1

I want to have two ellipses with constant radial gap in between. Basically if the ellipse 1 is:

$$r_1\left( \theta\right)=\frac{1}{\frac{{\cos^2\theta}}{a_1^2}+\frac{{\sin^2\theta}}{b_1^2}}$$

and ellipse 2 is

$$r_2\left( \theta\right)=\frac{1}{\frac{{\cos^2\theta}}{a_2^2}+\frac{{\sin^2\theta}}{b_1^2}}$$

Now for each $a_1$ and $b_1$ how can I find $a_2$ and $b_2$ to have

$$r_2\left( \theta\right)=r_1\left( \theta\right)+c$$

P.S. Even approximate solutions would also do. I mean even if there is no analytical solution for a constant $c$ radial gap between two ellipses, maybe there is a $c(\theta)$ where it's variation is minimum from 0 to $2\pi$. In other words if we define average gap size as:

$$c_{avg}=\frac{1}{2\pi}\int^{2\pi}_0\left( r_2\left( \theta\right)-r_1\left( \theta\right) \right)d\theta$$

How to find the $a_2$ and $b_2$ in a way to minimize absolute deviation of gap size from average:

$${d}_{abs}=\frac{1}{2\pi}\int^{2\pi}_0\left| c\left( \theta\right)-c_{avg} \right|d\theta$$

Foad
  • 431
  • @Mathmo123 Thanks a lot for editing the tags. – Foad Apr 19 '18 at 11:08
  • 6
    When an ellipse is painted with constant thickness, the new boundary (containing the paint) is no longer an ellipse anymore. See the case for ellipsoid here. It's the case for same *radial thickness* except for circle. – Ng Chung Tak Apr 19 '18 at 12:20

1 Answers1

3

radial constant

There is an analytic expression for a constant radial gap around an ellipse. You start by writing the ellipse equation and getting its gradient. So for example, let the ellipse be $$\frac{x^2}{1}+\frac{y^2}{2}=1.$$ Then its gradient will be $$\nabla=\left(\begin{array}{c}2x\\y\end{array}\right)$$ We will use that in a bit, but understand that it gives the normal direction away from the ellipse for every point. Next convert the ellipse equation to parametric form. $$\left(\begin{array}{c}x\\y\end{array}\right)=\left(\begin{array}{c}CP_{x}\\CP_{y}\end{array}\right)+\left(\begin{array}{c}a\cdot cos(t)\\b\cdot sin(t)\end{array}\right)$$ In this case, I let $a=1$ and $b=\sqrt{2}$ so we get $$\left(\begin{array}{c}x\\y\end{array}\right)=\left(\begin{array}{c}0\\0\end{array}\right)+\left(\begin{array}{c}cos(t)\\\sqrt{2}sin(t)\end{array}\right)$$ Next, I just use the gradient to write the new locus of points that are a constant distance, $r$, from the existing ellipse.

$$\left(\begin{array}{c}x_{1}\\y_{1}\end{array}\right)=\left(\begin{array}{c}0\\0\end{array}\right)+\left(\begin{array}{c}cos(t)\\\sqrt{2}sin(t)\end{array}\right)+r\frac{\left(\begin{array}{c}2\cdot cos(t)\\\sqrt{2}sin(t)\end{array}\right)}{\left\Vert \left(\begin{array}{c}2\cdot cos(t)\\\sqrt{2}sin(t)\end{array}\right)\right\Vert }$$ Hopefully, my picture shows an example for this ellipse with $r=1/4$. I am pretty sure that the new locus is not an ellipse.

Narlin
  • 1,201
  • +1 It's not an ellipse. See comment from @NgChunTak . – Ethan Bolker Apr 19 '18 at 23:48
  • 1
    I had to modify my answer to use a "unit" vector times r (duh! of course) and it made the picture look better. The new locus is still not an ellipse, but the looks could fool someone. – Narlin Apr 20 '18 at 10:15