3
$\begingroup$

This Wikipedia article listing trigonometric identities, states the following identity under half angles: $$ \sin{\frac{\theta}{2}} = \text{sgn}\bigg(2\pi-\theta+4\pi\bigg\lfloor\frac{\theta}{4\pi}\bigg\rfloor \bigg)\sqrt{\frac{1-\cos{\theta}}{2}} $$ where $\text{sgn}(x)$ is a function that returns $+1$ if $x>0$ and $-1$ if $x<0$.

I know and understand the part of the formula that says $\sqrt{\frac{1-\cos{\theta}}{2}}$ and understand the fact that the remaining part is to decide the $\pm$ sign in the formula. From where, however, does the sign-deciding part of the expression come from? And why should it make sense?

$\endgroup$
5
  • 1
    $\begingroup$ The sign thing is just a shorthand to express that the value is positive when $\theta \in [4k\pi, (4k+2)\pi]$ for some integer $k$, and negative otherwise. It's just a notation trick, not something derived from anywhere. $\endgroup$ – user3733558 Apr 4 at 8:29
  • $\begingroup$ @user3733558 Why $4\pi$, though? $\endgroup$ – soupless Apr 4 at 8:34
  • $\begingroup$ @user3733558 Won't any $\theta$ lie in that range? (put $k=0$ and the range becomes $[0, 2\pi]$) Edit: I understand what you are saying now. But I still do not understand how the notational trick incorporates the same. $\endgroup$ – Aadi Prasad Apr 4 at 8:39
  • $\begingroup$ Thanks @AadiPrasad, and a last note (because I feel the tone of my comments and answers don't reflect it): I want to applaud your curiosity, and your questioning of details like this one. That's a proper attitude towards anything mathematical. Well done. $\endgroup$ – user3733558 Apr 4 at 9:06
  • 1
    $\begingroup$ Thank you very much to you too, @user3733558. And no, your answer or your comments do not reflect the 'it is how it is' tone at all. Not everything has to have a very profound explanation. $\endgroup$ – Aadi Prasad Apr 4 at 10:26
5
$\begingroup$

I imagine it happened this way. In the past, someone saw $$ \sin \frac{\theta}{2} = \pm \sqrt{\frac{1-\cos{\theta}}{2}} $$ and thought "Wait a minute, this isn't really satisfactory. There shouldn't be uncertainties here. We know what sign the expression should be for a given $\theta$". And indeed, $$ \sin \frac{\theta}{2} \geqslant 0 \text{ for } 0 \leqslant \theta \leqslant 2\pi \\ \sin \frac{\theta}{2} \leqslant 0 \text{ for } 2\pi \leqslant \theta \leqslant 4\pi $$ and this pattern is periodic of period $4\pi$. So that someone sat down and thought "What function can I cook up that would give me $+1$ for inputs in $[0,2\pi]$ and $-1$ for inputs in $[2\pi,4\pi]$, and would also be $4\pi$ periodic?". And he found the expression you can see on Wikipedia.

This is not the only way to achieve the same goal. This is not a forced (derived) result. This is an artificial invention, made to achieve a specific goal, and it works. But don't look for a deeper meaning behind it, there isn't any.

$\endgroup$
2
$\begingroup$

Personally, I would go with $$ (-1)^{\left\lfloor\frac\theta{2\pi}\right\rfloor}=\left\{\begin{array}{rl}1&\text{when }\frac\theta{2\pi}\in[0,1)\pmod{2}\\-1&\text{when }\frac\theta{2\pi}\in[1,2)\pmod{2}\end{array}\right.\tag1 $$ However, what was done on Wikipedia was to create a sawtooth wave, which again, I would give as $$ \frac12-\left\{\frac\theta{4\pi}\right\}\left\{\begin{array}{}\gt0&\text{when }\frac\theta{4\pi}\in\left(0,\frac12\right)\pmod{1}\\\lt0&\text{when }\frac\theta{4\pi}\in\left(\frac12,1\right)\pmod{1}\end{array}\right.\tag2 $$ However, since $\{x\}=x-\lfloor x\rfloor$, $(2)$ can be written as $$ \frac12-\left\{\frac\theta{4\pi}\right\}=\frac12-\left(\frac\theta{4\pi}-\left\lfloor\frac\theta{4\pi}\right\rfloor\right)\tag3 $$

enter image description here

Note that all three functions have the same sign; negative in the gray regions and positive in the white regions.

The sign in $(3)$ is preserved upon multiplication by $4\pi$: $$ 2\pi-\theta+4\pi\left\lfloor\frac\theta{4\pi}\right\rfloor\tag5 $$ which has the same sign as $(2)$ and is what appears in Wikipedia.

$\endgroup$
5
  • $\begingroup$ Can you elaborate a bit more? How did you come up with $(1)$, or $(2)$? All of it just looks Greek to me. $\endgroup$ – Aadi Prasad Apr 5 at 3:20
  • $\begingroup$ Ok. We need to establish some context. Have you encountered the floor function, $\lfloor x\rfloor$? $\{x\}$ is the (periodic) fractional part function, defined as $x-\lfloor x\rfloor$. It goes from $0$ to just less than $1$ as $x$ goes from $n$ to just less than $n+1$ for $n\in\mathbb{Z}$. $\endgroup$ – robjohn Apr 5 at 4:25
  • $\begingroup$ I know all the definitions you mention. But why would you 'go with $(1)$'? From the way it is written, it seems as if equation $(1)$ should be as obvious and natural as $2+2 = 4$ is. It isn't so to me. (I am a high schooler who will be learning (single variable) calculus this year.) $\endgroup$ – Aadi Prasad Apr 5 at 10:37
  • $\begingroup$ Look at $\left\lfloor\frac\theta{2\pi}\right\rfloor$. It jumps up $1$ every $2\pi$. That means it goes from even ($0$) on $[0,2\pi)$, to odd ($1$) on $[2\pi,4\pi)$, to even ($2$) on $[4\pi,6\pi)$, and so on. Thus, $(-1)^\left\lfloor\frac\theta{2\pi}\right\rfloor$ goes from $1$ on $[0,2\pi)$ to $-1$ on $[2\pi,4\pi)$, to $1$ on $[4\pi,6\pi)$. This is the same pattern as the sign of $\sin\left(\frac\theta2\right)$ (see the graph). $\endgroup$ – robjohn Apr 5 at 12:54
  • $\begingroup$ Ooh! Now I get it — both, $(1)$ and $(2)$. Thanks a lot! $\endgroup$ – Aadi Prasad Apr 6 at 9:19
1
$\begingroup$

The function inside the $\operatorname{sgn}$ function has zeroes only when:

$$\theta=2\pi \pm 4k\pi$$

and the sign alternates between them, which gives the same pattern as the $\sin\frac{\theta}{2}$ function.

$\endgroup$
0
$\begingroup$

It is from the process of deriving the identity itself.

First, let us use the double-angle formula $\cos2\theta = 2\cos^{2}\theta - 1$. Solving for $\cos^{2}\theta$, we get $$\cos^{2}\theta = \frac{1 + \cos 2\theta}{2}.$$

Replacing $\theta$ by $\frac{\theta}{2}$, we get \begin{align*}\cos^{2}\left(\frac{\theta}{2}\right) &= \frac{1 + \cos\theta}{2} \\ \cos\left(\frac{\theta}{2}\right) &= \pm\sqrt{\frac{1 + \cos\theta}{2}}\end{align*}

Using the identity $\sin^{2}\theta + \cos^{2}\theta = 1$ to get $\sin \frac{\theta}{2}$, \begin{align*}\sin^{2}\left(\frac{\theta}{2}\right) + \cos^{2}\left(\frac{\theta}{2}\right) &= 1 \\ \sin^{2}\left(\frac{\theta}{2}\right) + \frac{1 + \cos\theta}{2} &= 1 \\ \sin^{2}\left(\frac{\theta}{2}\right) &= 1 - \frac{1 + \cos\theta}{2} \\ \sin^{2}\left(\frac{\theta}{2}\right) &= \frac{2 - 1 - \cos\theta}{2} \\ \sin^{2}\left(\frac{\theta}{2}\right) &= \frac{1 - \cos\theta}{2} \\ \sin\left(\frac{\theta}{2}\right) &= \pm\sqrt{\frac{1 - \cos\theta}{2}}\end{align*}

As you can see, the squares on the equation force it to take the plus-minus sign.


Edit: I don't have an idea about the $\displaystyle \mathrm{sgn}\left(2\pi - \theta + 4\pi\left\lfloor\frac{\theta}{4\pi}\right\rfloor\right)$. I just used the basic ones.

$\endgroup$
1
  • $\begingroup$ Thank you very much for writing out the entire derivation. But I am already aware of it — the issue I have is with what is inside the $\text{sgn}$ function. $\endgroup$ – Aadi Prasad Apr 4 at 8:32

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.