2

Consider all the ways to prove $\sin(2x)=2\sin(x)\cos(x)$. There are many!

I personally would use Euler's formula involving complex numbers due to it being extremely simple and straightforward. We know $\sin(x)=\frac{e^{ix}-e^{-ix}}{2i}$ and $\cos(x)=\frac{e^{ix}+e^{-ix}}{2}$. The LH and RH sides become $\frac{e^{i2x}-e^{-i2x}}{2i}=2(\frac{e^{ix}-e^{-ix}}{2i})(\frac{e^{ix}+e^{-ix}}{2})$. And now this proof can be done with some basic algebra!

At the same time, I have always been bothered by rewriting $\sin(x)$ and $\cos(x)$ this way. To rewrite everything in terms of $e$ feels like an utter avoidance to use the words "sine" and "cosine." It also seems like an avoidance to represent such formulas along the unit circle to prove them. The proofs involving the sum and difference formulas that use the unit circle have always appeared so clean to me.

This got me thinking... is there a dependable way to prove results like these using the unit circle? Or is there a way to rewrite everything in terms of sine and cosine another way?

I have been curious about this for a long time, and I would be curious as to what other dependable ways are used to prove trigonometric results such as these without $i$ and infinite sequences. What do other people do?

N. F. Taussig
  • 76,571
W. G.
  • 1,766
  • Many of these results can be proved using a unit circle, which is a nice method. However, I disagree with the idea that writing everything in terms of $e$ is a cop-out. Fundamentally, $\sin$ and $\cos$ are exponential functions. In fact, all elementary functions are. – Joe Feb 20 '21 at 19:52
  • @Joe All elementary functions? Even rational functions? – Allawonder Feb 25 '21 at 22:02
  • @Joe And the sine and the cosine are not exponential functions, but certain combinations of exponentials. – Allawonder Feb 25 '21 at 22:03
  • @Allawonder: Rational functions are quotients of polynomials, and polynomials are the sum of powers of $x$, and powers of $x$ are defined by the exponential and the logarithm. In my comment, when I said 'exponential function', I meant it in a looser sense, rather than a function $f$ of the form $f(x) = ab^x$. – Joe Feb 26 '21 at 18:39
  • @Joe Pardon me, but I don't see how $x^2,$ for example, is of the form $a^x,$ for some constant $a.$ They seem to be essentially different to me. – Allawonder Feb 26 '21 at 21:51
  • @Allawonder: It's an exponential function in the sense that $x^2 := \exp(2\log x)$. You're of course right that it's not the exponent that is variable. My point was simply that almost everything goes back to the exponential function and the logarithm, even if we have to take a circuitous root. – Joe Feb 26 '21 at 21:54
  • @Allawonder: I agree that I'm using the term exponential function in a non-standard way; I just wanted to emphasise that the function $\exp$ is at the heart of the elementary functions. – Joe Feb 26 '21 at 21:59
  • @Joe In that case then any function $f(x)$ can be trivially written as $e^{\log f(x)},$ elementary or otherwise. – Allawonder Feb 26 '21 at 23:56
  • @Allawonder: Sure, but the reason that $x^2$ is connected to the exponential is not because it can trivially be written in terms of it, but because $x^2$ is defined as $\exp(2\log x)$ for positive $x$. Also, consider how most of the elementary functions can be defined as solutions to differential equations. And the exponential function is at the heart of those solutions. – Joe Feb 27 '21 at 00:11
  • @Joe I don't think I follow. The functions $x^a,$ where $a$ is constant, can be defined however we want. But my point is that there are several (equivalent) definitions. What's more, we can choose to define any function $f(x)$ as $e^{\log f(x)}$ as well. Also, the solutions to differential equations can be functions of almost any kind -- all the special functions arise to our knowledge only because we study differential equations. Finally, exactly how do you define exponential function, because $\exp(2\log x)$ isn't exponential to me, but a composition of an exponential and a logarithm. – Allawonder Feb 27 '21 at 08:25
  • @Allawonder: I'm not making a precise statement about how $x^2$ is an exponential function. Rather, I'm saying that it's related to it because $x^2 = \exp(2\log x)$ (and this is the most common way of defining $x^y$). And the differential equations that the elementary functions are solutions to are all connected to the exponential function. If we try to solve $f''(x) = -f(x), f'(0) = 1, f(0)=0$, we get a solution written in terms of $e^x$. It's true that the solutions of differential equations can be functions of any kind. But the differential equations that the elementary functions are... – Joe Feb 27 '21 at 10:30
  • ...solutions to are all intimately connected to the exponential function. – Joe Feb 27 '21 at 10:30
  • @Joe I don't see that you have made this "intimate relationship" between the exponential function and the elementary functions explicit -- indeed, everything you've said so far applies to any function at all, elementary or not. – Allawonder Feb 27 '21 at 12:45

5 Answers5

3

A delightful application of linear algebra:

Let $R(\theta)$ be the linear map $\mathbb{R}^2 \rightarrow \mathbb{R}^2$ that rotates points counterclockwise an angle $\theta$ about the origin. It is well-known that $R(\theta)$ has the matrix representation:

$$\begin{bmatrix} \cos \theta & -\sin \theta \\ \sin \theta & \cos \theta \end{bmatrix}$$

It is obvious from inspection that $R(\theta + \psi) = R(\theta) \circ R(\psi)$. That is to say:

$$\begin{bmatrix} \cos \theta & -\sin \theta \\ \sin \theta & \cos \theta \end{bmatrix}\begin{bmatrix} \cos \psi & -\sin \psi \\ \sin \psi & \cos \psi \end{bmatrix} = \begin{bmatrix} \cos (\theta+\psi) & -\sin (\theta+\psi) \\ \sin (\theta+\psi) & \cos (\theta +\psi)\end{bmatrix}$$

Hence matrix multiplication gives you the angle addition identities:

$$\sin (\theta + \psi) = \sin\theta\cos\psi + \cos\theta\sin\psi$$

for example. Of course, identifying $a+bi \equiv \begin{bmatrix} a & -b \\ b & a \end{bmatrix}$ takes a little bit of the shock out of this, but I would argue that this is an elegant way of deriving trigonometric identities without relying on Euler's formula.

While I Am
  • 2,380
1

Partial Answer

Firstly,

The proofs involving the sum and difference formulas that use the unit circle have always appeared so clean to me.

$$\sin(2x)=\sin(x+x)=\sin(x)\cos(x)+\cos(x)\sin(x)=2\sin(x)\cos(x)$$

Most of the trigonometry formulas can be derived after establishing the basic ones (especially the sum and difference identities). In the unit circle, by rotating an angle (in the standard position) by the multiples of 90 degrees counterclockwise around the origin (or by using congruent triangles), you can prove the phase shift identities and much more. Sketching graphs often helps too. For example, $\sin(x-180°)=-\sin(x)$ can be seen by sketching $\sin(x-180°)$, noting that it is simply a sine curve with a horizontal shift of $180°$. Complex numbers are especially useful for double/triple angle formulae and power reduction etc.

krazy-8
  • 2,224
1

$\def\a{\alpha} \def\b{\beta} \def\t{\theta} \def\ar{\text{area}}$

Here's an unusual way to derive the addition formula for $\sin(\a+\b)$, based on areas of triangles. First, recall that the area of a triangle with two sides $a$ and $b$ and included angle $\t$ is $A=\frac12ab\sin\t$. The steps are outlined.

Consider the diagram below:

Addition law for sin

The points $D$ and $E$ are obtained by dropping perpendiculars from $B$ and $A$, respectively, onto $\overline{OC}$.

$\bullet$ Show that $\triangle ACE \sim \triangle BCD$.

$\bullet$ Show that $(AE)(CD) = (BD)(CE)$ and conclude that $\ar\triangle ACD = \ar\triangle BCE$.

$\bullet$ Explain why \begin{align*} \sin(\a+\b) &= 2\,\ar\triangle OAB = 2\big(\ar\triangle OAC+\ar\triangle OCB\big) \\ &= 2\big(\ar\triangle OAD + \ar\triangle OEB\big). \end{align*}

$\bullet$ Find $OD$ (in terms of $\b$) and $OE$ (in terms of $\a$), and deduce the formula for $\sin(\a+\b)$.

Ted Shifrin
  • 115,160
1

Let $R_{\theta}$ be a rotation matrix that rotates an arbitrary vector by an angle of $\theta$ radians anticlockwise about the origin: $$ R_{\theta} = \begin{bmatrix} \cos \theta & -\sin\theta \\ \sin \theta & \cos\theta \end{bmatrix} $$ Clearly, $R_{\theta}R_{\theta}=R_{2\theta}$, since two rotations of size $\theta$ has the same result as a rotation of size $2\theta$: $$ R_{2\theta}= \begin{bmatrix} \cos 2\theta & -\sin2\theta \\ \sin 2\theta & \cos2\theta \end{bmatrix} \, . $$ On the other hand, if we multiply out the matrices the long way, we get $$ R_{\theta}R_{\theta}= \begin{bmatrix} \cos \theta & -\sin\theta \\ \sin \theta & \cos\theta \end{bmatrix} \begin{bmatrix} \cos \theta & -\sin\theta \\ \sin \theta & \cos\theta \end{bmatrix} = \begin{bmatrix} \cos^2\theta-\sin^2\theta & -2\sin\theta\cos\theta \\ 2\sin\theta\cos\theta & \cos^2\theta - \sin^2\theta \end{bmatrix} \, , $$ and comparing the two gives us $\cos 2\theta \equiv \cos^2\theta - \sin^2\theta$ and $\sin 2\theta \equiv 2\sin\theta\cos\theta$.

Joe
  • 19,636
0

A rather twisted idea: $y = \sin 2x$ and $y = 2\sin x\cos x$ are solutions of the ODE $y'' + 4y = 0$ with initial conditions $y(0) = 0$, $y'(0) = 2$. Now, apply unicity...