Sufficient condition for strong mixing.

306 Views Asked by At

Given a measure-preserving transformation $T:X\to X,$ with $(X,\mathcal{A},\mu)$ a probability space, we say $T$ is strongly mixing if $$\mu(T^{-n}(A)\cap B)\to \mu(A)\mu(B)$$ when $n\to\infty$ for every $A,B\in \mathcal{A}.$

I know this definition to be equivalent to that for every $f,g\in L^{\infty}(X,\mathcal{A},\mu)$ we have $$(*) \int_X (f\circ T^n)g d\mu \to \int_X f d\mu \int_X g d\mu$$ when $n\to\infty.$

Now, I was asked to prove (Brin/Stuck exercise 4.3.6) that if $\mu$ is a Borel measure, $X$ is a compact topological space and for every $f,g$ continuous with zero integrals we have $$\int_X (f\circ T^n)g d\mu \to 0,$$ then $T$ is strongly mixing.

Certainly, every continuous function in $X$ is bounded, and I presume I have to get to (*), but I still haven't come with an idea of how to prove it for every function with nonzero integral. Any idea or suggestion will be appreciated.

1

There are 1 best solutions below

0
On BEST ANSWER

Any continuous function can be decomposed as $f = ( f - \int f ) + \int f $ where the first term has $0$ intergral, and the second is constant. You should be able to use your hypothesis and the linearity of integration to get ($*$) for all continuous functions.

Now that you have it for continuous functions, use the Stone-Weierstrass Approximation Theorem and X compact to finish the proof.