If the ellipse $\dfrac{x^2}{A}+\dfrac{y^2}{B}=1$ is to enclose the circle $x^2+y^2=2y$, what values of $A,B>0$ minimize the area of the ellipse?

So far I've put the circle equation into the standard form: $x^2+(y-1)^2=1$, and I know that $A,B$ represents the length of semi-major and semi-minor axis. I'm not sure how to make sure the circle is enclosed in the ellipse.

  • 1
    $\begingroup$ Shouldn't $A$ be equal to $a^2$ and $B$ equal to $b^2$ ? Than $\sqrt{A}$ and $\sqrt{B}$ would represent the length of the semi-major and semi-minor axis. $\endgroup$ – Curious Aug 9 '16 at 16:23

We require the two conics to touch each other. Eliminating $x^2$ leads to the quadratic equation $$(A-B)y^2+2By-AB=0$$

Assuming $A\neq B$, applying the condition that the discriminant is zero leads to the equation $$A^2-AB+B=0$$

We now need to minimize the area $\Delta=\pi\sqrt{AB}$.

Therefore we can differentiate $$\Delta^2=\pi^2AB=\pi^2\frac{A^3}{A-1}$$

Setting the derivative to zero will give $$A=\frac 32, B=\frac 92$$

It is readily seen that this will provide the minimum area since there is no maximum. Therefore the semiaxes are $$\sqrt{A}=\sqrt{\frac 32}, \sqrt{B}=\frac{3}{\sqrt{2}}$$

The minimum ellipse area is then $$\frac{3\sqrt{3}}{2}\pi$$

  • $\begingroup$ How did you eliminate x squared? I've been playing around with both equations but I get to a point where I can't factor anything. $\endgroup$ – Jack A Mar 19 '19 at 17:37

The problem has already been solved. If I put a second circle $x^2+(y+1)^2=1$ in the diagram, the problem is:

What is the ellipse of smallest area that can enclose two non-overlapping unit circles?

On Erich Friedman's Packing Centre the following answer by James Buddenhagen is given:

Two circles in a minimum-area ellipse

The ellipse that solves both the problem above and the original question has semi-major axis $\frac{3}{\sqrt2}$, semi-minor axis $\sqrt\frac32$ and area $\frac{3\sqrt3\pi}2$.

  • $\begingroup$ A proof would be nice. The link does not have one. $\endgroup$ – marty cohen Aug 9 '16 at 20:29
  • $\begingroup$ @martycohen Is given by Quinn above. $\endgroup$ – Parcly Taxel Aug 10 '16 at 4:26
  • $\begingroup$ The link does not have a proof. $\endgroup$ – marty cohen Aug 10 '16 at 4:48
  • $\begingroup$ I meant Quinn's proof, on this same page! $\endgroup$ – Parcly Taxel Aug 10 '16 at 4:49

From the equation $x^2+(y-1)^2=1$, the circle has center $(0, 1)$ and radius $1$. Therefore the points on it with minimum and maximum $x$ and $y$ values are $(0, 0), (1, 1), (0, 2), (-1, 1)$.

Since the ellipse has center at the origin, it has to have a max $y$ value of at least $2$ and a max $x$ value of at least $1$.

For the equation $\dfrac{x^2}{A}+\dfrac{y^2}{B}=1$, this means that $A \ge 1$ and $B \ge 4 = 2^2$.

However, if the ellipse has greater curvature at the top than the circle, it will intersect the circle. The radius of curvature of the ellipse with $A=1, B=4$ is $\frac{a}{b} =\frac14 $, which is smaller than that of the circle, which is $1$. So, we have to modify the values of $A$ and $B$ so that the ellipse is tangent to the circle.

An easy way to do this is to make the ellipse a circle of radius $2$, so the values are $A=B=4$.

Another possibility is to make $A$ slightly larger and $B$ larger so that the ellipse is tangent to the circle at two points. I think that, for any $A> 1$, the value of $B$ that makes this happen could be determined, but I do not feel like working this out. The desired answer would be the one that minimizes $AB$.


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.