Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. Join them; it only takes a minute:

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

On a measurable space, let $m_n, n=1, \dots, $ be a sequence measures on it and $\mu$ is another measure.

let $Z := \{ϕ∈L^2(\mu): ∫ ϕ\,d\mu = 0 \}$.

The first type of convergence of $m_i$'s is defined as $$ \lim_{n \to \infty} \sup_{\phi\in Z:\,\|\phi\|_2=1} \int \phi \, dm_n =0. $$ This is motivated by the definition of a stochastic process being ρ-mixing.

The second type of convergence of $m_i$'s is defined as $$ \lim_{n \to \infty} \sup_{\phi\in Z:\,\|\phi\|_\infty=1} \int \phi \, dm_n =0. $$ This is motivated by the definition of a stochastic process being α-mixing.

The third type of convergence of $m_i$'s is defined as $$ \lim_{n\to \infty} \sup_{0\leq\phi\leq1} \Big| \int \phi \, dm_n - \int \phi \, dQ \Big| = 0 $$ This is motivated by the definition of a stochastic process being β-mixing. A side question: what does $0\leq\phi\leq1$ really means?

Are there names for these types of convergence of measures? I would like to know about their relations with other types of convergence of a sequence of measures, such as weak and weak* convergence.

I have tried to read Chen, Xiaohong; Hansen, Lars Peter; Carrasco, Marine (2010). "Nonlinearity and temporal dependence". Journal of Econometrics 155 (2): 155–169. cited in the Wikipedia article for the three types of mixing, and also traced Mixing Conditions for Markov Chains Yu. A. Davydov Theory Probab. Appl. 18-2 (1974) cited in that paper for the definitions of β-mixing. But I failed to see answers to my above questions.


share|cite|improve this question

I think that $\rho$-convergence is $L^2$ convergence to zero. For simplicity assume that the measures $m_n$ are absolutely continuous with respect to $\mu$, with Radon-Nikodym density $q_n$. Then you want $$ \int q_n^2 d \mu \rightarrow 0. $$ By Holder's inequality this implies the conergence you give, and if this $L^2$ convergence doesn't hold, you can construct a $\phi$ to violate your condition.

$\alpha$-mixing looks like $L^1$ convergence to zero. This is $$ \int q_n d \mu \rightarrow 0. $$ We don't need absolute values since $q_n$ is nonnegative. A lot of this is related to duals of $L^p$ spaces. The dual of $L^2$ is $L^2$, while the dual of $L^\infty$ is $L^1$.

In your definition of $\beta$-mixing, what is $Q$.

Note, for example that $\rho$-mixing is stronger than weak convergence in $L^2$. That would be equivalent to saying that for each $\phi\in L^2$, $$ \int \phi q_n d \mu \rightarrow 0, $$ without saying anything about convergence holding uniformly over all such $\phi$ with $||\phi||_2 = 1$.

share|cite|improve this answer
Thanks! In the definition of β-mixing, Q is the measure $M_n$'s converge to. – Tim Dec 1 '12 at 4:40

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.