2
$\begingroup$

I apologize if this is a well known problem and I've just missed the answer, but I have searched fairly extensively. Perhaps there is another way to phrase it that I've been missing?

This question was posted four days ago at Math.SE, but with no answer or comments.

Suppose there are two submanifolds $M_1$ and $M_2$ embedded within $SO(3)$.

$M_1 =\left\{ P_1 e^{\, \mathbf \xi_1 t} | t \in \left[0, ^{2 \pi}/_{|\xi_1|} \right] \right\}$

$M_2 =\left\{ P_2 e^{\, \mathbf \xi_2 t} | t \in \left[0, ^{2 \pi}/_{|\xi_2|} \right] \right\}$

with $P_1, P_2 \in SO(3)$ and $\mathbf \xi_1, \mathbf \xi_2 \in \mathfrak{so(3)}$

Is there an analytic way to determine the pair of points on $M_1$ and $M_2$ that minimizes the geodesic distance between them.

$\arg \min d(m_1, m_2)= \| \log m_1 m_2^\top\|\; : \; m_1 \in M_1, m_2 \in M_2$


The specific problem I'm trying to solve is for a rigid body with an initial attitude and a fixed rotation rate, what is the nearest attitude it will reach with a defined roll and pitch so $\xi_1$ would also point only in the direction of $i_z$ if that further constraint helps. I can solve this numerically, but am looking for an analytic solution to try and implement this on a resource constrained platform.

$\endgroup$
2
  • $\begingroup$ Should all those $\subseteq$'s be $\in$'s? Also, could you clarify what quantity is being minimized and what the quantifiers on $m_1$ and $m_2$ are? $\endgroup$
    – MTyson
    Commented Nov 5, 2017 at 23:00
  • $\begingroup$ @MTyson thanks, yes they should. i hope my edit makes the question more clear! $\endgroup$
    – peabody124
    Commented Nov 5, 2017 at 23:35

1 Answer 1

5
$\begingroup$

I'm assuming you meant $P_1, P_2 \in SO(3)$, and that $M_1$ and $M_2$ are 1-parameter subgroups. Assuming you want the biinvariant metric on $SO(3)$, the 1-parameter subgroups are also geodesics. If I were doing it, I'd lift them to the double cover (the 3-sphere), where they're great circles, apply an isometry to both of them so that the first moves to standard position, say the $xy$-plane (where $z=w=0$), and find the closest points there. Then undo your isometry.

$\endgroup$
2
  • $\begingroup$ and would finding the closest point in the 3-sphere allow an analytic approach to finding the closet points, or would it still be an iterative procedure? $\endgroup$
    – peabody124
    Commented Nov 6, 2017 at 12:50
  • $\begingroup$ It would be analytic. $\endgroup$ Commented Nov 6, 2017 at 18:30

You must log in to answer this question.

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