4

I am studying ellipses and need to prove the intersection of a plane and an elliptical cylinder is still an ellipse in 3D. My approach is to find its center and axes and have been struggling.

I have a parametric equation, $\left(x,\, y,\, z\right) = \left(\frac{1}{4}-\frac{3}{2}\cos t-\frac{1}{4}\sin t,\, 2\cos t,\, \sin t \right)$ . This is an intersection between a plane $4x+3y+z=1$ and an elliptical cylinder $y^2+4z^2=4$.

How do I find its center and two axes? Equivalently, how can I express it in the vector form:

$$\mathbf x (t)=\mathbf c+(\cos t)\mathbf u+(\sin t)\mathbf v$$

where $\mathbf u$ and $\mathbf v$ are orthogonal vectors from the center $\mathbf c$ whose norms represent the lengths of the axes and whose directions represent the directions of the axes.

Wei
  • 143
  • 1
    Welcome to [math.se] SE. Take a [tour]. You'll find that simple "Here's the statement of my question, solve it for me" posts will be poorly received. What is better is for you to add context (with an [edit]): What you understand about the problem, what you've tried so far, etc.; something both to show you are part of the learning experience and to help us guide you to the appropriate help. You can consult this link for further guidance. – Another User Jul 09 '22 at 12:03
  • Ellipse. It only has 2 axes. – Wei Jul 09 '22 at 12:17
  • I am studying ellipses and need to prove the intersection of a plane and an elliptical cylinder is still an ellipse, which is true. I am just struggling with the proof. – Wei Jul 09 '22 at 12:30
  • All ellipse shapes are circles viewed from an angle not perpindicular to the circumference of the circle. So the trick is to figure out the viewing angle that is perpindicular or more properly, the z axis of the circle centre, hence the trigonometric angles in calculations to express the shadow and dimension of the ellipse. – Rhodie Jul 09 '22 at 20:45

3 Answers3

3

$$\left(x,\, y,\, z\right) = \left(\frac{1}{4}-\frac{3}{2}\cos t-\frac{1}{4}\sin t,\, 2\cos t,\, \sin t \right)=(\frac{1}{4},0,0)+(\frac{-3}{2},2,0)\cos t+(\frac{-1}{4},0,1)\sin t$$ So $c=(\frac{1}{4},0,0)$, $u=(\frac{-3}{2},2,0)$ and $v=(\frac{-1}{4},0,1)$

3

The center of the ellipse is $C=(1/4,0,0)$. Given a generic point $P(t)$ on the ellipse, you can write the squared distance $(P-C)^2$ as a function of $t$ and find for which values of $t$ it is a minimum or maximum (one can do that with or without calculus). That will give you the vertices of the ellipse.

I can write the explicit result, if needed, but it isn't particularly appealing.

EDIT.

To prove that a curve is an ellipse, it isn't necessary to use the "canonical" parametrization you are looking for, in the sense that $\mathbf u$ and $\mathbf v$ needn't be orthogonal. One can show that the curve is still an ellipse for any pair of (non parallel) vectors $\mathbf u$ and $\mathbf v$, which then define two conjugate semidiameters.

In fact you can check that the line tangent to the ellipse at $\mathbf c+\mathbf u$ (an endpoint of the first diameter, corresponding to $t=0$) is parallel to $\mathbf v$, while the line tangent to the ellipse at $\mathbf c+\mathbf v$ (an endpoint of the second diameter, corresponding to $t=\pi/2$) is parallel to $\mathbf u$, which is the definition of conjugate diameters.

If you want then to find the axes of the ellipse, you can use the formulas explained here.

Intelligenti pauca
  • 50,470
  • 4
  • 42
  • 77
  • Thank you. I know the center is $(1/4, 0, 0)$ but cannot prove it. Is there a theorem for determining it? – Wei Jul 09 '22 at 16:29
  • Well, $(1/4,0,0)$ is the average value of $(x,y,z)$ in your parametrization, hence.... And it's also the intersection of the cylinder axis with the plane. – Intelligenti pauca Jul 09 '22 at 16:46
2

Let $ r = [x, y, z]^T $

Then

$ r = r_0 + \cos(t) v_1 + \sin(t) v_2 = r_0 + [v_1, v_2] U(t) $

where $r_0 = \begin{bmatrix} \frac{1}{4} \\ 0 \\ 0 \end{bmatrix},\hspace{20pt} v_1 = \begin{bmatrix} -1.5 \\ 2 \\ 0 \end{bmatrix} , \hspace{20pt} v_2 = \begin{bmatrix} -0.25 \\ 0 \\ 1 \end{bmatrix} ,\hspace{20pt} U(t) = \begin{bmatrix} \cos(t) \\ \sin(t) \end{bmatrix} $

Using Gram-Schmidt orthogonalization, we get

$ [v_1, v_2] = [w_1, w_2] G $

where

$ w_1 = \begin{bmatrix} -0.6 \\ 0.8 \\ 0 \end{bmatrix} , w_2 = \begin{bmatrix} -0.156892908 \\ -0.117669681 \\ 0.980580676 \end{bmatrix} $

and

$ G = \begin{bmatrix} 2.5 && 0.15 \\ 0 && 1.019803903 \end{bmatrix} $

Note that the set $\{ w_1, w_2 \} $ is an orthonormal set, therefore, $r$ can be described by a $2D$ coordinate vector $ u $, i.e.

$ r = r_0 + W u $

where $W = [w_1, w_2]$ and $ u = G U(t) \Longleftrightarrow U(t) = G^{-1} u $

To find the equation governing $u$, we use the fact that

$ U^T(t) U(t) = 1 $

From which it follows that

$ u^T G^{-T} G^{-1} u = 1 \hspace{20pt} (1a)$

And we have

$ G^{-T} G^{-1} =\begin{bmatrix} 0.16 && -0.023533936 \\ -0.023533936 && 0.965 \end{bmatrix} $

Diagonalizing $G^{-T}G^{-1} $ into $ R D R^T $, we find that

$ D = \begin{bmatrix} 0.159312579 && 0 \\ 0 && 0.965687421 \end{bmatrix}$

$ R = \begin{bmatrix} 0.999573668 && -0.029197307 \\ -0.029197307 && 0.999573668 \end{bmatrix} $

Now equation $(1a)$ can be re-written as

$ u^T R D R^T u = 1 \hspace{20pt} (1b) $

So that

$ D^{1/2} R^T u = U(s) $

where $U(s)$ is the unit vector

$U(s) = \begin{bmatrix} \cos(s) \\ \sin(s) \end{bmatrix} $

Thus, finally, the coordinate vector $u$ can be expressed as,

$ u = R D^{-1/2} U(s) $

And with that,

$ r = r_0 + W u = r_0 + W R D^{-1/2} U(s) $

So the orthogonal vectors that represent the semi-major and semi-minor axes of the ellipse $r$ are the columns of the $3 \times 2 $ matrix $ W R D^{-1/2} $ which is given by

$ W R D^{-1/2} = \begin{bmatrix} -1.514068635 && -0.14176096 \\ 1.994848166 && -0.14346009 \\ 0.071730043 && 0.997424083 \end{bmatrix} $

The first column of the above matrix has a length of $2.505387841 $ and the second column has a length of $1.017610813 $

Therefore, the semi-major axis is along the the first column, and the semi-minor axis is along the second column.

The center, of course, is $r_0$.


Note that

$ r - r_0 = [v_1, v_2] U(t) = W G U(t) = W R D^{-1/2} U(s) $

Hence,

$ U(t) = G^{-1} R D^{-1/2} U(s) $

But $G^{-T} G^{-1} = R D R^T $, so it follows that $G^{-1} = R' D^{1/2} R^T $ for some rotation matrix $R'$, which is given by $R' = G^{-1} R D^{-1/2} $

Therefore,

$ U(t) = R' U(s) $

i.e. $\begin{bmatrix} \cos(t) \\ \sin(t) \end{bmatrix} = \begin{bmatrix} \cos(\theta) && - \sin(\theta) \\ \sin(\theta) && \cos(\theta) \end{bmatrix} \begin{bmatrix} \cos(s) \\ \sin(s) \end{bmatrix} $

From this it follows that $ t = s + \theta + 2 k \pi , \ k $ integer.

Hosam Hajeer
  • 21,978
  • Thank you! This is the answer I have been seeking. And it can be generalized to higher dimensions as well. The only clarification is that it should be $r - r_0= W u$ given the notation here. – Wei Jul 10 '22 at 12:56
  • 1
    Yes. Thanks for pointing this out. Corrected. – Hosam Hajeer Jul 10 '22 at 13:57