Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. 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 behavior of Dirac deltafunctions within multivariate integrals. Here is a simple example to which I do not know the answer:

$$\iint\limits_{[0,1]\times [0,1]} \delta\left(x - y\right)\, dA$$

This integrates the constant function $1$ over the line $y=x$. Is there a prefactor involed though due to the delta? I can imagine this evaluating either to $1$ or to $\sqrt2$.

The information from the relevant Wikipedia page suggests the value $\sqrt2$ while WolframAlpha suggests the value $1$.

How is the behavior of a Dirac deltafunction within a multivariate integral defined? My goal is to create an algorithm to solve a broad class of these problems in general.

share|cite|improve this question
I'm not totally sure this is absolutely correct, but I think the iterated integral you have listed above has value $1$. I believe the other answer, $\sqrt{2}$, maybe the value of $\iint\limits_{[0,1]\times[0,1]} \delta(x-y)\,dA$. There's no reason to believe Fubini's theorem should apply here. – Bill Cook Dec 22 '11 at 16:31
I'm particularly interested in the single integral over a region, not an iterated integral. I'll change the question. – MRocklin Dec 22 '11 at 16:51
@Bill: If by that integral you mean what the Wikipedia article defines it to mean, then the factor $1/|\nabla g|=1/\sqrt2$ makes this come out as $1$, too. (Sorry about the multitude of comments I wrote and deleted -- I was trying to figure out what you meant by that integral.) Generally speaking, we should be able to treat an integral with a one-dimensional delta as a limiting case with a single parameter going to $0$, so there's no question of interchange of limits and the result shouldn't depend on technicalities like the order of integration. – joriki Dec 22 '11 at 17:28

In this case, where the coordinates appear linearly in the argument of the delta function, no special treatment is required; you can just evaluate the inner integral as you'd normally evaluate an integral over a delta function, and then evaluate the outer integral:

$$\int_{0}^{1}\int_{0}^{1} \delta\left(x - y\right)\, \mathrm dx\, \mathrm dy=\int_0^11\,\mathrm dy=1\;.$$

In case you got the value $\sqrt2$ from the length of the diagonal, you didn't take into account the factor $1/|\nabla g|=1/\sqrt2$.

share|cite|improve this answer
Where does the $1/ \sqrt 2$ appear in your solution? What is the missing intermediate step? – MRocklin Dec 22 '11 at 16:48
@MRocklin: There's no need for that factor in my solution. That last sentence wasn't related to my solution; it was just a speculation on what might have gone wrong when you tried to follow the Wikipedia article and got $\sqrt2$. If it doesn't make sense, you can ignore it. – joriki Dec 22 '11 at 17:04
Ah, yes, in contrast to my (deleted after @joriki's correction) earlier comment, reducing to the 1-D case by rotating gives $\delta(\sqrt{2}x)$ along the diagonal of length $\sqrt{2}$, so the $\sqrt{2}$'s go away. – paul garrett Dec 22 '11 at 17:20

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.