Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. It's 100% free, no registration required.

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

Let $(X,\mathcal{B},\mu)$ be a standard probability space, and let $T:X\rightarrow X$ be a measurable, measure-preserving transformation, i.e. for every $A\in\mathcal{B}$, $\mu(T^{-1}(A))=\mu(A)$. Consider the operator $U_T:L_2(\mu)\rightarrow L_2(\mu)$ given by $U_T (f)=f\circ T$. Clearly it is a linear isometry. My question is whether $U_T$ is unitary.

share|cite|improve this question
For reference, the mapping $T\longrightarrow U_T$ is called the Koopman representation. – 1015 Jan 14 '13 at 13:21
up vote 3 down vote accepted

The answer is yes when $T$ is invertible and with measurable inverse. In this case, $(U_T)^*=U_{T^{-1}}$ gives the adjoint.

More generally, in Peter Walters' book An introduction to Ergodic theory, we find the following result:

Theorem 2.8. Let $T$ be a measure preserving transformation on the probability space $(\Omega,\mathcal B,\mu)$. Then $U_T\colon L^2(\mu)\to L^2(\mu)$ is surjective if and only if $M\colon (\widetilde{\mathcal B},\widetilde{\mu})\to (\widetilde{\mathcal B},\widetilde{\mu})$is surjective, where $\widetilde{\mathcal B}$ is $\mathcal B$ where we identify set $A$ and $B$ when $\mu(A\Delta B)=0$, $\widetilde{\mu}$ is defined canonically and $M(B)=T^{-1}(B)$.

If $U_T$ is bijective and $A\in\mathcal B$, let $f$ such that $U_Tf=\chi_{\widetilde A}$. Then $U_T(f\cdot f)=U_T(f)$ so $f\cdot f=f$ and $f=\chi_{\widetilde C}$ for some $C\in\mathcal B$. So $\widetilde A=M(\widetilde C)$ and $M$ is surjective.

The latter condition is not always satisfied, as 5PM points out. Take $X=[0,1]$, $\mathcal B$ the Borel $\sigma$-algebra, $\mu$ the Lebesgue measure and $T(x):=2x\operatorname{ mod }1$. In this case $U_T$ is not onto as the function which are not $1/2$-periodic are not in the range.

share|cite|improve this answer
When $T$ is invertible it is clear. But if it is not invertible, what happens then? – user25640 Dec 27 '12 at 13:23
So: we should show that if $T$ is not injective (a.e. or something), then not everything in $L_2$ can be written as $f \circ T$ a.e. ... – GEdgar Dec 27 '12 at 13:27
In the book: "Recurrence in Ergodic Theory and Combinatorial Number Theory" by H. Furstenberg this operator $U_T$ is reffered to as a unitary operator and I didn't see an assumption that $T$ is invertible. – user25640 Dec 27 '12 at 14:16
@Davide Giraudo Can you think of a measure-preserving transformation $T$ of a standard probability space such that the induced function $M$ is not surjective? – user25640 Dec 29 '12 at 16:45
@user25640 $T(x)=2x \mod 1$ is a measure-preserving transformation on $[0,1]$ equipped with the Lebesgue measure. That is, for every measurable set $E\subset [0,1]$ we have $|T^{-1}E|=|E|$. Only certain functions can be written as $u\circ T$: those that are $1/2$-periodic. – user53153 Dec 29 '12 at 17:45

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

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