold has the invariance property. Is this a matter of writing down and doing the calculations or is there a difficult idea behind it?

The second question is about the density argument. How can we prove that each line (in fact it is the CLASS of the line on the quotient ${\mathbb{R}^2}/{\mathbb{Z}^2}$, right? ) $W^u(x)$ is dense in $\mathbb{T}^2$ ?

Any hint will be much appreciated. Thank you a lot! :)

  • @user539887 Thank you a lot for your answer. Well,ok I proved the first fact about the invariance. Also I did prove that the set ${ ka (\mod 1) : k \in \mathbb{Z} }$ is dense in $[0,1]$ and in fact I can understand intuitively why $W^u(x)$ is dense in the torus, but I cannot write it down on the paper formally. How can we relate the density of the above set to the density of our manifold? – Petros Karajan Mar 21 '19 at 20:09
  • @user539887 Thank you a lot for your answer!! Can I ask you something? Do you know if it's possible to write you a message here on your inbox or,if not,do you want to send me your email address so I can ask you in more detail some things about hyperbolic sets and the horseshoe map ? – Petros Karajan Mar 27 '19 at 12:43
  • @user539887 ok,no problem. Listen,can I also ask you something related to the density argument. Ok,the manifold passing through the point $(0,0)$is indeed dense. But why does this imply that another manifold passing through a point $(x,y)$ is also dense?? – Petros Karajan Mar 29 '19 at 11:59
  • @user539887 Yes,this is true,but why does this imply the density of these translates also? Also,in your first argument,we showed that the points $(0,k a)$ belong to $W^{u}(0,0)$,but this does not imply that the set ${(k a,m a : k,m\in \mathbb{N} }$ also belongs to the manifold (which would in turn imply the density). – Petros Karajan Mar 30 '19 at 12:43
  • Ok then let's see if someone else can explain it more precisely,have a nice day! – Petros Karajan Mar 31 '19 at 19:25

0 Answers0