|
3 | 3 | Using and Designing Coordinate Representations |
4 | 4 | ---------------------------------------------- |
5 | 5 |
|
6 | | -As described in :ref:`astropy-coordinates-overview`, the actual |
7 | | -coordinate data in `~astropy.coordinates` is represented via |
8 | | -"Representation classes"... |
| 6 | +As described in :ref:`astropy-coordinates-overview`, the actual coordinate |
| 7 | +data in `astropy.coordinates` frames is represented via |
| 8 | +"Representation classes". These can be used to store 3-d coordinates in |
| 9 | +various representations, such as cartesian, spherical polar, cylindrical, and |
| 10 | +so on. The built-in representation classes are: |
| 11 | + |
| 12 | +* `~astropy.coordinates.CartesianRepresentation`: cartesian |
| 13 | + coordinates ``x``, ``y``, and ``z`` |
| 14 | +* `~astropy.coordinates.SphericalRepresentation`: spherical |
| 15 | + polar coordinates represented by a longitude (``lon``), a latitude |
| 16 | + (``lat``), and a distance (``distance``). The latitude is a value ranging |
| 17 | + from -90 to 90 degrees. |
| 18 | +* `~astropy.coordinates.UnitSphericalRepresentation`: |
| 19 | + spherical polar coordinates on a unit sphere, represented by a longitude |
| 20 | + (``lon``) and latitude (``lat``) |
| 21 | +* `~astropy.coordinates.PhysicsSphericalRepresentation`: |
| 22 | + spherical polar coordinates, represented by an inclination (``theta``) and |
| 23 | + azimuthal angle (``phi``), and radius ``r``. The inclination goes from 0 to |
| 24 | + 180 degrees, and is related to the latitude in the |
| 25 | + `~astropy.coordinates.SphericalRepresentation` by |
| 26 | + ``theta = 90 deg - lat``. |
| 27 | +* `~astropy.coordinates.CylindricalRepresentation`: |
| 28 | + cylindrical polar coordinates, represented by a cylindrical radius |
| 29 | + (``rho``), azimuthal angle (``phi``), and height (``z``). |
| 30 | + |
| 31 | +Instantiating and converting |
| 32 | +^^^^^^^^^^^^^^^^^^^^^^^^^^^^ |
| 33 | + |
| 34 | +Representation classes should be instantiated with `~astropy.units.Quantity` |
| 35 | +objects: |
9 | 36 |
|
10 | | -.. todo:: @astrofrog writes this |
11 | | - |
12 | | -* describe creating/using representation objects. Emphasize that they |
13 | | - are mostly used inside frame classes |
14 | | -* describe how to get other representations, esp. cartesian, and at least a brief |
15 | | - mention of how things work with/without `distance` |
16 | | -* brief discussion of how users can create their own for use |
17 | | - |
18 | | - |
19 | | - |
20 | | - |
21 | | -The text below is the old ``distances.rst`` - that's not necessarily |
22 | | -relevant because a lot of it isn't that important, but some of it can be |
23 | | -rolled into the above. Use as needed then remove when done |
24 | | - |
25 | | -REMOVE THIS AFTER ABOVE SECTION IS WRITTEN |
26 | | -========================================== |
27 | | - |
28 | | -Coordinates can also have line-of-sight distances. If these are provided, a |
29 | | -coordinate object becomes a full-fledged point in three-dimensional space. |
30 | | - |
31 | | -The `~astropy.coordinates.Distance` class is provided to represent a |
32 | | -line-of-sight distance for a coordinate. It must include a length unit to |
33 | | -be valid.:: |
34 | | - |
35 | | - >>> from astropy.coordinates import Distance, ICRS, CartesianRepresentation |
36 | 37 | >>> from astropy import units as u |
37 | | - >>> d = Distance(770.) |
38 | | - Traceback (most recent call last): |
39 | | - ... |
40 | | - UnitsError: No unit was provided for Distance |
41 | | - >>> d = Distance(770., u.kpc) |
42 | | - >>> d |
43 | | - <Distance 770.0 kpc> |
44 | | - >>> c = ICRS('00h42m44.3s', '+41d16m9s', distance=d) |
45 | | - >>> c |
46 | | - <SkyCoord (ICRS): ra=10.6845833333 deg, dec=41.2691666667 deg, distance=770.0 kpc> |
47 | | - |
48 | | -Because `~astropy.coordinates.Distance` is a subclass of |
49 | | -`~astropy.units.Quantity`, in general a `~astropy.units.Quantity` with units |
50 | | -of length may be provided and it will automatically convert to a |
51 | | -`~astropy.coordinates.Distance`:: |
52 | | - |
53 | | - >>> ICRS('00h42m44.3s', '+41d16m9s', distance=770*u.kpc) |
54 | | - <SkyCoord (ICRS): ra=10.6845833333 deg, dec=41.2691666667 deg, distance=770.0 kpc> |
55 | | - |
56 | | -By default, `~astropy.coordinates.Distance` values must be non-negative. For some cosmologies, |
57 | | -distance metrics may become zero or negative; negative distance values are supported |
58 | | -by setting the ``allow_negative`` keyword argument to ``True``:: |
59 | | - |
60 | | - >>> d = Distance(-42., u.kpc) |
61 | | - Traceback (most recent call last): |
62 | | - ... |
63 | | - ValueError: Distance must be >= 0. Set the kwarg 'allow_negative=True' to |
64 | | - allow negative values. |
65 | | - >>> d = Distance(-42., u.kpc, allow_negative=True) |
66 | | - |
67 | | -If a ``distance`` is present, the coordinate can be converted into Cartesian |
68 | | -coordinates using the `~astropy.coordinates.CartesianRepresentation` class, or |
69 | | -the :attr:`~astropy.coordinates.CartesianRepresentation` attribute shorthand:: |
70 | | - |
71 | | - >>> cart = c.represent_as(CartesianRepresentation) |
72 | | - >>> cart.x |
73 | | - <Quantity 568.71288821656... kpc> |
74 | | - >>> cart.y |
75 | | - <Quantity 107.30093596881... kpc> |
76 | | - >>> cart.z |
77 | | - <Quantity 507.88990924863... kpc> |
78 | | - >>> cart = c.cartesian |
79 | | - >>> cart.x |
80 | | - <Quantity 568.71288821656... kpc> |
81 | | - |
82 | | -If a ``distance`` is not present, the Cartesian coordinates are still |
83 | | -available, but the point is interpreted as lying on the (dimensionless) |
84 | | -unit sphere:: |
85 | | - |
86 | | - >>> c2 = ICRS('00h42m44.3s', '+41d16m9s') |
87 | | - >>> c2.cartesian.x |
88 | | - <Quantity 0.73858816651502...> |
89 | | - >>> c2.cartesian.y |
90 | | - <Quantity 0.13935186489455...> |
91 | | - >>> c2.cartesian.z |
92 | | - <Quantity 0.65959728473848...> |
93 | | - |
94 | | - |
95 | | -.. note:: |
96 | | - |
97 | | - The location of the origin is different for different coordinate |
98 | | - systems, but for common celestial coordinate systems it is often |
99 | | - the Earth center (or for precision work, the Earth/Moon barycenter). |
100 | | - |
101 | | -This Cartesian representation can also be used to create a new coordinate |
102 | | -object through a `~astropy.coordinates.CartesianRepresentation` object:: |
103 | | - |
104 | | - >>> cart = CartesianRepresentation(x=568.7129*u.kpc, y=107.3009*u.kpc, z=507.8899*u.kpc) |
105 | | - >>> ICRS(cart) |
106 | | - <ICRS Coordinate: ra=10.6845796179 deg, dec=41.2691659084 deg, distance=769.99999759 kpc> |
107 | | - |
108 | | -Finally, two coordinates with distances can be used to derive a real-space |
109 | | -distance (i.e., non-projected separation):: |
110 | | - |
111 | | - >>> c1 = ICRS('5h23m34.5s', '-69d45m22s', distance=49*u.kpc) |
112 | | - >>> c2 = ICRS('0h52m44.8s', '-72d49m43s', distance=61*u.kpc) |
113 | | - >>> sep3d = c1.separation_3d(c2) |
114 | | - >>> sep3d |
115 | | - <Distance 23.056848146957... kpc> |
116 | | - >>> sep3d.kpc |
117 | | - 23.056848146957... |
118 | | - >>> sep3d.Mpc |
119 | | - 0.023056848146957... |
120 | | - >>> sep3d.au |
121 | | - 4755816315.663... |
| 38 | + >>> from astropy.coordinates.representation import CartesianRepresentation |
| 39 | + >>> car = CartesianRepresentation(3 * u.kpc, 5 * u.kpc, 4 * u.kpc) |
| 40 | + >>> car |
| 41 | + <CartesianRepresentation x=3.0 kpc, y=5.0 kpc, z=4.0 kpc> |
| 42 | + |
| 43 | +Representations can be converted to other representations using the |
| 44 | +``represent_as`` method:: |
| 45 | + |
| 46 | + >>> from astropy.coordinates.representation import SphericalRepresentation, CylindricalRepresentation |
| 47 | + >>> sph = car.represent_as(SphericalRepresentation) |
| 48 | + >>> sph |
| 49 | + <SphericalRepresentation lon=1.0303... rad, lat=0.6012... rad, distance=7.0710... kpc> |
| 50 | + >>> cyl = car.represent_as(CylindricalRepresentation) |
| 51 | + >>> cyl |
| 52 | + <CylindricalRepresentation rho=5.8309... kpc, phi=1.0303... rad, z=4.0 kpc> |
| 53 | + |
| 54 | +All representations can be converted to each other without loss of |
| 55 | +information, with the exception of |
| 56 | +`~astropy.coordinates.UnitSphericalRepresentation`. This class |
| 57 | +is used to store the longitude and latitude of points but does not contain |
| 58 | +any distance to the points, and assumes that they are located on a unit and |
| 59 | +dimensionless sphere:: |
| 60 | + |
| 61 | + >>> from astropy.coordinates.representation import UnitSphericalRepresentation |
| 62 | + >>> sph_unit = car.represent_as(UnitSphericalRepresentation) |
| 63 | + >>> sph_unit |
| 64 | + <UnitSphericalRepresentation lon=1.0303... rad, lat=0.6012... rad> |
| 65 | + |
| 66 | +Converting back to cartesian, the absolute scaling information has been |
| 67 | +removed, and the points are still located on a unit sphere: |
| 68 | + |
| 69 | + >>> sph_unit = car.represent_as(UnitSphericalRepresentation) |
| 70 | + >>> sph_unit.represent_as(CartesianRepresentation) |
| 71 | + <CartesianRepresentation x=0.4242... , y=0.7071... , z=0.5656... > |
| 72 | + |
| 73 | +Array values |
| 74 | +^^^^^^^^^^^^ |
| 75 | + |
| 76 | +Array `~astropy.units.Quantity` objects can also be passed to |
| 77 | +representations:: |
| 78 | + |
| 79 | + >>> import numpy as np |
| 80 | + >>> x = np.random.random(100) |
| 81 | + >>> y = np.random.random(100) |
| 82 | + >>> z = np.random.random(100) |
| 83 | + >>> car_array = CartesianRepresentation(x * u.m, y * u.m, z * u.m) |
| 84 | + >>> car_array # doctest: +SKIP |
| 85 | + <CartesianRepresentation (x, y, z) in m |
| 86 | + [(0.7093..., 0.7788..., 0.3842...), |
| 87 | + (0.8434..., 0.4543..., 0.9579...), |
| 88 | + ... |
| 89 | + (0.0179..., 0.8587..., 0.4916...), |
| 90 | + (0.0207..., 0.3355..., 0.2799...)]> |
| 91 | + |
| 92 | +Creating your own representations |
| 93 | +^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^ |
| 94 | + |
| 95 | +To create your own representation class, your class should: |
| 96 | + |
| 97 | +* Your class should inherit from the ``BaseRepresentation`` class. |
| 98 | +* Define ``__init__`` |
| 99 | +* Define a ``from_cartesian`` class method which should take a |
| 100 | + `~astropy.coordinates.CartesianRepresentation` object and |
| 101 | + return an instance of your class. |
| 102 | +* Define a ``to_cartesian`` method which should return a |
| 103 | + `~astropy.coordinates.CartesianRepresentation` object |
| 104 | +* Define a ``components`` property that returns a list of the names of the |
| 105 | + coordinate components (such as ``x``, ``lon``, and so on). |
| 106 | + |
| 107 | +In pseudo-code, this means that your class will look like:: |
| 108 | + |
| 109 | + class MyRepresentation(BaseRepresentation): |
| 110 | + |
| 111 | + def __init__(self, ...): |
| 112 | + ... |
| 113 | + |
| 114 | + @classmethod |
| 115 | + def from_cartesian(self, cartesian): |
| 116 | + ... |
| 117 | + return MyRepresentation(...) |
| 118 | + |
| 119 | + def to_cartesian(self): |
| 120 | + ... |
| 121 | + return CartesianRepresentation(...) |
| 122 | + |
| 123 | + @property |
| 124 | + def components(self): |
| 125 | + return [...] |
| 126 | + |
| 127 | +Once you do this, you will then automatically be able to call |
| 128 | +``represent_as`` to convert other representations to/from your representation |
| 129 | +class. |
0 commit comments