Define $\sin x $ and $\cos x$ via their infinite series: $$ \sin x = \sum_n (-1)^{n}\frac{x^{2n+1}}{(2n+1)!}, \qquad \cos x = \sum_n (-1)^n \frac{x^{2n}}{(2n)!}. $$ Is there a short, clever proof that $\cos(x+y) = \cos x \cos y - \sin x \sin y$ for all real $x,y$? I can prove it using product series, or by showing that both sides (with $y$ fixed) are solutions of $f''(x) = -f(x)$, $f(0) = \cos y$, $f'(0) = - \sin y$. Does anyone know other (preferably slick!) proofs?
-
There's probably a proof involving complex numbers. – Joe Z. Apr 02 '13 at 20:12
-
Not exactly what you want, but still, you might be interested in the geometric proof. – dtldarek Apr 02 '13 at 20:24
-
Allow me to clarify. I'm wondering if there is a proof using real analytic methods (e.g. power series). I'm not assuming anything about $\sin$ and $\cos$ other than what can be derived from their definitions as series. – Umberto P. Apr 02 '13 at 20:25
-
2I happen to have answered a very similar question today. You will find there in particular a proof of the two Addition Laws, rather condensed, using the series definition. – André Nicolas Apr 02 '13 at 20:46
-
ProofWiki features a geometric proof. As a PW herald, I'd like to invite any users to add their proofs (as far as they're different from the present quadruple) to it as well! – Lord_Farin Apr 02 '13 at 21:39
-
@AndréNicolas: Beautiful! Exactly what I wanted to see. I'm not sure the proofs get simpler than that. – Umberto P. Apr 03 '13 at 01:53
-
The diagram in this answer links the series definition with the unit circle definition, and then this diagram illustrates the identity. – Blue May 07 '13 at 11:36
9 Answers
New answer to an old question. This one, maybe the slickiest of them all, is due to Erhard Schmidt.
Define
$$f(t)=\cos(x+y-t)\cos(t)-\sin(x+y-t)\sin(t).$$
Verify $f'(t)=0$, hence $f$ is constant. Now the desired angle sum identity follows from $f(0)=f(y)$

- 18,103
- 3
- 32
- 49
One way is to use the fact that $$\cos(\theta) = \dfrac{e^{i \theta}+e^{-i \theta}}2$$ $$\cos(x+y) = \dfrac{e^{i(x+y)}+e^{-i(x+y)}}2 = \left(\dfrac{e^{ix}+e^{-ix}}2 \right) \left(\dfrac{e^{iy}+e^{-iy}}2\right) - \left(\dfrac{e^{ix}-e^{-ix}}{2i} \right) \left(\dfrac{e^{iy}-e^{-iy}}{2i}\right)$$
-
Can $e^{\imath x}$ be defined without series? The proofs that $e^{\imath(x+y)} = e^{\imath x}e^{\imath y}$ seem to involve product series or uniqueness of ODE, which I'm seeing if I can avoid. – Umberto P. Apr 02 '13 at 20:20
-
I don't believe it can. Are you looking for, say, an elementary geometry proof? – Joe Z. Apr 02 '13 at 20:21
-
look at Mertens' theorem http://en.wikipedia.org/wiki/Cauchy_product and you need only case when both series are absolutely convergent, which should be easier to prove. If I can remember right you will need that absolutely convergent series can be summed in any order. – tom Apr 02 '13 at 21:52
-
@tom: I know the product series proof as stated in the question. I'm looking for novel proofs. – Umberto P. Apr 03 '13 at 01:49
This was addressed in the question already. I leave it so that the method is fully explained.
LEMMA Let $f$ be a function with second derivative everywhere such that $f''+f=0$ and $f'(0)=0$; $f(0)=0$. Then $f$ is identically zero everywhere.
P We have that $$f''+f=0$$ Then $$f'f''+ff'=0$$ or $$(f')^2+f^2=C$$
But the initial conditions force $f'^2+f^2=0$ everywhere, which means $f\equiv 0$. $\blacktriangle$.
PROP Let $f$ be a function with second derivative everywhere such that $f''+f=0$, and $f'(0)=a$, $f(0)=b$. Then $$f=a\sin+b\cos $$
P Let $g=f-a\sin+b\cos$. Then $g''+g=0$ and $g'(0)=0$, $g(0)=0$. The lemma implies $g\equiv 0$, so that $f=a\sin+b\cos$. $\blacktriangle$.
Differentiate with respect to one variable and use the uniqueness of the solution of a second degree ODE with initial conditions.
That is, your cosine on the left vetifies $$f''+f=0$$ and $f'(0)=–\sin y$, $f''(0)=\cos y$. Then it must coincide with the unique solution $$f'(0) \sin+f(0)\cos$$

- 122,002
I always had a hard time to memorize that formula. But actually, that's not really needed, because there it is an easy way to reconstruct it from the from the laws of exponentiation applied to complex exponentiation: $$e^{i(x + y)} = e^{ix} \cdot e^{iy}.$$
Using the complex multiplication rule $\operatorname{Re}(ab) = \operatorname{Re}(a)\operatorname{Re}(b) - \operatorname{Im}(a)\operatorname{Im}(b)$, taking the real part gives
$$\operatorname{Re}(e^{i(x + y)}) = \operatorname{Re}(e^{ix})\operatorname{Re}(e^{iy}) - \operatorname{Im}(e^{ix})\operatorname{Im}(e^{iy}).$$
So by $\cos(x) = \operatorname{Re}(e^{ix})$ and $\sin(x) = \operatorname{Im}(e^{ix})$ $$\cos(x + y) = \cos(x)\cos(y) - \sin(x)\sin(y).$$

- 22,696
The way I learned it as a kid was geometric, and probably looked like the proof seen here on Wikipedia.
The segment $OP$ has length $1$. We have the $\sin(\alpha + \beta) = PB = PR + RB = \cos(\alpha) \sin(\beta) + \sin(\alpha) \cos(\beta)$.
Then, to prove the cosine identity we can use that $\cos(\alpha + \beta) = \sin(\alpha + \beta + \pi/2)$ and use the sine identity.

- 22,445
- 3
- 51
- 82
Let $\vec{u},\vec{v}\in\mathbb{R}^2$ unitary vectors such that $$ \vec{u}=\big(\cos(x),\sin(x)\big)\quad \mbox{ and } \quad \vec{v}=\big(\cos(-y),\sin(-y)\big) $$ Here $x$ and $-y$ are the smallest angle formed between the x-axis and the vectors $\vec{u}$ and $\vec{v}$ respectively. Then \begin{align} \cos\big( x+y\big) = & \cos\big( x-(-y)\big)\\ = & \frac{\vec{u}\bullet\vec{v}}{\|\vec{u}\|\cdot \|\vec{v}\|} \\ = & \vec{u}\bullet\vec{v}\\ = & \cos(x)\cdot\cos(-y)+\sin(x)\cdot\sin(-y)\\ = & \cos(x)\cdot\cos(y)-\sin(x)\cdot\sin(y)\\ \end{align}

- 14,658
$$\begin{array} {rcl} \cos(x + y) + i \sin(x + y)& = & e^{i(x + y)} \\ &=& e^{ix}e^{iy} \\ &=& (\cos(x) + i\sin(x))(\cos(y) + i\sin(y)) \\ &=& (\cos(x)\cos(y) - \sin(x)\sin(y)) + i(\sin(x)\cos(y) + \sin(y)\cos(x)) \\ \end{array}$$
Equating real and imaginary parts you get
$$\cos(x + y) = \cos(x)\cos(y) - \sin(x)\sin(y)$$ $$\sin(x + y) = \sin(x)\cos(y) + \sin(y)\cos(x)$$

- 23,556
-
Thanks for the answer - this question is nearly 6 years old. At the time I was interested in proving the trig angle-sum formulas using nothing but the series definition of sine and cosine. The issue is that the identity $\cos x + i \sin x = e^{ix}$ is outside the scope of the question, and the justification of the further identity $e^{w+z} = e^w e^z$ is requires tools that essentially prove what was being asked in the first place. – Umberto P. Mar 13 '19 at 18:59
For variety, here is a different proof. Unfortunately it might be considered circular since it relies on differentiation of trig functions. But maybe you know their derivatives without using this identity, if say $\sin$ and $\cos$ have been defined by their Taylor series.
Apply $\frac{d^2}{dx^2}$ to each side (viewing $y$ as some constant), and you see that each side a solution to $\frac{d^2}{dx^2}f(x)=-f(x)$
Both sides are in agreement at $x=-y$. Also the first derivatives of each side are in agreement at $x=-y$. Therefore they are the same expression.

- 54,717
Suppose we know that $\sin(x+y) = \sin x \cos y + \cos x \sin y$, then we can do this to get a proof:
$\cos (x + y) \\ = \sin (x + y + \frac{\pi}{2}) \\ = \sin x \cos (y + \frac{\pi}{2}) + \cos x \sin (y + \frac{\pi}{2}) \\ = \sin x (-\sin y) + \cos x \cos y \\ = \cos x \cos y - \sin x \sin y$
By the way, both have a geometric proof.

- 151