Packed Ellipse

Calculus Level 2

Three identical circles are packed in an ellipse, as shown. Let A , O , B A,O,B be the centers of these circles (from left to right).

What is the angle A O B \angle AOB such that the area of the ellipse is minimized?

2 tan 1 30 2\tan^{-1}\sqrt{30} 2 tan 1 31 2\tan^{-1}\sqrt{31} 2 tan 1 32 2\tan^{-1}\sqrt{32} 2 tan 1 33 2\tan^{-1}\sqrt{33} 2 tan 1 34 2\tan^{-1}\sqrt{34} 2 tan 1 35 2\tan^{-1}\sqrt{35}

This section requires Javascript.
You are seeing this because something didn't load right. We suggest you, (a) try refreshing the page, (b) enabling javascript if it is disabled on your browser and, finally, (c) loading the non-javascript version of this page . We're sorry about the hassle.

4 solutions

Mark Hennings
Oct 14, 2018

If A O B = 2 θ \angle AOB = 2\theta , then A , O , B A,O,B can be given coordinates ( 2 sin θ , 0 ) (-2\sin\theta,0) , ( 0 , 2 cos θ ) (0,-2\cos\theta) and ( 2 sin θ , 0 ) (2\sin\theta,0) respectively. Then the ellipse has semi-minor axis b = 1 + 2 cos θ b = 1+2\cos\theta , and has equation x 2 a 2 + y 2 ( 1 + 2 cos θ ) 2 = 1 \frac{x^2}{a^2} + \frac{y^2}{(1 + 2\cos\theta)^2} \; = \; 1 The circle ( x 2 sin θ ) 2 + y 2 = 1 (x - 2\sin\theta)^2 + y^2 \; = \; 1 must touch the ellipse at exactly two points, so that equation x 2 a 2 + 1 ( x 2 sin θ ) 2 ( 1 + 2 cos θ ) 2 = 1 [ ( 1 + 2 cos θ ) 2 a 2 ] ( x + 2 a 2 sin θ ( 1 + 2 cos θ ) 2 a 2 ) 2 = 4 a 2 ( 1 + cos θ ) ( 1 + 2 cos θ ) 2 a 2 [ ( 1 + 2 cos θ ) 2 a 2 cos θ ] \begin{aligned} \frac{x^2}{a^2} + \frac{1 - (x - 2\sin\theta)^2}{(1 + 2\cos\theta)^2} & = \; 1 \\ \big[(1 + 2\cos\theta)^2 - a^2\big]\left(x + \frac{2a^2\sin\theta}{(1 + 2\cos\theta)^2-a^2}\right)^2 & = \; \frac{4a^2(1 + \cos\theta)}{(1 + 2\cos\theta)^2 - a^2}\big[(1 + 2\cos\theta)^2 - a^2\cos\theta\big] \end{aligned} has exactly one solution in x x , which means that a = 1 + 2 cos θ cos θ a \; = \; \frac{1 + 2\cos\theta}{\sqrt{\cos\theta}} The ellipse and the circle so defined will in fact only meet if ( 1 + 2 sin θ ) cos θ < 1 (1+2\sin\theta)\cos\theta < 1 , or θ > 1.21556 \theta > 1.21556 . The diagram, as drawn, is only possible for sufficiently large values of θ \theta . Simple calculus then shows that the ellipse area A = π a b = π ( 1 + 2 cos θ ) 2 cos θ A \; =\; \pi ab \; = \; \pi \frac{(1 + 2\cos\theta)^2}{\sqrt{\cos\theta}} is minimized when cos θ = 1 6 \cos\theta = \tfrac16 , and hence when tan θ = 35 \tan\theta = \sqrt{35} . Thus the desired answer is 2 tan 1 35 \boxed{2\tan^{-1}\sqrt{35}} .

I had the right idea then. I figured I was missing some nice piece of geometry that would make the algebra easier but no it’s just a monster :p

Zac Mann - 2 years, 7 months ago

Log in to reply

Well, we have done some elementary trigonometry/geometry to get the coordinates of A , O , B A,O,B . The next bit of the problem (finding the value of (a)) is basically requiring a quadratic to have a repeated root. I did this by completing the square, but you could set the discriminant to zero. The last step is a simple bit of calculus. Is this really a monster?

Mark Hennings - 2 years, 7 months ago

Log in to reply

I guess not. I think part of my problem was that I left the radii of the circles as a variable instead of just letting it be 1. That and I went away from the circle intersecting the ellipse at two points method (which I couldn't see how to guarantee two solutions but now I see you can let the discriminant be 0), to a method where I tried to find a formula for the distance of a point from an ellipse, the minimum of which I would set to R. I did manage to get an equation relating a and theta but I don't think it was correct.

Zac Mann - 2 years, 7 months ago

In the first line, should "semi-major" not read "semi-minor"?

Dennis Engel - 2 years, 7 months ago

Log in to reply

Well spotted!

Mark Hennings - 2 years, 7 months ago

"The circle must touch the ellipse at exactly two points, so that equation ... eq in x only ... has exactly one solution in x x ..."

This is not exactly right, as y might be imaginary for that particular x. This is indeed what happens for approximately θ < 1.21556 \theta<1.21556 . For those θ \theta the correct ellipse touches right circle at ( x , y ) = ( 1 + 2 s i n θ , 0 ) (x,y)=(1+2sin\theta,0) , and the correct semi-major axis is a = 1 + 2 s i n θ a=1+2sin\theta .

Simple example is for θ = π / 4 \theta=\pi/4 , where the correct ellipse is a circle.

Darko Simonovic - 2 years, 6 months ago

Log in to reply

The problem with your comment is that the circles constructed for these values of θ \theta look nothing like those of the diagram (and that the circles are packed in the ellipse, "as shown", was a key part of the question). For small values of θ \theta , the circles with centres ( ± 2 sin θ , 0 ) (\pm2\sin\theta,0) are not meant to touch the ellipse on the major axis, which is what is going on in your examples. Agreed, we need ( 2 sin θ + 1 ) cos θ < 1 (2\sin\theta + 1)\cos\theta < 1 , or θ > 1.21556 \theta > 1.21556 for the given diagram to be viable, but only these values of θ \theta are relevant to the problem.

Mark Hennings - 2 years, 6 months ago

Log in to reply

My initial understanding of the image and text is that 3 circles are packed in an ellipse, not necessarily that circles are touching an ellipse exactly 5 times (on a second thought, for cases circles are touching an ellipse at 3 points, one would have to give further comments, whether side circles stay on the semi-major axis, or the circles are snuggly packed), but all this is irrelevant.

To me reasons and regions where "one solution to quadratic equation" fails is more interesting than the problem itself. This was not evident to me. In a similar problem, whatever number came, I would have pronounced it as a solution. Now, I might remember to check it back if the complete solution remained real, and look when and why solution ceases to work.

Darko Simonovic - 2 years, 6 months ago
David Vreken
Oct 22, 2018

Let the circles be unit circles, let the ellipse be centered at the origin C C , and let A O B = 2 θ \angle AOB = 2\theta . Also, let D D be one of the points of tangency between the ellipse and circle B B , and E E be the point of tangency between the ellipse and circle O O , as shown below.

Then B O C = θ \angle BOC = \theta , and B C = 2 sin θ BC = 2 \sin \theta and C O = 2 cos θ CO = 2 \cos \theta . This means the minor axis of the ellipse b = C E = 2 cos θ + 1 b = CE = 2 \cos \theta + 1 and B B has coordinates ( 2 sin θ , 0 ) (2 \sin \theta, 0) .

Let D D have coordinates ( p , q ) (p, q) and let a a be the major axis of the ellipse. Since D D is on the ellipse, p 2 a 2 + q 2 b 2 = 1 \frac{p^2}{a^2} + \frac{q^2}{b^2} = 1 . Since D D is on circle B B , ( p 2 sin θ ) 2 + q 2 = 1 (p - 2 \sin \theta)^2 + q^2 = 1 . Since the slope of B D BD must be the same as the slope of the normal of the ellipse at ( p , q ) (p, q) , q p 2 sin θ = a 2 q b 2 p \frac{q}{p - 2 \sin \theta} = \frac{a^2q}{b^2p} . Solving these 3 3 equations, along with the identity sin 2 θ + cos 2 θ = 1 \sin^2 \theta + \cos^2 \theta = 1 , we find that a a reduces to a = b 2 b 1 a = b\sqrt{\frac{2}{b - 1}} .

The area of an ellipse is A = π a b A = \pi ab , so in this ellipse, A = π b 2 2 b 1 A = \pi b^2 \sqrt{\frac{2}{b - 1}} . Using a derivative we find that this area is minimized when b = 4 3 b = \frac{4}{3} . Since b = 2 cos θ + 1 b = 2 \cos \theta + 1 , cos θ = 1 6 \cos \theta = \frac{1}{6} , which makes tan θ = 35 \tan \theta = \sqrt{35} , and A O B = 2 tan 1 35 \angle AOB = \boxed{2 \tan^{-1} \sqrt{35}} .

Solving those 3 equations to get a=f(b) is quite messy. Did you use any tricks there?

Yathish Dhavala - 2 years, 7 months ago

Log in to reply

Here's what I did:

The ellipse equation p 2 a 2 + q 2 b 2 = 1 \frac{p^2}{a^2} + \frac{q^2}{b^2} = 1 can be rearranged to q 2 = b 2 b 2 p 2 a 2 q^2 = b^2 - \frac{b^2p^2}{a^2} .

The slope equation q p 2 sin θ = a 2 q b 2 p \frac{q}{p - 2 \sin \theta} = \frac{a^2q}{b^2p} can be rearranged to p 2 sin θ = b 2 p a 2 p - 2 \sin \theta = \frac{b^2p}{a^2} .

Substituting these into the circle equation ( p 2 sin θ ) 2 + q 2 = 1 (p - 2 \sin \theta)^2 + q^2 = 1 , we have b 4 p 2 a 4 + b 2 b 2 p 2 a 2 = 1 \frac{b^4p^2}{a^4} + b^2 - \frac{b^2p^2}{a^2} = 1 , which can be rearranged to p 2 b 2 ( a 2 b 2 ) = a 4 ( b 2 1 ) p^2b^2(a^2 - b^2) = a^4(b^2 - 1) .

The slope equation q p 2 sin θ = a 2 q b 2 p \frac{q}{p - 2 \sin \theta} = \frac{a^2q}{b^2p} can also be rearranged to p = 2 a 2 sin θ a 2 b 2 p = \frac{2a^2 \sin \theta}{a^2 - b^2} , so substituting this in gives 4 a 4 sin 2 θ ( a 2 b 2 ) 2 b 2 ( a 2 b 2 ) = a 4 ( b 2 1 ) \frac{4a^4 \sin^2 \theta}{(a^2 - b^2)^2}b^2(a^2 - b^2) = a^4(b^2 - 1) , so a 2 = 4 b 2 sin 2 θ + b 4 b 2 b 2 1 a^2 = \frac{4b^2 \sin^2 \theta + b^4 - b^2}{b^2 - 1} .

Since b = 2 cos θ + 1 b = 2 \cos \theta + 1 , cos θ = b 1 2 \cos \theta = \frac{b - 1}{2} . This means sin 2 θ = 1 cos 2 θ = 1 ( b 1 ) 2 4 \sin^2 \theta = 1 - \cos^2 \theta = 1 - \frac{(b - 1)^2}{4} . Substituting this in the above gives a 2 = 4 b 2 ( 1 ( b 1 ) 2 4 ) + b 4 b 2 b 2 1 a^2 = \frac{4b^2 (1 - \frac{(b - 1)^2}{4}) + b^4 - b^2}{b^2 - 1} which simplifies to a = b 2 b 1 a = b\sqrt{\frac{2}{b - 1}} .

David Vreken - 2 years, 7 months ago

why is CE the minor axis,b?

Yuanhui Zhu - 2 years, 7 months ago

Log in to reply

Since circles A and B are tangent to two places on the ellipse, their centers must lie on the ellipse's horizontal axis. Since the circles are congruent, the diagram has vertical symmetry, so the midpoint C (between A and B) is the center of the ellipse, which makes CE the minor axis.

David Vreken - 2 years, 7 months ago

Log in to reply

got it. thanks

Yuanhui Zhu - 2 years, 7 months ago

My solution appears almost identical, except I did not use any trigonometric functions until the very end.

Ken Hill - 2 years, 4 months ago

This is simpler way of doing this problem using the options given.

Consider the half triangle of the bigger triangle AOB and take the corresponding sides of the triangle as x x and y y

Then by Pythagoras Theorem,

x 2 + y 2 = 4 r 2 x^{2}+y^{2} = 4r^{2} where r r is the radius of the individual circles.

Now, we know that, tan ( A O B / 2 ) = x y \tan( AOB/2) = \dfrac {x}{y}

So the following conditions should satisfy-

1 ] x 2 + y 2 1]x^{2}+y^{2} should be divisible by 4 4 and

2 ] 2] The divisible number x 2 + y 2 4 \dfrac {x^2+y^2}{4} should be a perfect square number. Check all options and only the last one satisfies.

A N S W E R : 2 tan 1 35 ANSWER :\boxed{2\tan^{-1}\sqrt{35}}

N o t e : Note: This method works only if you consider r r to be an I n t e g e r Integer

Why could you not have (for example) x = 31 u y = u x \; = \; \sqrt{31}u \hspace{1cm} y \; = \; u where 32 u 2 = x 2 + y 2 = 4 32u^2 = x^2 + y^2 = 4 (since the circles have unit radius), and so u = 1 2 2 u = \tfrac{1}{2\sqrt{2}} ? The fact that 1 + 35 1 + 35 is a perfect square is not enough to exclude the other options.

Mark Hennings - 2 years, 7 months ago

Log in to reply

The fact is about (35+1)/4 =9 being a perfect square not just 35+1. Have a look, (31+1)/4=8 is not a perfect square. I suggest you carefully read my solution. I said (x^2+y^2)/4 being a perfect square number.

A Former Brilliant Member - 2 years, 7 months ago

Log in to reply

8 is square of 2^(3/2) also r^2 doesn't need to be a whole number(It can be positive real number really)

A Former Brilliant Member - 2 years, 7 months ago

Log in to reply

@A Former Brilliant Member Read the reply I gave to Mark Hennings. I assure you better solutions next time.

A Former Brilliant Member - 2 years, 7 months ago

If you look at my answer you will find that the actual value of y y is 1 3 \tfrac13 . Your argument is based on the assumption that all variables must be integers.

Mark Hennings - 2 years, 7 months ago

Log in to reply

@Mark Hennings Mine is a bit not that satisfactory approach .I actually used one more approach to this and added surety of both I came up with this answer. See the area of triangle AOB is directly related to that of ellipse. More elongated the angle AOB more the area of the Ellipse. Moreover, difference between inverse of tan of (√35/1) and (√31/1) is not significant and clearly they satisfy the above condition. I am a student, not a mathematician, more of a self-learner not having a promising Maths Teacher. I am in learning stage and I will come up with better solutions in future with support of you guys! :)

A Former Brilliant Member - 2 years, 7 months ago
Vinod Kumar
Oct 22, 2018

Answer: Using Pictorial Reference, James Buddenhagen, June 2004, and plotting with WolframAlpha the following is obtained as the answer:

Ellipse with

semi-major axis=8/√6,

semi-minor axis=4/3,

(1) Center Unit Circle = ( 0, -1/3),

(2) Right Unit Circle = ((1/3)√35,0),

(3) Left Unit Circle. = (-(1/3)√35,0)

0 pending reports

×

Problem Loading...

Note Loading...

Set Loading...