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 $f(x,y)$ be some bounded with its derivatives continuous function on $\Omega \times \overline{\Omega}$, where $\Omega$ is a domain in $\mathbb{R}^n$. Let $f(\,\,\cdot\,,\,y) \in \mathcal{E}(\Omega)$ for any fixed $y$. Let $L_{x} \in \mathcal{E}'(\Omega)$. Is it true that

$$\int\limits_{\overline{\Omega}} L_{x}f(x,y) \, \mu(dy) = L_{x}\int\limits_{\overline{\Omega}} f(x,y) \, \mu(dy)$$

holds for any probability measure $\mu$ in $\overline{\Omega}$? If it is true, how to show it?

If $f(x,y) \in \mathcal{E}(\Omega \times \Omega')$ where domain $\Omega'$ is such that $\overline{\Omega} \subseteq \Omega'$ then the equality holds by virtue of the tensor product of distributions theorem.

share|cite|improve this question
Do you need that the derivatives of $f(\cdot,y)$ to be bounded, in other to ensure us that $x\mapsto \int_{\overline \Omega}f(x,y)\mu(dy)$ is infinitely many times differentiable? – Davide Giraudo May 8 '12 at 15:08
Thanks, I've corrected the question. – Nimza May 8 '12 at 17:46
up vote 5 down vote accepted

Suppose first that $L_x\in C^\infty_0(\Omega)$. Then the equality you ask about is Fubini's theorem.

Suppose now that $L_x$ is not necessarily smooth. Choose a sequence $\newcommand{\ve}{\varepsilon}$ $L_{\ve,x}\in \mathscr{E}'(\Omega)$ that converges to $L_x$ in the weak sense. Then one needs to prove that

$$ \lim_{\ve\to 0} L_{\ve,x}\int_\Omega f(x,y)d\mu(y)=L_x\int_\Omega f(x,y)d\mu(y), \tag{A} $$

$$ \lim_{\ve\to 0}\int_\Omega (L_{\ve,x}-L_x)f(x,y)d\mu(y)=0. \tag{B} $$

The equality (A) is an immediate consequence of the weak convergence. The equality (B) requires an additional assumption on $f$.

Denote by $K$ a compact set containing the support of $L_x$ and $L_{\ve, x}$, $\ve$ sufficiently small. If we assume that for any multi-index $\alpha$ we have

$$ \sup_{x\in K, y\in \Omega} \partial^\alpha_x f(x,y) <\infty, \tag{C} $$

then (B) follows by invoking the uniform boundedness principle for $\mathscr{E}'(\Omega)$ which states that if a sequence $u_n \in \mathscr{E}'(\Omega)$ converges weakly to $0$, then $ u_n(\phi)\to 0$ uniformly for $\phi$ in a bounded subset of $\mathscr{E}(\Omega)$.

I recall that a subset $\Phi\subset \mathscr{E}(\Omega)$ is bounded if for any compact $K\subset \Omega$ and any multi-index $\alpha$ we have

$$ \sup_{x\in K, \phi\in \Phi} \partial^\alpha_x\phi(x) <\infty. $$

Update. Let me set $\phi_y:=f(x,y)$. To insure the integrability of $y\mapsto L(\phi_y)$ for any $L\in\mathscr{E}'(\Omega)$ it suffices to assume that the map $\Omega\ni y\mapsto \phi_y\in\mathscr{E}(\Omega)$ is continuous, i.e., for any $y_0\in \Omega$, any $\ve>0$, any compact $K\subset \Omega$ and any multi-index $\alpha$ there exists $\delta>0$ such that

$$|y-y_0|<\delta \Rightarrow \sup_{x\in K}\left|\partial^\alpha_x\bigl(\; \phi_y(x)-\phi_{y_0}(x)\;\bigr )\right| <\ve. $$

share|cite|improve this answer
Great thanks! Can you give me please some reference on such version of uniform boundness principle? – Nimza May 8 '12 at 15:34
The most complete reference would be Francois Treve, Topological Vector Sapeces, Distributions and Kernels, now in Dover. However this is a bit harder to digest given its generality. A more readable and helpful source would be vol. 2 of Gelfand and Shilov's treatise on generalized functions. There they investigate complete countably normed spaces, or Frechet spaces and they prove among other things the uniform boundedness principle. The space $\mathscr{E}(\Omega)$ is such a space. – Liviu Nicolaescu May 8 '12 at 16:00
Thanks for update and for references. – Nimza May 8 '12 at 17:49

Davide's right. Neither integral makes sense because the function was only continuous. If you assume $f$ is a smooth test function, then a priori it's only clear when $\mu$ is a finite linear combination of point masses. Thus, you cannot avoid using a Riemann sums trick: approximate the more general measure with a sequence of finitely supported measures $\mu_n$. My impression is that the general theory of distributions cannot start without taking Riemann-type sums at some point and that any argument probably has this maneuver underlying it somewhere.

For the right hand side, you need to prove that $\int f(x,y) d\mu_n \to \int f(x,y) d\mu(y)$ in some $C^k$ topology as functions of $x$ over the support of $L_x$. For the left hand side, check that $\mu_n \rightharpoonup \mu$ weakly and $L_x f(x,y)$ is continuous in $y$. This step again uses that $L_x$ is continuous with respect to $C^k$ convergence for some $k$.

Note, if you establish this identity when $\mu$ is, say, an absolutely continuous measure with a smooth density function, then you can pass to the limit for a general finite measure by using a mollifying kernel (analogous to taking $L_{\epsilon, x}$ in Liviu's argument, but this is a mollification in the $y$ variable, and is just measure theoretic).

share|cite|improve this answer
@ Phil The collection of functions $\Phi:=(\; f(-,y)\;)_{y\in \Omega}$ is bounded in $\mathscr{E}(\Omega)$ due to (C). Set $$s(\varepsilon):=\sup_{y\in \Omega} |(L_{\varepsilon, x}-L_x)f(x,y)|$$ The uniform boundedness principle implies $s(\varepsilon)\to 0$. Hence $$\left|\;\int_\Omega(L_{\varepsilon, x}-L_x)f(x,y)d\mu(y)\;\right| \leq \int_Omega s(\varepsilon) d\mu(y)=s(\varepsilon)\to 0. $$ – Liviu Nicolaescu May 8 '12 at 16:23
(It looks like there is a coding error in the comment.) I think I'm starting to see the relationship between these two arguments. On the one hand, your argument appeals to Fubini to get started; the argument that I suggested actually avoids Fubini's theorem by taking Riemann sums. I actually thought that you couldn't get away without this move (for the reason in my post), but it looks like you can just get by with the continuity in the $y$ variable? Thoughts? – Phil Isett May 8 '12 at 20:35

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.