I have a rather simple problem which I can't make rigorous:

Let $\varphi \colon \mathbb{R}^2 \to \mathbb{R}$ be a continuous function. Then it holds that

$$ \lim_{m \to \infty} \, \, \,\lim_{n \to \infty} mn \int_{x}^{x+1/m} \int_y^{y+1/n} f(u,v) \, du dv = \lim_{m \to \infty} m \int_{x}^{x+1/m} f(u,y) \, du = f(x,y)$$.

Intuitively, it should also hold that

$$ \lim_{m \to \infty} m^2 \int_{x}^{x+1/m} \int_y^{y+1/m} f(u,v) \, du dv = f(x,y).$$

How can one prove this?

  • $\begingroup$ Note that in the limit, you're essentially convolving $f$ with the top-hat representation of the Dirac delta function. $\endgroup$ – John Barber May 1 '18 at 2:21

Use the mean value theorems for definite integrals, and you should have

$$ \int_x^{x+1/m} \int_y^{y+1/m} f(u, v) \,dudv = \frac{1}{m^2} f(\alpha_m, \beta_m),$$ where $\alpha_m \in [x, x+1/m], \,\beta_m \in [y, y+1/m]$.

The continuous property of $f(u, v)$ implies that

$$\lim_{m \to \infty} m^2 \int_x^{x+1/m} \int_y^{y+1/m} f(u, v) \,dudv = \lim_{m \to \infty} f(\alpha_m, \beta_m) = f(x, y).$$

References: The Mean Value Theorem for Double Integrals

  • $\begingroup$ Thanks, an argument like this was what I was looking for! $\endgroup$ – herrsimon May 1 '18 at 2:19
  • $\begingroup$ Why don't replace one $m$ with $n$? $\endgroup$ – Danny Pak-Keung Chan May 1 '18 at 2:32

No. However intuitive you find it, it is simply NOT true that you can combine limits like that. Take the following as a counterexample:

$$\lim_{m \rightarrow \infty}\lim_{n \rightarrow \infty}\biggl(1+\frac 1n\biggl)^m=\lim_{m \rightarrow \infty}1^m=1$$


$$\lim_{n \rightarrow \infty}\biggl(1+\frac 1n\biggl)^n=e$$

  • $\begingroup$ What you say is true in general but your "counterexample" doesn't apply to the situation here. $\endgroup$ – herrsimon May 1 '18 at 2:18

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.