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

I'm interested in the following property, for a positive and locally bounded function $\omega:\mathbb{R}\to\mathbb{R}^d$, $d\ge 1$: there exists a countable sequence of open and pairwise disjoint sets $A_j$, such that

  1. $\mathbb{R}^d\setminus\bigcup_{j=1}^\infty A_j=N$, with $\mathcal {L} ^d(N)=0$ ($\mathcal L^d$ is the $d$-dimensional Lebesgue measure),
  2. $a_j\le\omega(x)\le b_j$ , for any $x\in A_j$, with $\frac{b_j}{a_j} \le C$ for some constant $C>0$ and for any $j$.

Since it is pretty involved and abstract, as a definition, I would like to have some criterion that ensures it to be true. At the moment I've understood the following:

  1. if $C_1\le\omega(\cdot)\le C_2$, with $C_1,C_2 >0$ then the property is obviously satisfied, with $A_1=\mathbb R^d$;
  2. taking $A_j=\omega^{-1}\big(]2^n,2^{n+1}[\big)$ , $n\in \mathbb Z$, the property is true, with $C=2$, for radial functions with continuous and injective radial part (since $N=\bigcup_n \omega^{-1}(2^n)$ is a union of spheres of dimension $d-1$);
  3. since if $\mathcal L^d(\omega^{-1}(2^n))>0$ then $2^n$ has to be a critical value, by the Sard Lemma, if we move a little the extrema of the above intervals, the property is true also for $\mathcal C^d(\mathbb R^d)$ functions ;
  4. if $f(\cdot)\le\omega(\cdot)\le g(\cdot)$, with $f$ and $g$ satisfying this property, then also $\omega$ satisfies this property.

Due to the very bad behavior of continuous functions, I think that I cannot extend directly the arguments of point 2 and 3 to general continuous functions (which is where I hoped this would be true). Is this true?

If this is the case, then there are some results that says when can I bound $\omega$ with $\mathcal C^d(\mathbb R^d)$ or continuous and radially injective functions?

share|cite|improve this question
Erm... Isn't it just a fancy way to say that $\log\frac{\limsup_x\omega}{\liminf_x\omega}\in L^\infty$? – fedja Sep 10 '12 at 23:15
It can be, but I do not completely get it. This property implies immediately that $\frac{\limsup_x \omega}{\liminf_x \omega}\in L^\infty$. Reasoning on a compact covering of $\mathbb R^d$, it seems to me that this is indeed a characterization. Why do you say that the $\log$ is needed? Sorry but my measure theory is pretty rusty! – Dario Prandi Sep 11 '12 at 10:10
Yes, it seems that one characterization is $\frac{\lim_{r \to 0}\sup_{B(x,r)} \omega}{\lim_{r \to 0}\inf_{B(x,r)} \omega}\in L^\infty$. (The converse follows by standard covering arguments.) Log does not make any difference, since this is always bounded below by $1$. – Tapio Rajala Sep 11 '12 at 10:56

In the case $\omega \colon \mathbb R^d \to \mathbb R$ is continuous, you can take $A_j = \omega^{-1}(]t_j,t_{j+1}[)$ with $t_i \in ]2^i,2^{i+1}[$ chosen so that $\mathcal L^d(\omega^{-1}(t_i)) = 0$ for all $i \in \mathbb Z$. The existence of such $t_i$ follows since $(\omega^{-1}(t))_{t \in \mathbb R}$ are disjointed and only for countably many $t \in \mathbb R$ they can have positive $\mathcal L^d$-measure.

share|cite|improve this answer

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.