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 posted the following question also here, but thought that I can get more answers in MO. Let $(\Omega,\Sigma)$ be a measurable space and $\nu_1$, $\nu_2$ two probability measures on it. For $i=1,2$, let $L_{\nu_i}:L^{1}(\nu_i)\rightarrow\mathbb R$ be the map defined by $L_{\nu_i}(g)=\int_\Omega gd\nu_i$. Denote by $[1_E]_{\nu_i}$ the equivalence class in $L^1(\nu_i)$ of the characteristic function $1_E$. Does there always exist a linear map $M:L^1(\nu_1)\rightarrow L^1(\nu_2)$ such that $$L_{\nu_2}\circ M([1_E]_{\nu_1})=L_{\nu_1}([1_E]_{\nu_1})$$ for all $E\in\Sigma$? When $\nu_1$ is absolutely continuous to $\nu_2$ then, by the Radon-Nykodim Theorem, the map $M$ can be given explicitly, i.e. there exists a $\Sigma$-measurable function $f:\Omega\rightarrow\mathbb R$, such that $M(g):=f\cdot g$ will satisfy the equation given above. So the question becomes interesting when $\nu_1$ and $\nu_2$ are singular to each other, and the spaces $L^1(\nu_1)$ and $L^1(\nu_2)$ are not isometric. The way I think of it is, that it could characterize in some/all cases a Radon-Nykodim derivative when the measures are singular to each other...comments on this are welcome!

share|cite|improve this question
up vote 1 down vote accepted

How about $M \colon L^1(\nu_1)\rightarrow L^1(\nu_2) \colon g \mapsto (x \mapsto \int g~d\nu_1)$?

share|cite|improve this answer
ok, you are right. so the above equality does not really charachterize the radon-nikodym derivative! – Andy Teich Nov 1 '12 at 13:23
one should impose that $M$ is injective (???) on the set of all characterisitc functions in $L^1(\nu_1)$ in order to make my question reasonable... – Andy Teich Nov 1 '12 at 14:47
Then the answer would be no (in general): take $\nu_1 = \mathcal{L}|_{[0,1]}$ and $\nu_2 = \delta_0$ (with $\Omega = [0,1]$). Then the $M$ above is the only possible mapping with the desired properties. – Tapio Rajala Nov 1 '12 at 14:54

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.