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


After reading the previous post, I still have some doubts. Let's consider everything on $R$ to avoid complications.

  1. Can we say that any distribution $\mu\in\mathcal{D}'(R)$ of zero order is a signed radon measure?

  2. Since $\mu\in\mathcal{D}'(R)$ which is non-negative on non-negative test functions $C_c^\infty(R)$ is a positive radon measure, it is natural to ask what is corresponding part for the Schwartz distribution $\mu\in\mathcal{S}'(R)$ which is non-negative on non-negative test functions $\mathcal{S}(R)$? Intuitively, it is the radon measure whose mass grows slowly at infinity. Is there a name for this measure?

  3. For a radon measure $\mu$ on $R$, can we apply the Lebesgue's decomposition locally with a compact set fixed (say $K=[-a,a]$)? Then $\mu_K=\mu_{ac}+\mu_{sc}+\mu_{pp}$. The absolutely continuous part $\mu_{ac}$ corresponds to an absolutely continuous function; the pure point part $\mu_{pp}$ corresponds to sum of delta functions. How about the singular continuous part $\mu_{sc}$? EDIT: This question can also be put in the following way: Let $f$ be a singular function, what can we say $\int f\psi d x$ with $\psi\in\mathcal{D}(R)$?

Thank you for your help! :-)

Best Anand

share|cite|improve this question
The singular part corresponds to a singular continuous function(!). That is: a continuous function such that for any eps>0, there exists a disjoint collection of intervals whose combined length is 2a-eps on which the variation is less than epsilon. – Anthony Quas Jul 12 '11 at 17:00
Thanks Prof. Anthony Quas. Then can we view the singular continuous function as a distribution? – Anand Jul 12 '11 at 18:04
up vote 7 down vote accepted

To answer the first question: yes, at least locally. That is, given a distribution $u$ of order $0$, compactly supported for simplicity, the "order 0" condition asserts that $u$ factors through the space $C^o_c(U)$ of continuous compactly supported functions _with_the_corresponding_topology_. Then invoke the Riesz representation theorem to obtain the corresponding measure. Reduce to this case by a locally finite smooth partition of unity.

I cannot instantly answer the second question, but would try to reduce it to the compactly-supported case by a partition of unity argument.

I think the comment deals with the question about the "singular continuous" part.

share|cite|improve this answer
Thanks Prof. Paul Garrett. I am clear now for the first one. – Anand Jul 14 '11 at 19:40

About the second question:

The Schwartz space $\mathcal{S}(R)$ is a Frechet space, i.e. it's topology is given by a countable sequence of seminorms. Any Schwartz distribution (continuous linear functional on $\mathcal{S}(R)$) has to bounded by one of these seminorms. If the Schwartz distribution is of "order zero" then the only relevant seminorms are $\|f\|_n = \sup (1+|x|)^n |f(x)|$. If $T\in \mathcal{S}'(R)$ is bounded by $\|\ldots\|_n$, then this means $T$ is of the form $\langle T, f\rangle = \int (1+|x|)^{n} f(x) \mu(dx)$ for some finite measure on $R$.

So I would think measures which are Schwartz distributions are measures "of polynomial growth."

share|cite|improve this answer
Thanks Sigurd Angenent. I totally agree with you. It is also my guess. This statement can be made further for zero order distributions in $\mathcal{O}_c'$ (distributions with fast decrease, I am using Schwartz notations). Then the measure should have fast decrease (faster than any polynomials). Thanks a lot! :-) – Anand Jul 15 '11 at 9:29

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.