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

Suppose $\mu$ is a positive Borel measure on $\mathbb{R}^n$ and $V$ is an open set. Define the map $\varphi\colon \mathbb{R}^n\to [0,\infty]$ by $\varphi(x)=\mu(x+V)$, where $x+V$ is the translation of $V$ by $x$.

I am told that $\varphi$ is lower semi-continuous, provided that $\mu$ is finite. I have produced a proof of this, but my proof doesn't seem to use the finiteness assumption. This goes roughly as follows: suppose $x_0\in\varphi^{-1}((\alpha,\infty))$ isn't an interior point. This gives a sequence $y_n$, converging to $x_0$, with $\varphi(y_n)\leq\alpha$. Since $V$ is open, each $x\in x_0+V$ lies in the sets $y_n+V$ from some point onward. Write $x_0+V$ as an increasing union, based on the index at which points land in the sequence $y_n+V$. But then $\varphi(x_0)$ is the limit of a sequence, bounded above by $\alpha$. Contradiction.

I have spent some time looking for an oversight on my part in the proof, but I can't find one. Could someone please confirm that the finiteness assumption is unnecessary, or alternatively, point to a flaw in my proof or give a counterexample?

share|cite|improve this question
By the way, what do you mean by an infinite measure? Is it one that assigns infinite measure to some sets? For example, like Lebesgue measure on the line. – Patrick Mar 7 '12 at 22:45
@Patrick Yes, by an infinite measure I mean a measure that assigns the whole of $\mathbb{R}^n$ infinite measure. Try saying that three times fast. – Miha Habič Mar 7 '12 at 22:50
Your argument should begin with $x_0\in\varphi^{-1}((\alpha,\infty])$, but I think it is OK for $\mathbb R^n$. Maybe the assertion you saw was for some more general setting? – GEdgar May 7 '12 at 21:48
@GEdgar You're right, $x_0+V$ could have infinite measure. I don't think it matters in any case, since the boundary points of the upper level sets have low $\varphi$-value. – Miha Habič May 8 '12 at 6:14
The poser shows that the indicator functions converge ptwise, and then he could quote Fatou, which does not require finiteness. – mike May 10 '12 at 16:57
up vote 5 down vote accepted

For the sake of having an answer, I'll note that Mike's comment about Fatou's lemma works.

Take a sequence $x_n \to x$. Note that $\liminf 1_{V + x_n}(y) \ge 1_{V + x}(y)$ for every $y$. In other words, if $y \in V + x$, then $y \in V + x_n$ for all but finitely many $n$. The converse need not hold, so the inequality can be strict for some $y$, but that is okay.

Now Fatou's lemma says $$\liminf \int 1_{V + x_n} \ge \int \liminf\, 1_{V + x_n} \ge \int 1_{V + x}$$ which is to say $\liminf \varphi(x_n) \ge \varphi(x)$.

This does not require any assumptions on the measure, except that it be positive.

share|cite|improve this answer
Thanks to Mike and Nate for this solution. It's very elegant; I admit Fatou's lemma hadn't crossed my mind, but then again I don't recall ever using it for anything other than proving dominated convergence. – Miha Habič May 14 '12 at 15:48

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.