Conic Sections

A conic section is the curve that follows the intersection of a cone (or double cone) and a plane. Objects whose orbits are well explained by Newtons law of gravitation (i.e. everything for which GR isn’t needed) follow conic sections. Which type of conic section they follow depends on their orbital parameters.

Here we visualize the intersection of a cone and a plane and briefly run through some the math of how it works.

Cone opening angle (θ)
Cone rotation angle (Φ)

The Math

A plane is defined as the set of all points that satisfy,

Ax+By+Cz=D Ax + By + Cz = D

Where AA , BB , CC , and DD are constants (e.g. WolframAlpha). The plane above is defined by,

z=1 z = 1

A cone with the vertex at the origin and with an opening surrounding the zz axis is defined by,

x2+y2=tan2(θ)z2 x^2 + y^2 = \tan^2(\theta) z^2

Where θ\theta is the opening angle (e.g. WolframAlpha). If we then rotate the cone through some angle ϕ\phi around the xx axis (which is toward us in the diagram above), we can apply the following rotation matrix,

(1000cos(ϕ)sin(ϕ)0sin(ϕ)cos(ϕ)) \begin{pmatrix} 1 & 0 & 0 \\ 0 & \cos(\phi) & -\sin(\phi) \\ 0 & \sin(\phi) & \cos(\phi) \end{pmatrix}

and therefore,

x2+(cos(ϕ)y+sin(ϕ)z)2=tan2(θ)(cos(ϕ)zsin(ϕ)y)2 x^2 + (\cos(\phi)y + \sin(\phi) z)^2 = \tan^2(\theta) (\cos(\phi) z - \sin(\phi) y)^2

(e.g. WolframAlpha).

We can solve for the intersection of these surfaces by substituting z=1z = 1 into the definition of the cone,

x2+(cos(ϕ)y+sin(ϕ))2=tan2(θ)(cos(ϕ)sin(ϕ)y)2 x^2 + (\cos(\phi)y + \sin(\phi))^2 = \tan^2(\theta) (\cos(\phi) - \sin(\phi) y)^2

which gives us a function of xx and yy . The intersection is the set of points that satisfy this equation. Unfortunately this solution is a bit messy, but it can be written1

x2=ay2+by+c x^2 = a y^2 + b y + c (1)

where

a=tan2(θ)sin2(ϕ)+sin2(ϕ)1b=2tan2(θ)sin(ϕ)cos(ϕ)2sin(ϕ)cos(ϕ)c=tan2(θ)sin2(ϕ)sin2(ϕ)+tan2(θ) \begin{align} a &= \tan^2(\theta) \sin^2(\phi) + \sin^2(\phi) - 1 \\ b &= -2 \tan^2(\theta) \sin(\phi) \cos(\phi) - 2 \sin(\phi) \cos(\phi) \\ c &= -\tan^2(\theta) \sin^2(\phi) - \sin^2(\phi) + \tan^2(\theta) \end{align} (2)

This solution results in three different shapes depending on the choices of ϕ\phi and θ\theta .

Ellipse

The most general form of an ellipse is,

Ax2+Bxy+Cy2+Dx+Ey+F=0 Ax^2 + Bxy + Cy^2 + Dx + Ey + F = 0 (3)

as long as the coefficients satisfy,2

B24AC<0 B^2 - 4AC < 0

The intersection is of the correct form (we can rewrite (1) as x2ay2byc=0x^2 -ay^2 - b y - c = 0 ), and so we just need to check the second condition. B=0B = 0 (as we have no xyxy term), and A=1A = 1 (the coefficient of x2x^2 ), and so the inequality becomes,

C=a<0tan2(θ)sin2(ϕ)+sin2(ϕ)1<0sin2(ϕ)(tan2(θ)+1)<1sin2(ϕ)<cos2(θ) \begin{align} -C = a &< 0 \\ \tan^2(\theta) \sin^2(\phi) + \sin^2(\phi) - 1 &< 0 \\ \sin^2(\phi) (\tan^2(\theta) + 1) &< 1 \\ \sin^2(\phi) &< \cos^2(\theta) \end{align}

What does this tell us? Well, for small opening angles, cos2(θ)\cos^2(\theta) is large and so the inequality is almost always satisfied. When is it not? When ϕπ/2\phi \approx \pi / 2 , specifically when ϕπ/2<θ|\phi - \pi / 2| < \theta .3

When the cone is opening up along the plane we do not form an ellipse. How close to the plane this happens depends on the opening angle.

Hyperbola

What happens in the region where the ellipse is invalid? It becomes a hyperbola, the most general form of which is the same as an ellipse,

Ax2+Bxy+Cy2+Dx+Ey+F=0 Ax^2 + Bxy + Cy^2 + Dx + Ey + F = 0

But has the opposite requirement,

B24AC>0 B^2 - 4AC > 0

Following the same argument as before the intersection is a hyperbola when, ϕπ/2>θ|\phi - \pi / 2| > \theta .

Parabola

So we have an ellipse when θ+ϕ<π2\theta + \phi < \frac{\pi}{2} , and a hyperbola when the sum of the angles is >π2> \frac{\pi}{2} . What happens at the crossover point when one arm of the cone is parallel to the plane at θ+ϕ=π/2\theta + \phi = \pi / 2 ? In that case, tan(θ)=cot(ϕ)=cos(ϕ)sin(ϕ)\tan(\theta) = \cot(\phi) = \frac{\cos(\phi)}{\sin(\phi)} . Substituting this into (2) we get,

Which gives us a parabola after we substitute into (1) and solve for yy .


  1. Most of the algebra from here was first done using Maple. I don’t know my trig identities nearly well enough to do it by hand. Once I have the solution, I’ve tried to back out enough steps so that it makes sense.↩︎

  2. This comes from the fact that the general ellipse (3) can be written as (xy)(AB/2B/2C)(xy)+(DE)(xy)+F=0 \begin{pmatrix} x & y \\ \end{pmatrix} \begin{pmatrix} A & B/2 \\ B/2 & C \\ \end{pmatrix} \begin{pmatrix} x \\ y \\ \end{pmatrix} + \begin{pmatrix} D & E \\ \end{pmatrix} \begin{pmatrix} x \\ y \\ \end{pmatrix} + F = 0 where Q=(AB/2B/2C) Q = \begin{pmatrix} A & B/2 \\ B/2 & C \\ \end{pmatrix} is the matrix of the quadratic form. The determinant of this matrix, ACB2/4AC - B^2/4 (which usually appears in the scaled form 4ACB24AC - B^2 ), is the discriminant and determines the properties of the solutions, namely an ellipse if it is >0> 0 and a hyperbola if it is <0< 0 . One way to think about why this works is that in Ax2+Bxy+Cy2+F=0Ax^2 + Bxy + Cy^2 + F= 0 , if AA or BB are large, solutions with large xx or yy are impossible. The squared terms will grow quicker than the cross term. The resulting shape is therefore closed (ellipse). However, if the squared coefficients are small and the cross coefficient is large, solutions with with large xx and yy (where one is large negative, the other large positive) are possible and the shape is not closed (hyperbola). You can play with this on Desmos and I’m sure you can find a more elegant explanation somewhere.↩︎

  3. The conic section is also a parabola when θϕ=π/2\theta - \phi = \pi / 2 and θ±ϕ=3π/2\theta \pm \phi = 3 \pi / 2 . The arguments for the other 3 cases are identical to the one shown.↩︎