Change of variables on the torus

176 Views Asked by At

Consider $A = \begin{bmatrix} 2 && 1 \\ 1 && 1 \end{bmatrix}$. Then $A$ acts on $\mathbb{T}^2 = (\mathbb{R}/\mathbb{Z})^2$ by matrix multiplication. If $f \in C(\mathbb{T}^2)$, then let $(\Phi_Af)(x) = f(Ax) $.

I want to compute a formal adjoint of $\Phi_A$, i.e. I want a map $L: C(\mathbb{T}^2) \to C(\mathbb{T}^2)$ with

$$ \int \Phi_A(f) g = \int f L(g)$$

I'm thinking you just change variables from $x$ to $A^{-1}x$ to conclude that $L(g)(x) = g(A^{-1}x)$. But, now, I'm worried about the exact domain of integration, since on the left I'm really integrating over $[0,1]^2$, while after the change of variables I get a different domain on the right. My gut tells me that because of the periodicity, the domain and domain change should not really matter. However, I'm having trouble exactly justifying why.

1

There are 1 best solutions below

2
On

You may argue in different ways. You may do it by hand: take the unit square S and transform it (to a parallelogram). This you may chop it into 4 pieces and translate those back to precisely cover S. A more abstract approach (valid for general $A\in GL_n({\Bbb Z})$) is to prove that $f$ is a diffeomorphism that preserves Lebesgue measure. For this you need to show e.g. that $Ax-Ax'\in {\Bbb Z}^2$ iff $x-x'\in {\Bbb Z}^2$ which is due to the fact that $A$ and its inverse preserve the lattice.