5

I'm finding the area of an ellipse given by $\frac{x^2}{a^2}+\frac{y^2}{b^2} = 1$. I know the answer should be $\pi ab$ (e.g. by Green's theorem). Since we can parameterize the ellipse as $\vec{r}(\theta) = (a\cos{\theta}, b\sin{\theta})$, we can write the polar equation of the ellipse as $r = \sqrt{a^2 \cos^2{\theta}+ b^2\sin^2{\theta}}$. And we can find the area enclosed by a curve $r(\theta)$ by integrating

$$\int_{\theta_1}^{\theta_2} \frac12 r^2 \ \mathrm d\theta.$$

So we should be able to find the area of the ellipse by

$$\frac12 \int_0^{2\pi} a^2 \cos^2{\theta} + b^2 \sin^2{\theta} \ \mathrm d\theta$$

$$= \frac{a^2}{2} \int_0^{2\pi} \cos^2{\theta}\ \mathrm d\theta + \frac{b^2}{2} \int_0^{2\pi} \sin^2{\theta} \ \mathrm d\theta$$

$$= \frac{a^2}{4} \int_0^{2\pi} 1 + \cos{2\theta}\ \mathrm d\theta + \frac{b^2}{4} \int_0^{2\pi} 1- \cos{2\theta}\ \mathrm d\theta$$

$$= \frac{a^2 + b^2}{4} (2\pi) + \frac{a^2-b^2}{4} \underbrace{\int_0^{2\pi} \cos{2\theta} \ \mathrm d\theta}_{\text{This is $0$}}$$

$$=\pi\frac{a^2+b^2}{2}.$$

First of all, this is not the area of an ellipse. Second of all, when I plug in $a=1$, $b=2$, this is not even the right value of the integral, as Wolfram Alpha tells me.

What am I doing wrong?

Ali Caglayan
  • 5,726
Eric Auld
  • 28,127
  • 1
    One can obtain the answer without doing much thinking: to obtain the ellipse, we stretch the circle $a$ units through one axis and $y$ units through the other, which amounts to $ab$ times the area of the unit circle, which is $\pi$, so $ab\pi$. Alternatively, the transformation $(x,y)\mapsto (ax,by)$ has Jacobian $ab$. – Pedro Sep 14 '13 at 03:20
  • @Eric Auld: see http://math.stackexchange.com/questions/1933765/reconciling-x-mathrmdy-y-mathrmdx-x2y2-mathrmd-theta – 926reals Nov 09 '16 at 03:02

5 Answers5

10

There are already a lot of good answers here, so I'm adding this one primarily to dazzle people w/ my Mathematica diagram-creating skills.

As noted previously,

$x(t)=a \cos (t)$

$y(t)=b \sin (t)$

does parametrize an ellipse, but t is not the central angle. What is the relation between t and the central angle?:

cool, huh?

Since y is bSin[t] and x is aCos[t], we have:

$\tan (\theta )=\frac{b \sin (t)}{a \cos (t)}$

or

$\tan (\theta )=\frac{b \tan (t)}{a}$

Solving for t, we have:

$t(\theta )=\tan ^{-1}\left(\frac{a \tan (\theta )}{b}\right)$

We now reparametrize using theta:

$x(\theta )=a \cos (t(\theta ))$

$y(\theta )=b \sin (t(\theta ))$

which ultimately simplifies to:

$x(\theta)=\frac{a}{\sqrt{\frac{a^2 \tan ^2(\theta )}{b^2}+1}}$

$y(\theta)=\frac{a \tan (\theta )}{\sqrt{\frac{a^2 \tan ^2(\theta )}{b^2}+1}}$

Note that, under the new parametrization, $y(\theta)/x(\theta) = tan(\theta)$ as desired.

To compute area, we need $r^2$ which is $x^2+y^2$, or:

$r(\theta )^2 = (\frac{a}{\sqrt{\frac{a^2 \tan ^2(\theta )}{b^2}+1}})^2+ (\frac{a \tan (\theta )}{\sqrt{\frac{a^2 \tan ^2(\theta )}{b^2}+1}})^2$

(note that we could take the square root to get r, but we don't really need it)

The above ultimately simplifies to:

$r(\theta)^2 = \frac{1}{\frac{\cos ^2(\theta )}{a^2}+\frac{\sin ^2(\theta )}{b^2}}$

Now, we can integrate $r^2/2$ to find the area:

$A(\theta) = (\int_0^\theta \frac{1}{\frac{\cos ^2(x )}{a^2}+\frac{\sin ^2(x )}{b^2}} \, dx)/2$

which yields:

$A(\theta) = \frac{1}{2} a b \tan ^{-1}\left(\frac{a \tan (\theta )}{b}\right)$

good for $0\leq \theta <\frac{\pi }{2}$

Interestingly, it doesn't work for $\theta =\frac{\pi }{2}$ so we can't test the obvious case without using a limit:

$\lim_{\theta \to \frac{\pi }{2}} \, \frac{1}{2} a b \tan ^{-1}\left(\frac{a \tan (\theta )}{b}\right)$

which gives us $a*b*Pi/4$ as expected.

4

Your question has been answered, so now we look at how to find the area, using your parametrization, which is a perfectly good one.

The area is the integral of $|y\,dx|$ (or alternately of $|x\,dy|$. over the appropriate interval.

We have $y=b\sin\theta$ and $dx=-a\sin\theta\,d\theta$. So the area is $$\int_0^{2\pi} |-ab\sin^2\theta|\,d\theta.$$ Using $\sin^2\theta=\frac{1-\cos 2\theta}{2}$, we find that the area is $$\int_0^{2\pi} ab\frac{1-\cos 2\theta}{2}\,d\theta.$$ This is indeed $\pi ab$.

André Nicolas
  • 507,029
3

HINT:

Putting $x=r\cos\theta,y=r\sin\theta$

$$\frac {x^2}{a^2}+\frac{y^2}{b^2}=1,$$

$$r^2=\frac{a^2b^2}{b^2\cos^2\theta+a^2\sin^2\theta}=b^2\frac{\sec^2\theta}{\frac{b^2}{a^2}+\tan^2\theta}$$

  • Great. So I see that $\vec{r}(t) = (a\cos{t}, b\sin{t})$ does not parameterize the ellipse by the angle. Can you spot my error in evaluating the integral as I set it up? – Eric Auld Sep 14 '13 at 03:18
  • 1
    @EricAuld, clearly you can see the value of $r$ here in terms of $\theta$. We can't set polar equation in term of $\theta$ in that way. You can see some other methods here : http://en.wikipedia.org/wiki/Ellipse#Area – lab bhattacharjee Sep 14 '13 at 03:23
3

This is another way to do it when one know the area of a circle: Consider the area of a circle with radius 1 in coordinates $(\xi, \eta)$ this is:

$$ \int d\xi d \eta = \pi $$

now if you define new coordinates in your ellipse equation $\xi = \frac{x}{a}, \quad \eta= \frac{y}{b}$ you obtain a circle of radius one: $\xi^2 + \eta^2 =1$

The area of the ellipse you want is $ \int dx dy = ab \int d\xi d\eta = \pi ab$.

2

Here you go - this person even made your mistake, then someone else corrected it.

Link

Glorfindel
  • 3,955
  • Awesome! So I was wrong in assuming that that integral represented the area. Can you see how I have erred in evaluating the integral? – Eric Auld Sep 14 '13 at 03:14