-2
$\begingroup$

I'm stuck on a passage I do not understand, which reads:

$$\int_{r<|y|<1} \bigg| \frac{1}{(|y|^2 - r^2)^s |y|^n} - \frac{1}{|y|^{n+2s}}\bigg|\ \text{d}y$$ $$\int_1^r \bigg| \frac{1}{(t^2 - r^2)^s} - \frac{1}{t^{2s}}\bigg|\ \frac{\text{d}t}{t}$$ $$r^{s}\int_1^{1/r} \bigg| \frac{1}{(\tau^2 - 1)^s} - \frac{1}{\tau^{2s}}\bigg|\ \frac{\text{d}\tau}{\tau}$$

I understood there is a sort of rescaling between the second and the third line, but I cannot find a way to get the passage between the first and the second line. I thought about a change of variable like $|y|^n = t$ but there is something that I miss, and it doesn't work.

This is an integral in $\mathbb{R}^n$, where $0<s<1$.

$\endgroup$

1 Answer 1

2
$\begingroup$

Note that the integrand only depends on the absolute value of $y$, and not on the direction of the vector. That means it is a good idea to substitute $t = |y|$. Let's denote $$f(t) = \bigg| \frac{1}{(t^2 - r^2)^s t^n} - \frac{1}{t^{n+2s}}\bigg|$$ for positive real $t$. Denote $t\mathbb S^n$ to be the sphere of radius $t$. Then we have $$\int_{r<|y|<1}f(|y|) dy = \int_r^1\int_{t\mathbb S^n}f(t)dudt = \int_r^1\text{Vol}(t\mathbb S^n)f(t)dt = \text{Vol}(\mathbb S^n)\int_r^1f(t)t^{n-1}dt.$$ This reduces to almost the same formula that you want. It seems to me that the volume of the unit sphere has been neglected in your text, and the integration goes from $1$ to $r$ which is strange as $r<1$.

For the third line you can substitute $\tau = \frac tr$, but it seems you already knew how to do that.

$\endgroup$
3
  • $\begingroup$ I must be doing the substitution wrong. After setting $\tau = t/r$, don’t you pick up a factor of $r^{-2s}$ in front, not $r^s$? The text also seems to reverse the bounds of integration between line 2 and 3. $\endgroup$
    – Vik78
    Commented Dec 14, 2021 at 10:21
  • $\begingroup$ Hmm, you are right. This is another mistake in the text I think. The bounds of integration are incorrect in line 2 and correct again in line 3. $\endgroup$
    – Squala
    Commented Dec 14, 2021 at 10:24
  • $\begingroup$ This is very helpful thank you. Also my mistake: there were indeed a factor $r^{2s}$ in front of the integral. Also the volume of the sphere has been taken out from the integral as a constant $c(n)$, which I forgot to write! $\endgroup$
    – Heidegger
    Commented Dec 14, 2021 at 12:12

Not the answer you're looking for? Browse other questions tagged .