[Kstars-devel] Fwd: Plans for threading in KStars

[Akarsh Simha]

Hi Alexey

> I think that euqlidean vectors with norm 1 are better suited for
> representation of points on the sphere. They are more compact (3
> instead of 4 numbers), do not suffer from ambiguity and easier to
> correct for refraction.

I didn't get you here.

> AFAIU precession, nutation, coordinates transformations (horizontal,
> equatorial, ecliptic, galactic etc.), viewpoint change all could be
> expressed as sphere rotation. So all of these could be naturally
> represented as quaternions. Even better all transformations could be
> fused into one quaternion.

Yes, I agree! That should work!

> Two exceptions are proper motion of the stars and refraction. They
> have to be applied for each star but it shouldn't be any slower than
> now.
> 
> 
> Even better. There is no need to recalculate horizontal coordinates
> for each star. Coordinate transformation could be hidden into
> quaternion.
> In pseudocode:
> 
> >q = q_{to horiz.} * q_{precession} * q_{nutation}
> >x_{projected} = q * x * q^{-1}
> >(x,y) = project( x_{projected} )

Yep!

> Projections should be faster as well. There is no need to use
> trigonometry. Equirectngular projection is an exception it'll became
> slower asin/acos are expensive.
> 
> As for quaternion implementaion IMHO we should use eigen.
> 
> 
> P.S. I really should to write summary which describe all the math.

Yes, that would help. I came across quarternions in my mathematical
physics class, but I don't know very much about them. We were talking
about how they represent the group of rotations in 3-D space etc.

Regards
Akarsh