Shortest distance between point and a line in n-dimensions using generalised skew-symmetric matrix

359 Views Asked by At

Consider a line which passes through the origin and with direction vector $\vec g= G \vec f$ where $G$ is an $n+1$ by $n$ matrix of rank $n$

Consider a point $p$ in $n+1$ space

I'd like to find the $f$ which minimises the distance between the line and point $p$ over the sphere $f^{T}f=1$

If $g$ is three dimensional (it isn't in general) then I could use the distance equation:

$\frac {\Vert p\times Gf \Vert}{\Vert Gf \Vert}$

And i can introduce the skew-symmetric matrix for $p$ and also minimise the square instead to give:

$\frac {\Vert f^{T}G^{T} [p_\times]^{T}[p_\times] Gf \Vert}{\Vert f^{T}G^{T}Gf \Vert}$

where $[p_x]$ is the equivalent skew-symmetric matrix of cross product with $p$

I could then transform $f$ using cholseky decomposition of $G^{T}G$ and solve...

But i don't know how to do this in n-dimensions (i've tried to use the dot product formulation of distance to line without success).