Calculate $$\int_C \frac{2xy^2dx-2yx^2dy}{x^2+y^2},$$ where $C$ is the ellipse $3x^2 +5y^2 = 1$ taken in the positive direction.

I tried to calculate the integral using green theorm.

now i need to build enclosier that doesn't enclose $(0,0)$ i am having hard time guessing what to build.

a circle and ellipse might be perfect but then the domain is not easy to write. can i have hint please ?

  • 3
    $\begingroup$ Are you sure it is a minus between the two terms in the numerator? Green's theorem doesn't seem to work if it is a minus, since $$P_y=\frac{4xy}{x^2+y^2}\\Q_x=-\frac{4xy}{x^2+y^2}$$So $P_y\ne Q_x$ $\endgroup$ – John Doe Jan 7 at 22:19
  • $\begingroup$ oh right , anyway can i apply green therom here ? even if its not $0$ or the integral is too complex to calculate ? $\endgroup$ – Mather Jan 7 at 22:22
  • $\begingroup$ i have also tried to calculate it with basic parametrization of the ellipse but i got harsh integrand $\endgroup$ – Mather Jan 7 at 22:23
  • $\begingroup$ @JohnDoe are you sure? Shouldn't there be a $(x^2+y^2)^2$ in the denominator from quotient rule. I don't think it cancels. $\endgroup$ – AHusain Jan 7 at 22:24
  • $\begingroup$ @AHusain Oh whoops, I didn't differentiate that properly, you''re right! However, even if you do differentiate it properly, it still doesn't give $P_y=Q_x$, the minus sign will still mess this up $\endgroup$ – John Doe Jan 7 at 22:34

We have

$$\int_\mathcal C \frac{2xy^2\mathrm dx-2x^2y\mathrm dy}{x^2+y^2}=\int_{3x^2+5y^2=1}\frac{\mathrm d(x^2)y^2-x^2\mathrm d(y^2)}{x^2+y^2}\tag1$$

Let $u=x^2, v=y^2$. Then we have

$$\int_{3u+5v=1,\quad u,v\ge0}\left(\frac{v}{u+v}\mathrm du-\frac{u}{u+v}\mathrm dv\right)\tag2$$

We compute the first integral. The path is $3u+5v=1,\quad u,v\ge0$, so $v=\frac15-\frac35 u$ and $u\in[0,\frac13]$.

$$\int_\mathcal C\frac v{u+v}\mathrm du=\int_0^\frac13\frac{1-3u}{1+2u}=\cdots=\frac14\left(5\log\frac53-2\right)$$

We compute the second integral. The path is $u=\frac13-\frac53 v$ with $v\in[0,\frac15]$.

$$\int_\mathcal C\frac u{u+v}\mathrm dv=\int_0^\frac15\frac{1-5v}{1-2v}\mathrm dv=\cdots=\frac14\left(2-3\log\frac53\right)$$

Hence the amount that $(2)$ contributes to the total integral $(1)$ is$$\frac14\left(5\log\frac53-2-2+3\log\frac53\right)=2\log\frac53-1$$


As pointed out in the comments, this change of co-ordinates was not a bijective map, so we have not included all the points on the ellipse by doing this calculation, we have only done it for $x,y>0$. For $x,y<0$, we would have the same path, but traversed backwards (i.e. the integration limits will be swapped), which would directly cancel out what we have computed here.

For $x>0,y<0$, we have the same path as in $(2)$, while for $x<0,y>0$, we have this path traversed backwards (you can see this by considering the ellipse traversed clockwise, seeing whether $x$, and correspondingly $u$, is increasing in magnitude or not). So again, these two contributions cancel. This is what gives $0$.

  • $\begingroup$ shouldn't it be $0$ ? $\endgroup$ – Mather Jan 8 at 9:40
  • 1
    $\begingroup$ @JohnDoe This isn't entirely correct. Your new coordinate system only maps a quarter of the ellipse. To go all the way around the ellipse, you must go forwards and backwards on the line in $(u,v)$ twice, making the integral $0$. $\endgroup$ – Dylan Jan 8 at 12:38
  • $\begingroup$ @mather see the above comment ^ I'll edit my answer in a little while $\endgroup$ – John Doe Jan 8 at 13:20
  • $\begingroup$ @Dylan I have edited the question, thanks for pointing that out $\endgroup$ – John Doe Jan 8 at 15:40

You can evaluate the line integral directly by taking $\mathbf r(t) = (\cos(t)/\sqrt 3, \sin(t)/\sqrt 5)$: $$I = \int_C \mathbf F \cdot d\mathbf r = -\int_0^{2 \pi} \frac {\sin 2 t} {4 + \cos 2 t} dt = -\int_0^\pi \frac {\sin 2 t} {4 + \cos 2 t} + \int_0^\pi \frac {\sin 2 t} {4 + \cos 2 t} dt = 0.$$ Green's theorem still holds for $\mathbf F$ even though $\mathbf F$ doesn't have continuous partial derivatives at $(0, 0)$: $$I = -\iint_{3 x^2 + 5 y^2 \leq 1} \frac {4 x y} {x^2 + y^2} dx dy.$$ This form makes it clearer that the result is zero because of the symmetries wrt the coordinate axes.


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.