15

Why can complex numbers be written in exponential form? $z=r(\cos \theta+i\sin \theta)$ is $z=re^{i\theta}$.

I have studied that the exponential form of a complex number $z=r(\cos \theta+i\sin \theta)$ is $z=re^{i\theta}$.

Can someone explain why?

Guy Fsone
  • 23,903
Anne
  • 2,931
  • @A---B: I have reverted the edit which changed the title of the question to “how to change from polar to exponential form…” — it does not seem to be what the question is asking, so changes the meaning of the question. – ShreevatsaR Sep 19 '17 at 19:32
  • @ShreevatsaR Check now. –  Sep 19 '17 at 19:33
  • 3
    @A---B I am not sure “Why complex numbers are written in expoenential form?” is fine either (besides introducing a typo), because the question asks “I have studied that … can be written… can somebody explain why?”, so it may be asking about why complex numbers can be written in exponential form, rather than why they are. – ShreevatsaR Sep 19 '17 at 19:37
  • @ShreevatsaR Yes I think you are correct. –  Sep 19 '17 at 19:44
  • There is deeper justification for the equivalence of polar form and exponential form of a complex number (and this is beyond algebra). For an introduction to complex numbers this equivalence can be thought of a mnemonic to help remember rule for multiplication of complex numbers: multiply moduli and add argument. – Paramanand Singh Sep 20 '17 at 04:28
  • https://math.stackexchange.com/questions/3510/how-to-prove-eulers-formula-eit-cos-t-i-sin-t – Guy Fsone Nov 18 '17 at 18:44

6 Answers6

16

Here's a rather elegant proof.

The function $f : t\mapsto \cos t+i\sin t$ is differentiable and satisfies \begin{align*} f'(t) &= i\,f(t) \\ f(0) &= 1 \end{align*} Now let's solve it.

We have $f(0) = 1$ and

$$f'(t) = (\cos t+i\sin t)' = -\sin t+ i\cos t = i(\cos t+i\sin t) =if(t) $$ Now let us solve this differential equation $$f'(t) = if(t)\Longleftrightarrow e^{-it}f'(t) -ie^{-it} f(t)=0 \Longleftrightarrow \frac{d}{dt}\left(e^{-it} f(t)\right) = 0$$

That is $$e^{-it} f(t) = c\Longleftrightarrow f(t) = ce^{it}$$

But $f(0)=1 $ i.e $c=1$. Hence $f(t)=e^{it}$.

Guy Fsone
  • 23,903
  • +1 ugh, came here just to post a similar thing, but you beat me :) this is the most satisfying answer IMO. – user541686 Sep 20 '17 at 05:52
  • 4
    It's not clear what you mean by "$\ln(f(t))$". Someone wondering why $e^{it} = \cos t + i \sin t$ probably doesn't know what complex logarithm is; even worse, they may not even know that they don't know what it is! (And thus miss the fact that there's something nontrivial to say.) – Najib Idrissi Sep 20 '17 at 11:38
  • 6
    I don't understand what you are trying to prove here, and what is your definition of $e^{it}$ while you do it. – Federico Poloni Sep 20 '17 at 12:45
  • 1
    It is a little easier to show that $t \mapsto e^{-it} f(t)$ is a constant. Then the proof becomes a one liner. – copper.hat Sep 20 '17 at 14:07
  • @Mehrdad Thank you for editing this but I think we should be carefull to use logarithm together with complex. so I now propose an easier Proof. – Guy Fsone Sep 20 '17 at 15:50
  • @FedericoPoloni Maybe you check that edit now. Then ask further question if you still don't understand. – Guy Fsone Sep 20 '17 at 16:10
  • It would be better if you could supplement your answer with a specific definition of $e^{it} $. – Paramanand Singh Sep 20 '17 at 16:10
  • @NajibIdrissi Please check the first edited proof it does not use complex logarithm. I arised this issues myself – Guy Fsone Sep 20 '17 at 16:12
  • @ParamanandSingh You are right this depend on with definition we are of $e^{it}$. Here Our definition of $e^z$ is just the analytical extension of the real exponential $e^x$. But if you want to use the definition $e^z=\sum_{n} \frac{z^n}{n!}$ the to solve we will need to use Picard iteration which is far a bite beyond the scope of this question – Guy Fsone Sep 20 '17 at 16:16
  • @NajibIdrissi, GuyFsone: You can just assume $i$ is an arbitrary constant (a real-valued one if you so desire) and then at the end just naturally extend the same definition to complex values by assuming $i = \sqrt{-1}$. It's quite intuitive that this is the only definition that would make sense, even if you've never heard the term "analytical extension" before. That's kind of what I was going for. – user541686 Sep 20 '17 at 17:44
  • 2
    When you use the power series definition of $e^z$ the result in question follows trivially from Taylor series for sine and cosine. So that part is easy. On the other hand how do you propose to analytically extend the definition of $e^{x} $? One of the ways of doing this is via power series. Other is to define $e^{x+iy} =e^{x} (\cos y+i\sin y) $ which again makes the question trivial. From your solution it appears that you assume that $g(t)=e^{it} $ satisfies the differential equation (in the beginning of your post). This makes it trivial as one can show that the equation has unique solution. – Paramanand Singh Sep 21 '17 at 05:25
  • Thus your approach needs to be supplemented with the following: a proper definition of $e^{it} $ with $t$ being real. And using this definition prove that $(e^{it}) '=ie^{it} $ and further $e^{-it} e^{it} =1$. – Paramanand Singh Sep 21 '17 at 05:28
  • 4
    Continuing Paramanand Singh's comment: if we don't know what $e^{it}$ is, how do we know what its derivative is? That is, how do we know that it satisfies $\frac{\mathrm{d}}{\mathrm{d}t}f(t)=i,f(t)$? – robjohn Sep 21 '17 at 05:43
  • @ParamanandSingh the writing $$e^{x+iy} = e^x(\cos y+i\sin y)$$ is what we means by analytics extension. and this extension carries also the properties of the extended function. Including the equation $y'=a y$ – Guy Fsone Sep 21 '17 at 07:30
  • @GuyFsone: if by analytic extension you mean the definition $e^{x+iy} =e^{x} (\cos y+i\sin y) $ then by putting $x=0$ we are done. – Paramanand Singh Sep 21 '17 at 07:32
  • One approach which I can suggest is to use the power series definition of $e^{z} $ and using it show that $(e^{it}) '=ie^{it} $ and also $e^{it} e^{-it} =1$ (these properties are very easily proved using power series) . Then your solution works and we get $e^{it} =\ cos t+i\sin t$ (this also gives Taylor series for since cosine for free). – Paramanand Singh Sep 21 '17 at 07:38
  • Another approach is the use of definition $e^{z} =\lim_{n\to\infty} (1+(z/n))^{n}$ which @robjohn has handled in excellent manner in the linked answer. – Paramanand Singh Sep 21 '17 at 07:44
  • Once you've set up the differential equation $\frac{dy}{dt} = i,y$, the standard way of solving it is to move all the y expressions to one side, integrate with respect to $t$, which gives you $\ln y$ on the left; so you take the exponential of both sides. This avoids the aha factor of multiplying by $e^{-it}$. – Papa Smurf Dec 02 '23 at 20:14
15

Lets consider a function from $\mathbb R\to \mathbb C$

$z(\theta) = \cos \theta + i\sin \theta\\ z(\theta)z(\phi) = (\cos \theta + i\sin \theta)(\cos \phi + i\sin \phi) = \cos(\theta + \phi) + i\sin (\theta+\phi) = z(\theta + \phi)$

That is a property of an exponential function. We do not know the base.

For some base:

$\exp (iy) = z(y) =\cos y + i\sin y$

and:

$\exp (x + iy) = \exp(x)\exp(iy) =\exp(x) (\cos y + i\sin y)$

And then you can define $e$ to be the required base. In much of complex analysis, it does not matter that it is the same $e$ as you have learned to be Euler's constant.

However, if you have taken calculus, you should recognize these Taylor expansions.

$e^x = \sum_\limits{n=0}^{\infty} \frac {x^n}{n!}\\ \cos x = \sum_\limits{n=0}^{\infty} \frac {(-1)^nx^{2n}}{(2n)!}\\ \sin x = \sum_\limits{n=0}^{\infty} \frac {(-1)^nx^{2n+1}}{(2n+1)!}$

what is

$e^{ix}$ ?

$e^{ix} = \sum_\limits{n=0}^{\infty} \frac {{ix}^n}{n!}\\ 1 + ix + \frac {(ix)^2}{2} + \frac {(ix)^3}{3!}+ \frac {(ix)^4}{4!} \cdots\\ 1 + ix + \frac {-x^2}{2} + \frac {-ix^3}{3!} + \frac {x^4}{4!} \cdots$

collect the real terms and the imaginary terms

$(1 - \frac {x^2}{2} + \frac {x^4}{4!}\cdots )+ i( x - \frac {x^3}{3!} + \frac {x^5}{5!} \cdots)\\ e^{ix} = \cos x + i\sin x$

Without calculus.

we can define $e = \lim_\limits{n\to\infty}(1+\frac {1}{n})^n\\ e^x =\lim_\limits{n\to\infty} (1+\frac {1}{n})^{nx} $

Make a substitution $m = nx$

$e^x =\lim_\limits{m\to\infty} (1+\frac {x}{m})^m $

Then look at what happens as $m = 1, 2,3, etc.$

We have already shown that multiplication of complex numbers multiplies the lengths and adds the angles.

enter image description here

As $m$ increases hopefully you can see how that sequence of line segments begins to lie on the curve of the circle.

and when $m$ is very large comes to rest on $\cos x + i\sin x$

A.M.
  • 3,944
Doug M
  • 57,877
9

There are several reasons. Even without going into the technical details of why it's correct, here is a small list of reasons for why might be a good idea:

  1. It's easy to use that form to read off the length and angle of your complex number
  2. It's easy to recognize the form and see that it is indeed meant to convey length and angle as opposed to, for instance, the width and height we see in the $a+bi$ form.
  3. The rules for complex multiplication means that hijacking the exponential notation and (mis)use the intuition you have from real exponentiation gives the correct results
Arthur
  • 199,419
6

In this answer, it is shown that $$ \lim_{n\to\infty}\left(1+\frac{i\theta}n\right)^n=\cos(\theta)+i\sin(\theta)\tag1 $$ Therefore, we can say $$ e^{i\theta}=\cos(\theta)+i\sin(\theta)\tag2 $$ We can also use the power series for $e^x$, $\cos(x)$, and $\sin(x)$ to derive $(2)$.

In any case, once we have $(2)$, any point on the unit circle in $\mathbb{C}$ can be represented as $e^{i\theta}$ for some $\theta\in\mathbb{R}/2\pi\mathbb{Z}$; $\theta$ is the argument of that point.

Furthermore, using $(2)$, we can write any point in $\mathbb{C}$ as $$ \begin{align} x+iy &=\overbrace{r\cos(\theta)}^x+i\,\overbrace{r\sin(\theta)}^y\\ &=re^{i\theta}\tag3 \end{align} $$ One important identity is $$ \begin{align} e^{i\theta}e^{i\phi} &=(\cos(\theta)+i\sin(\theta))(\cos(\phi)+i\sin(\phi))\\ &=(\cos(\theta)\cos(\phi)-\sin(\theta)\sin(\phi))+i(\sin(\theta)\cos(\phi)+\cos(\theta)\sin(\phi))\\ &=\cos(\theta+\phi)+i\sin(\theta+\phi)\\ &=e^{i(\theta+\phi)}\tag4 \end{align} $$ Equation $(4)$ tells us that we can combine imaginary exponents like we do real ones.

robjohn
  • 345,667
3

If $z=0$ it is clear that we can take $r=0$ and any value for $\theta$.

Note that any number on the complex unit circle can be written as $\cos t + i \sin t$ for some $t$.

Note that any non zero complex number $z$ can be written as $z= |z| {z \over |z|}$ and ${z \over |z|}$ lies on the complex unit circle.

If we let $r=|z|$, then we see that there is some $t$ such that $z = r (\cos t + i \sin t)$.

As to why $e^{it} = \cos t + i \sin t$, let $\phi(t) = e^{it} - ( \cos t + i \sin t ) $, note that $\phi(0) = 0$ and $\phi'(t) = i\phi(t)$. Hence $e^{-it} \phi(t)$ is a constant from which it follows that $\phi(t) = 0$ for all $t$.

copper.hat
  • 172,524
0

It's a definition.

As to why it's a good definition, the answer comes from the fact that the Taylor series for exp(x) stays the same when we let $x \in \mathbb{C}$.

Aaron P
  • 21
  • 3