MathOverflow is a question and answer site for professional mathematicians. 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

Let $\Theta\subseteq\mathbb{R}^d$ is open set and $(\cal X, \cal A)$ be a measurable space . For every $\theta\in\Theta$, suppose that $P_\theta$ is a probability measure on $(\cal X, \cal A)$. Suppose we have measurable function $J:\cal X\times \Theta\rightarrow\mathbb{R}^d$such that \begin{equation} \int{J(x,\theta)dP_{\theta}(x)}=0,\quad \forall \theta\in\Theta \end{equation} and \begin{equation}\int{|J(X,\theta)|^2dP_{\theta}}<\infty, \quad \forall \theta\in\Theta. \end{equation} where $|.|$ denotes the Euclidean norm. If we know the function $f$ defined by $f(\theta)=\int{J(x,\theta)J(x,\theta)^T}dP_{\theta}$ is continuous at $\theta_0\in\Theta$, then I'd like to prove that \begin{equation} \lim_{a\to\infty}\limsup_{\theta\to\theta_0}\int|J(x,\theta)|^21_{\{|J(x,\theta)|\geq a\}}dP_{\theta}=0. \end{equation} To prove this, I consider the simplest case when $d=1$. We have \begin{align*} L&=\lim_{a\to\infty}\limsup_{\theta\to\theta_0}\int|J(x,\theta)|^21_{\{|J(x,\theta)|\geq a\}}dP_{\theta}\\ &=f(\theta)-\lim_{a\to\infty}\limsup_{\theta\to\theta_0}\int|J(x,\theta)|^21_{\{|J(x,\theta)|<a\}}dP_{\theta}. \end{align*} The first term tends to $f(\theta_0)$ due to the continuity of $f$. I am sure that the last term also tends to $f(\theta_0)$ since $a$ is very big, but I could not justify this mathematically. Could anyone help me please? I am also confused what the relation between $|.|$ and $f$ is when $d>1$. Thank you in advance

share|cite|improve this question
I don't think that you can prove this without adding some continuity conditions on $J$ and $P_\theta$. Your problem smells like uniform integrability. In order to use such things one could try to find a dominating measure $\lambda$ for the family $P_\theta$ and write \begin{equation} f(\theta) =\int|J(x,\theta)|^2 \varphi(\theta,x) d\lambda(x) \end{equation} with the $\lambda$-densities $f(\theta,\cdot)$ of $P_\theta$. – Jochen Wengenroth Jan 31 '14 at 8:26
However, if you an $\varepsilon$ more integrability like $\int |J(x,\theta)|^{2+\varepsilon}dP_{\theta}(x) \le C$ for $\theta$ close to $\theta_0$ then the almost trivial estimate $\int Y^2 I_{\{|Y|\ge a\}}dP \le a^{-\epsilon} \int |Y|^{2+\epsilon}dP$ helps. – Jochen Wengenroth Jan 31 '14 at 8:34
up vote 3 down vote accepted

Here is (I believe) a counterexample:

For $\theta\in\Theta=(-1,1)$ let $$P_\theta= \frac{\theta^2}{2} \delta_{-1/\theta} +\frac{\theta^2}{2} \delta_{1/\theta}+(1-\theta^2)\delta_0$$ (where $\delta_x$ is the Dirac measure in $x$) if $\theta\neq 0$ and $P_0=(\delta_{-1}+\delta_1)/2$. For $J(\theta,x)= x$ your function $f$ is constant and for $\theta_0=0$ you have \begin{equation} \lim_{a\to\infty}\limsup_{\theta\to\theta_0}\int|J(x,\theta)|^21_{\{|J(x,\theta)|\geq a\}}dP_{\theta}(x)=1. \end{equation}

share|cite|improve this answer
Thank you very much for your help. I'll modify my assumption in this problem. – Jlamprong Jan 31 '14 at 11:20
If you think that this answers your question you could accept the answer and then ask another one. – Jochen Wengenroth Jan 31 '14 at 12:34

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.