Take the 2-minute tour ×
Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. It's 100% free, no registration required.

Let $x\in R^{n}$ be fixed ($x\neq 0$) and $r>0$. Evaluate the limit:$$\lim_{r\rightarrow \infty}\frac{V(B(x,r)\cap B(0,r))}{V(B(0,r))}$$

where $V$ stands for volume and $B(x,r)$ is the ball with center x and radius $r$.

Thanks in advance.

share|improve this question
    
Do you consider euclidian norm? –  Davide Giraudo Sep 27 '12 at 15:09

4 Answers 4

up vote 7 down vote accepted

The volume of $B(x,r)$ is $ar^n$ for some constant $a$ which dependends on $n$, but whose exact value will not concern us.

Let $d=||x||$ be the distance between the two centers $0$ and $x$. If $r>d$ then the triangular inequality implies that $$B_{r-d}(0)\subset B(x,r)\cap B(0,r)$$ Hence $${V(B(x,r)\cap B(0,r))\over V(B(0,r))}>{V(B_{r-d}(0))\over V(B(0,r))}={a(r-d)^n \over ar^n}=\Bigl(1-{d\over r}\Bigr)^n$$ When $r\to\infty$ then the last expression will $\to 1$.

share|improve this answer
1  
Don't you need also an upper bound? –  enzotib Sep 27 '12 at 15:39
    
Yes, but the ratio between the volumes is obviously $<1$, since one set is contained in the other. –  Per Manne Sep 27 '12 at 15:48
    
You're right, thanks. –  enzotib Sep 27 '12 at 15:50
    
I really liked your solution Per Manne. Thanks. –  Tomás Sep 27 '12 at 16:00

Yes. Use zooming $x\mapsto \displaystyle\frac1r\cdot x$. Then volume is shrinked by $\displaystyle\frac1{r^n}$, both in the enumenator and dominator, so the quotient is fix, $0$ stays, $r$ is constantly 1, and $x$ goes to $x/r \to 0$.

share|improve this answer

Yes if we work with euclidian norm. Let $I(r):=V(B(x,r)\cap B(0,r))$. Then $$\left|\frac{I(r)}{|B(0,r)|}-1\right|=\frac 1{|B(0,r)|}\int_{\Bbb R^n}\chi_{\{||x-y|>r\}}\chi_{\{y\leq |r\}}dy.$$ In the set on which we integrate, we have $|x|²-2\langle x,y\rangle+|y|^2\geq r^2\geq |y|^2$ hence $2\langle x,y\rangle\leq |x|^2$. This is a bounded set, hence the integral is uniformly bounded by an universal constant (independent of $r$). By the way, it gives a speed of convergence.

share|improve this answer

All Euclidean balls have the same volume, so we may as well use $V(B(\frac{x}{2},r))$ in the denominator. Now draw a picture to see that:

$$V\bigg(B\bigg(\frac{x}{2},r-\frac{|x|}{2}\bigg)\bigg)\leq V(B(x,r)\cap B(0,r))\leq V\bigg(B\bigg(\frac{x}{2},\sqrt{r^2 - \frac{|x|^2}{4}}\bigg)\bigg).$$

(enzotib points out below that it's easier to use a second radius of $r + |x|/2 > \sqrt{r^2 - |x|^2/4}$.)

Since the volume of a ball of radius $r$ is $Cr^n$, $|x|$ is fixed, and the radii of both the first and last balls are order $1$ in $r$, their volumes go as $Cr^n$. So when we divide through by $Cr^n$, we have that both the first and last expression go to $1$, forcing the middle to go to $1$ as desired.

share|improve this answer
    
The including ball can also have, more simply, $R=r+|x|/2$? –  enzotib Sep 27 '12 at 15:42
    
Oh yes, that's simpler. I drew a picture of two intersecting circles, with centers $0$ and $x$. The midpoint between the centers is $x/2$, and the two radii come from the largest circle centered at $x/2$ contained in the intersection and the smallest circle centered at $x/2$ which contains the intersection. This latter has radius $\sqrt{r^2 - |x|^2/4}$. –  Neal Sep 27 '12 at 15:47
    
My estimate comes from thinking of the union instead of the intersection, my fault. –  enzotib Sep 27 '12 at 15:51
    
Not a problem! The union is bigger than the intersection and $r + |x|/2$ still grows with order $1$ in $r$. :) –  Neal Sep 27 '12 at 15:53

Your Answer

 
discard

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.