Astronomical coordinate systems in Julia
julia> Pkg.add("git://github.com/kbarbary/SkyCoords.jl.git")julia> using SkyCoords
julia> c1 = ICRS(0., 0.) # ra, dec in radians
ICRS(0.0,0.0)
julia> c1.ra
0.0
julia> c2 = to_galactic(c1)
Galactic(1.681404315278054,-1.0504869904089078)
julia> c2.l # galactic coordinate fields are l, b rather than ra, dec
1.681404315278054
# FK5J2000 is a version of FK5 fixed at epoch J2000
julia> to_fk5j2000(c1)
FK5J2000(1.1102233710147402e-7,4.411803426976326e-8)These types include an extra field, equinox. The equinox must be specified
when creating the object or converting from another coordinate system:
julia> c2 = to_fk5(c1, 1970.)
FK5(6.276477890771361,-0.0029150998550493794,1970.0)
julia> to_icrs(c2)
ICRS(0.0,-8.673617379884037e-19) # pretty close to round-trippingFor small numbers of points, this package can be much faster than
astropy.coordinates in Python. The following is an example of
transforming points from ICRS to Galactic. The speed difference is
factor of approximately 10,000 for 10 coordinates, 80 for 1000
coordinates and 4 for 100,000 coordinates.
astropy.coordinates
In [1]: from numpy import pi
In [2]: from numpy.random import rand
In [3]: from astropy.coordinates import SkyCoord
In [4]: for n in [10, 1000, 100000]:
...: c = SkyCoord(2.*pi*rand(n), pi*rand(n)-pi/2, unit=('rad', 'rad'))
...: %timeit c.galactic
...:
100 loops, best of 3: 12.5 ms per loop
100 loops, best of 3: 13 ms per loop
10 loops, best of 3: 66.7 ms per loopSkyCoords.jl
julia> using SkyCoords
julia> using TimeIt
julia> for n in [10, 1000, 100000]
c = [ICRS(2pi*rand(), pi*(rand() - 0.5)) for i=1:n]
@timeit to_galactic(c)
end
100000 loops, best of 3: 1.33 µs per loop
1000 loops, best of 3: 163.94 µs per loop
10 loops, best of 3: 16.23 ms per loop