# Is $\rho(f) := \mu(|f| \geq \epsilon), \epsilon >0$ some kind of generalized norm?

Let $\rho(f) := \mu(|f| \geq \epsilon)$ where $\epsilon >0$ is a constant and $\mu$ is a measure and $f$ is a measurable function. Then

• $f\equiv 0 \rightarrow \rho(f) = 0$
• when $|a| \neq 0$, $\rho(af) = \mu(|f| \geq \epsilon/|a|)$ not necessarily equals $|a| \rho(f)$. But it holds when $|a|=1$.
• it is not true that $$\rho(f+g) = \mu(|f+g| \geq \epsilon) \leq \mu(|f| \geq \epsilon) + \mu(|g| \geq \epsilon) = \rho(f) + \rho(g)$$ because $\{|f+g| \geq \epsilon\}$ might be nonempty, while $\{|f| \geq \epsilon\}$ and $\{|g| \geq \epsilon\}$ can be empty.

What kinds of generalized norm is it? BTW: the generalized norms that I have heard of are quasinorms (which generalize the triangle inequality in some way), F-norms (which generalize the positive homogenity in some way) and seminorms (which generalize the positive definiteness in some way).

Motivation: I am trying to see if convergence in measure can be understood wrt the family of "generalized norms" $\{\rho_\epsilon, \epsilon >0\}$.

Thanks and regards!

-
Converrgencce in measure can not be even descibed in terms of topology. In other words there is no topology which gives rise convergencce in measure. – Norbert Dec 29 '12 at 18:19
@Norbert: Thanks! Isn't it that convergence in measure can be induced from the quasinorm $\int \frac{|f|}{1+|f|}$? Or is it just for convergence in probability? C.f. Pavel M's comments following this post. – Tim Dec 29 '12 at 18:31
@Norbert You may be thinking of convergence a.e. – user53153 Dec 29 '12 at 20:34
@PavelM M, you are right. This is convergence aproximately everywhere – Norbert Jan 4 '13 at 17:31
You could be interested in weak $L^p$ spaces. – Davide Giraudo Jan 18 '13 at 20:57