2
$\begingroup$

This is somewhat similar to my other question here.

Consider the two ellipses given by the equations

\begin{equation} \frac{x^2}{2^2} + \frac{(y-1)^2}{1^2} = 1 \end{equation}

and

\begin{equation} \frac{x^2}{1^2} + \frac{(y-4)^2}{(1/2)^2} = 1. \end{equation}

How do I find the coordinates for the two tangent points of their common tangent at the left side of the ellipses? (I hope the question makes sense.)

$\endgroup$
1
$\begingroup$

Let such a tangent line be $y=mx+c$

Solving simultaneously with the ellipses gives two quadratics in $x$ which are $$(4m^2+1)x^2+8mx(c-1)+4c^2-8c=0$$ and $$(4m^2+1)x^2+8mx(c-4)+4c^2-32c+63=0$$

Each of these must have double roots at the point of tangency and therefore the discriminant is zero, so this leads to two equations in $m$ and $c$ which are $$64m^2(c-1)^2=4(4m^2+1)(4c^2-8c)$$ and $$64m^2(c-4)^2=4(4m^2+1)(4c^2-32c+63)$$

Dividing these eliminates the $m$ terms and we get an equation for $c$ which is $$3c^2-30c+63=0\implies c=3,7$$

A diagram will confirm that the value $c=3$ gives us the tangents which pass between the ellipses, so we choose $c=7$ from which we get the gradient for the common tangent on the left as $$m=+\frac{\sqrt{35}}{2}$$

Using "$x=-\frac{B}{2A}$" the $x$-coordinates of the points of tangency which are $$x_1=-\frac{\sqrt{35}}{3}$$ and $$x_2=-\frac{\sqrt{35}}{6}$$

$\endgroup$
  • $\begingroup$ This looks really nice! Can you please extend your explanations, since I don't understand them to be honest? :-) $\endgroup$ – Svend Tveskæg Apr 7 '17 at 18:13
  • $\begingroup$ I have added some extra explanations, but have not included all the algebra. Please let me know if there is anything you don't understand. $\endgroup$ – David Quinn Apr 7 '17 at 18:55
  • 1
    $\begingroup$ You have accepted another answer, so I don't see why you need me to elaborate further on mine. But since you ask, the first quadratic equation is derived from multiplying out and simplifying $$\frac{x^2}{4}+(mx+c-1)^2=1$$ which is quicker for you to do yourself on paper than it is for me to type out in MathJax. :) $\endgroup$ – David Quinn Apr 14 '17 at 21:15
  • 1
    $\begingroup$ No the method is independent of the actual ellipse equations. When a line touches a curve there must be a double root... $\endgroup$ – David Quinn Apr 14 '17 at 21:42
  • 1
    $\begingroup$ It's the solution of $Ax^2+Bx+C=0$ when $B^2-4AC=0$ $\endgroup$ – David Quinn Apr 15 '17 at 5:29
2
$\begingroup$

Let's introduce the following new variables: $$2u=x\ \text{ and }\ y-1=v.$$

With these new variables, we have

$$u^2+v^2=1\ \text{ and }\ u^2+(v-3)^2=\frac14$$

that is, we have two circles as shown in the figure below.

enter image description here

We have similar triangles and we can see that $OD=6$. Also, by the Pythagorean theorem $DC=\frac{\sqrt{35}}2$ and by the similarity of $OBD$ and $O'CD$: $OB=\frac6{\sqrt{35}}$. So, the slope of the red straight lien is $-\sqrt{35}$. Don't forget a about the other tangent, not shown, whose slope is $\sqrt{35}$.

The two tangent lines in the $u,v$ system are

$$v=-\sqrt{35}u+6\ \text{ and } \ v=\sqrt{35}u+6.$$

Returning to the $x,y$ coordinate system, we get

$$y=-\frac{\sqrt{35}}2x+7\ \text{ and } \ y=\frac{\sqrt{35}}2x+7.$$

EDIT

Unforgivable! I forgave the other pair of tangents:

enter image description here

After similar calculations we get the equations of the other pair of tangent lines.

$$y=-\frac{\sqrt3}{2}x+3\ \text{ and }y=\frac{\sqrt3}{2}x+3 \ $$

$\endgroup$
  • 1
    $\begingroup$ It’s worth pointing out that this solution works because the ellipses are homothetic. $\endgroup$ – amd Apr 7 '17 at 19:29
  • 1
    $\begingroup$ @SvendTveskæg: I had to edit my answer. $\endgroup$ – zoli Apr 8 '17 at 12:06
  • 1
    $\begingroup$ @SvendTveskæg: Thank you, but I had one more mistake. Take a look at the equations of the blue tangents. $\endgroup$ – zoli Apr 9 '17 at 7:16
  • 1
    $\begingroup$ $OBD$ and $O'CD$ are similar because their angles equal. At $O$ and $C$ they have right angles and $D$ is a common angle. The third angles have to be equal because the sum of the angles is $180^{\circ}$. $\endgroup$ – zoli Apr 11 '17 at 20:35
  • 1
    $\begingroup$ @SvendTveskæg: Not a stupid question. I was wrong. Of course the right result is $\pm \sqrt{3}/2$. $\endgroup$ – zoli Apr 12 '17 at 10:24

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.