I'm not sure if this question already exists, because I'm not sure how to call it. I want to integrate over the line of a square. In a simplified problem, I'd like to solve:

$$ l=\oint\limits_c dl $$

Therefore I figured I could take one line and multiply the result by 4:

$$ l=4\oint\limits_{c'}dl $$

And then I wanted to use an angle ($d\alpha$) to represent $dl$ - however, that's where I failed. In the following figure I illustrated which values I can easily use to get $dl$ ($R$ is known and $r=\frac{R}{\cos\alpha})$: illustration With the help of the sine theorem I got the following: $$\frac{dl}{\sin(d\alpha)}=\frac{r}{\sin(180^\circ-d\alpha-(90^\circ-\alpha))}=\frac{r}{\sin(90^\circ+\alpha-d\alpha)}=\frac{r}{\cos(d\alpha-\alpha)}$$ and with the addition theorem: $$dl=\frac{r\sin(d\alpha)}{\cos(d\alpha-\alpha)}=\frac{r\sin(d\alpha)}{\sin(d\alpha)\sin\alpha+\cos(d\alpha)\cos\alpha}\simeq\frac{rd\alpha}{d\alpha\sin\alpha+\cos\alpha}$$ But I have no idea how to calculate the resulting integral: $$l=4\oint\limits_{c'}\frac{rd\alpha}{d\alpha\sin\alpha+\cos\alpha}$$ as $d\alpha$ is in the denominator.

Am I missing some trigonometric rule here?

Edit if I use $r=\frac{R}{cos\alpha}$, I'd have: $$dl\simeq\frac{Rd\alpha}{\sin\alpha\cos\alpha d\alpha+\cos^2\alpha}$$ which (as wolframalpha tells me) is $$dl\simeq\frac{2Rd\alpha}{\sin(2\alpha)d\alpha+\cos(2\alpha)+1}$$ However, that doesn't help me much, as $d\alpha$ still is in the denominator.

More with the help of guesswork then through my own thoughts, I found a $dl$ that would work: $$dl=|2\sin(2\alpha)|d\alpha$$ For a square with $R=1$ for example, I'd have: $$l=4\oint\limits_{c'}|2\sin(2\alpha)|d\alpha=4\int\limits_{-\frac{\pi}{4}}^{\frac{\pi}{4}}|2\sin(2\alpha)|d\alpha=4\cdot2=8$$ so this behaves as expected - however, I have no idea how this could be trigonometrically (or mathematically, for that matter) explained. Furthermore, this seems to contrast my previous thoughts about $dl$, as illustrated in the following graph from geogebra (with $d\alpha=0.1$ - green is my first attempt, blue is my guessed attempt): graph

I also did some more research and got here, where the line element in polar coordinates is given by: $$dl^2=dr^2+r^2d\alpha^2$$ this seems to use the pythagorean theorem, in the sense that $d\alpha$ is straigt for small values, which makes sense. Follwing that, one might say: $$\tan(d\alpha)=\frac{dr}{d\alpha}$$ implying the following for the line element: $$dl=\sqrt{d\alpha^2\tan^2(d\alpha)+d\alpha^2}=d\alpha\sqrt{\tan^2(d\alpha)+1}$$ however, I don't know how to use this, as $d\alpha$ also is in the Tangens, is squared, is a summand and is taken the root of. I guess there's something wrong with this solution, as the term $\alpha$ doesn't even appear in it, but even if it wasn't I could not use it any more than my first solution.

I have the feeling that I'm looking at this problem in a way to complicated manner, especially as I can't seem to find useful information in the internet. But if this is true, what is the obvious answer here?


$$l=R\tan\alpha\implies dl=\frac{R\,d\alpha}{\cos^2\alpha}.$$

  • $\begingroup$ Ok that actually makes sense. Could you nevertheless elaborate on that? $\endgroup$ – MetaColon Jun 26 at 4:29
  • $\begingroup$ @MetaColon What kind of elaboration do you mean? I thought your only problem was to express $dl$ via $d\alpha$... Of course the expression is valid only for one side of the square ($-\pi/4<\alpha<\pi/4$) but it is straightforward to write down similar expressions for 3 other sides as well. $\endgroup$ – user Jun 26 at 9:14
  • $\begingroup$ In my understanding it is a rather unusual way to express the variable to integrate over ($dl$) as a differential and hence applying the rules for differentiating on it. At least it's a method I've never seen in practice - maybe you could elaborate on why exactly this works? And also how it fits to my attempts, if it does at all. $\endgroup$ – MetaColon Jun 26 at 11:00
  • $\begingroup$ This is the usual way. In one-dimensional case one introduces a new variable $t$, expresses it as function of the old one $t=f(x)$, and finds the differential as $dt=f'(x)dx$. Quite oppositely I do not understand the background of your attempts. Please give a reference to description of this procedure and - if possible - to its successful application. $\endgroup$ – user Jun 26 at 11:24
  • $\begingroup$ To be honest I'm a physics student, hence we learned a kind of intuitive way to understanding things like $dl$. This answer for example shows how we explained volume elements in spherical coordinates. I tried to apply similar reasoning to get the line element in polar coordinates. $\endgroup$ – MetaColon Jun 26 at 14:05

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.