0

Suppose you have a cube of side length $2a$. You want to inscribe an ellipsoid in the cube such that its semi-axes are in the ratio $1:2:4$. Find a possible orientation of this ellipsoid.

This problem is closely related to this recent problem

My attempt

As discussed in the solutions of the above-referenced problem, the following relation holds

$ b^2 + (2 b)^2 + (4 b)^2 = 3 a^2 $

Where $b$ is the smallest semi-axis length. Therefore,

$ b = \dfrac{a} {\sqrt{7}} $

Now we have the three semi-axes lengths. From the results of this problem, it turns out that there are infinite possible orientations for the ellipsoid whose dimensions we now have.

The orientation is determined by three angles, and we have the following constraints on the angles

$ e_1^T R D^{-1} R^T e_1 = a^2 $

$ e_2^T R D^{-1} R^T e_2 = a^2 $

where

$ D^{-1} = \begin{bmatrix} b^2 && 0 && 0 \\ 0 && 4 b^2 && 0 \\ 0 && 0 && 16 b^2 \end{bmatrix}$

Matrix $R$ can be parameterized as follows

$ R = [ u_1 , u_2, u_3 ] R_0 $

where

$ u_3 = \begin{bmatrix} \sin \alpha \cos \beta \\ \sin \alpha \sin \beta \\ \cos \alpha \end{bmatrix} $

$ u_1 = \begin{bmatrix} \cos \alpha \cos \beta \\ \cos \alpha \sin \beta \\ - \sin \alpha \end{bmatrix} $

$ u_2 = \begin{bmatrix} - \sin \beta \\ \cos \beta \\ 0 \end{bmatrix} $

and

$ R_0 = \begin{bmatrix} \cos \gamma && -\sin \gamma && 0 \\ \sin \gamma && \cos \gamma && 0 \\ 0 && 0 && 1 \end{bmatrix} $

To obtain a finite number of possibilities we have to impose a constraint on these angles $\alpha, \beta, \gamma$.

For example, we can set $\alpha$ or $\beta$ or $\gamma$ to a specific value, then search for solutions for the remaining angles.

I still have to experiment with that.

Your comments on the above analysis, and alternative solutions are appreciated.

  • "As discussed in the solutions of the above-referenced problem" : being the author of one of these solutions, I would have been happy that you have some comment on my solution... – Jean Marie Jun 19 '24 at 21:50
  • @JeanMarie Your solution to that problem is correct. That why awarded it an upvote. This result is known to me before I read your solution. You can check the other referenced problem of mine. –  Jun 19 '24 at 23:34
  • Thanks for your answer. Please understand my comment : it is always interesting on such a site to have exchanges... It wasn't clear to me that you had already met the concept of "orthoptic sphere" of an ellipsoid. At least, I have contributed by giving web references.... – Jean Marie Jun 20 '24 at 05:00

1 Answers1

1

Fixing the angle $\gamma$ to a certain value, it is possible to solve for possible values of $\alpha$ and $\beta$. What I did then, is made $\gamma$ slide over the interval $[0, 2 \pi]$, and solved for $\alpha$ and $\beta$ as functions of $\gamma$. This generated the desired orientation for each value of $\gamma$. The result is shown below

enter image description here