0
$\begingroup$

Say I have one sphere and one plane

$$x^2+y^2+z^2 = 1$$ $$z = 2$$

We can easily calculate closest distance between these. It is easy exercise. Maybe first exercise in differential geometry or something like that. But is there some sense in which we can calculate "total closeness" of the objects? Like integrating over the objects calculating all distances and boil it down to one real number somehow.

$\endgroup$
5
  • 2
    $\begingroup$ en.wikipedia.org/wiki/Hausdorff_distance $\endgroup$
    – AO1992
    Commented May 23, 2019 at 8:21
  • 1
    $\begingroup$ Most points on the planes are very far away, so the "total closeness" is $\infty$. Even the "average closeness" is $\infty$ if you define it sensibly. $\endgroup$ Commented May 23, 2019 at 8:22
  • $\begingroup$ @AO1992 but it is not so much total distance but more like min of max or max of min distance, isn't it? $\endgroup$ Commented May 23, 2019 at 8:22
  • $\begingroup$ You can certainly define "total distance" the way you describe in your last sentence. It will be infinite, as @user10354138 explains. So I'm not sure what your question is. $\endgroup$
    – user856
    Commented May 23, 2019 at 10:05
  • $\begingroup$ It is quite possible to find definitions to get nice finite values. I've found several ways already. I am mostly looking for new ideas how to do so which I haven't thought of already. $\endgroup$ Commented May 23, 2019 at 10:10

2 Answers 2

1
$\begingroup$

I suggest to consider the orthogonal distance from a point of the sphere to the plan instead of the (non-orthogonal) distance to all points of the plan. So, the mean distance is not infinite.

If we consider the surface of the sphere, the orthogonal distance from the point $x,y,z$ to the plan $x=2$ is : $\big|2\pm\sqrt{1-y^2-z^2}\big|$. The sign depends on which hemisphere the point is located.

The mean distance is : $$\frac{\int_{-1}^1\int_{-\sqrt{1-z^2}}^{\sqrt{1-z^2}}\left(2+\sqrt{1-y^2-z^2}\right)dy\:dz+\int_{-1}^1\int_{-\sqrt{1-z^2}}^{\sqrt{1-z^2}}\left(2-\sqrt{1-y^2-z^2}\right)dy\:dz}{\int_{-1}^1\int_{-\sqrt{1-z^2}}^{\sqrt{1-z^2}} dy\:dz+\int_{-1}^1\int_{-\sqrt{1-z^2}}^{\sqrt{1-z^2}} dy\:dz}=\frac{8\pi}{4\pi}=2$$ Thus with this definition of the mean distance, the result is $2$. This was obvious because de mean distance of two symmetric points $(x,y,z)$ and $(-x,y,z)$ is $2$ and this is true for all couples of points of the sphere.

With the same definition and for the same raison the mean distance of all points inside the volume of the sphere will be found $=2$ by a triple integral.

Until the plan doesn't intersect the sphere the mean orthogonal distance is equal to the distance from the center of the sphere to the plan. If the plan intersect the sphere the calculus is more difficult. It should be interesting to compute the double and the triple integral of the absolute value $\big|2\pm\sqrt{1-y^2-z^2}\big|$ which is different from $\big(2\pm\sqrt{1-y^2-z^2}\big)$.

$\endgroup$
1
  • $\begingroup$ The shortest path is already orthogonal to the sphere, so "orthogonal distance" is adding a meaningless adjective here. On the other hand, you haven't integrated on the plane which is what really caused the $\infty$. $\endgroup$ Commented May 23, 2019 at 12:37
1
$\begingroup$

I can’t post a comment yet, but a possibly relevant and helpful piece of information is a classical theorem of Jacobi in Riemannian Geometry roughly stating that a geodesic segment minimizes the arc length with respect to neighboring curves with the same ends if it has no conjugate points. Calculating a geodesic requires solving a specific differential equation of degree two. For more information, see for example https://www.ime.usp.br/~gorodski/teaching/mat5771/ch5.pdf.

$\endgroup$
2
  • $\begingroup$ It is definitely helpful. Thank you. $\endgroup$ Commented May 23, 2019 at 9:31
  • $\begingroup$ You are welcome. $\endgroup$
    – user673903
    Commented May 23, 2019 at 9:35

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .