0
$\begingroup$

I was looking through this paper and found forward and backward transform equations for calculating radial distortion caused by fisheye lense abberation using the equations $f = \text{focal length}, r_u = \text{undistorted radius}, r_d = \text{distorted radius}$:

$r_d = f \arctan(\frac {r_u}{f})$

$r_u = f \tan(\frac {r_d}{f})$

I was previously confused by non polynomial representations of the distortion through there lack of explanation of how to plugin variables to the equation and get out a non distorted image (as if no abberation occured) like this one: in their words not mine...

$s =$ "a scalar"

$\lambda = $"controls the amount of distortion across the image"

$r_d = s \ln(1 + \lambda(r_u))$

$r_u = (e^\frac{r_d}s - 1)/\lambda$

I couldn't figure out what constants of lambda and s would get no distortion (this was important for programmic implementation and testing). Originally I didn't see how this was possible with the previous equation until I input higher and higher values of $f$, and I found out that the limit of $\tan(n/x)$ as $x$ approaches infinity always equal n, so you when $f = \infty, r_u = r_d$. (note the later equation set did not seem to have any consistency with inputs, so what ever makes $r_u = r_d$ isn't $0,1,\infty,$ or negation of any of those values)

I tried looking this up, and I was able to find this, however the logic used here cannot be applied for any numerator in tangent except 1, since $\sec^2(1/\infty) = \sec^2(0) = 1$ implies that $\sec^2(2/\infty)$ also $= 1$.

What are the steps to prove this/how would I even start formulating how this is true in my head?

EDIT: meant $x \tan(n/x)$, not $\tan(n/x)$, though you really shouldn't just be looking at the title and calling it a day :)

$\endgroup$
  • 2
    $\begingroup$ Your question doesn't match the Alpha input ! $\endgroup$ – Yves Daoust Mar 30 '17 at 21:30
  • $\begingroup$ Actually, $$\lim_{x\to\infty}\tan\left(\frac{n}{x}\right) = 0$$ for all values of $n$. $\endgroup$ – 5xum Mar 30 '17 at 21:31
  • $\begingroup$ I'm pretty sure that $$n=\lim_{x\to\infty}x\tan(n/x)\ne\lim_{x\to\infty}\tan(n/x)=0$$ If you meant the left and not the right, please clear this up for us, and while your at it, please learn how to format math here. $\endgroup$ – Simply Beautiful Art Mar 30 '17 at 21:32
  • $\begingroup$ @SimplyBeautifulArt Oops I meant the left one. Edited the title $\endgroup$ – opa Mar 30 '17 at 23:46
  • $\begingroup$ :-) Thanks for taking the time to fix that up and attempting to fill in the MathJax. I tried to fix what I could, but if anything contradicted your intentions, do fix. $\endgroup$ – Simply Beautiful Art Mar 31 '17 at 0:40
0
$\begingroup$

$$\lim_{x\to\infty}\tan\frac nx=0$$ because $$\lim_{x\to\infty}\frac nx=0.$$


The true question is

$$\lim_{x\to\infty}x\tan\frac nx=n$$ because $$\lim_{x\to\infty}x\frac nx=n,$$ using the Taylor development to the first order.

$\endgroup$
  • $\begingroup$ Ok I meant the second one, but your response still doesn't explain much, nor does it explain the inconsistency with the derivation for x tan(1/x), I really can't give this anything... $\endgroup$ – opa Mar 30 '17 at 23:45
  • $\begingroup$ @snb: which "inconsistency with the derivation for $x\tan(1/x)$" ??? $\endgroup$ – Yves Daoust Mar 31 '17 at 6:43
  • $\begingroup$ socratic.org/questions/… $\endgroup$ – opa Mar 31 '17 at 12:35
  • $\begingroup$ @snb: which inconsistency ? $\endgroup$ – Yves Daoust Mar 31 '17 at 13:57
0
$\begingroup$

Straight limit with L'Hopital: $$\lim_{x\to\infty}x\tan\Bigl(\frac{n}{x}\Bigr)=\lim_{x\to\infty}\dfrac{\tan\bigl(\frac{n}{x}\bigr)}{\frac{1}{x}}\to\frac{0}{0}\xrightarrow{\text{L'Hopital}}\lim_{x\to\infty}\dfrac{\tan\bigl(\frac{n}{x}\bigr)}{\frac{1}{x}}=\lim_{x\to\infty}\dfrac{n\sec^2\bigl(\frac{n}{x}\bigr)\cdot\left(-\frac{1}{x^2}\right)}{-\frac{1}{x^2}}=\lim_{x\to\infty}\dfrac{n}{\cos^2\bigl(\frac{n}{x}\bigr)}=n$$ As others have said, $\lim\limits_{x\to\infty}\tan\Bigl(\frac{n}{x}\Bigr)=0$

$\endgroup$
0
$\begingroup$

The derivative of $\tan(z)$ at $z=0$ is $1$. That is what your fact says: $$ 1 = \tan'(0) = \lim_{h \to 0}\frac{\tan(h)-\tan(0)}{h} $$ so for fixed $n$, let $h=n/x$, so that $h \to 0$ as $x \to \infty$. Thus $$ 1 = \lim_{x \to \infty} \frac{\tan(n/x) - 0}{n/x} = \lim_{x \to \infty} \frac{x}{n}\tan\frac{n}{x} = \frac{1}{n} \lim_{x \to \infty} x\tan\frac{n}{x} $$ equivalent to your result.

$\endgroup$
  • $\begingroup$ This makes a lot of sense, however, how does this fit into this post? what I've seen with $x tan(n/x)$ seems to contradict it, I'm also confused as to how the third step into the fourth step actually happens, doesn't seem to follow any trig identities. $\endgroup$ – opa Mar 31 '17 at 2:46
  • $\begingroup$ To clarify my above comment, when I talk about third and fourth steps, I'm not talking about your post, I'm talking about my link (yours makes 100% sense) $\endgroup$ – opa Mar 31 '17 at 2:55
0
$\begingroup$

Hint Set $h=\frac{n}{x}\to 0 $ as $x\to \infty.$ and use. $$\lim_{h\to 0}\frac{\tan h}{h} =1$$

$\endgroup$

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.