24b0ac3700
For #406 , add Time constructors from datetimes
...
This was previously an implicit power of `ts.utc()` which was more
difficult for users to find.
2020-07-17 09:41:48 -04:00
d23ce4e1ee
For #406 , mention more modern timezone sources
2020-07-17 08:18:50 -04:00
d2b105d6a0
Extend two-float precision to how UT1 is stored
2020-07-15 06:24:48 -04:00
b254335063
Fix #395 by making Time objects hashable
2020-07-02 12:47:07 -04:00
7e2fc1919f
Make matrix, vector multiply functions public
2020-06-12 14:18:08 -04:00
e62b0ca093
Update CHANGELOG and declare version 1.22
2020-06-08 10:46:59 -04:00
eda5feac32
Hide Time._nutation_angles_radians for now
...
I have still not settled on the API for offering alternative nutation
routines, so hiding this gives me the freedom to release later today
without making a public API commitment yet.
2020-06-07 17:09:13 -04:00
d1108c547f
Improve precisions of calendar TT and TAI times
2020-06-07 15:06:15 -04:00
ecbd5d5cff
Add JD fraction to TAI constructor, tdb_minus_tt()
...
And along the way, start simplifying docstrings that look a bit redundant.
2020-06-07 13:38:29 -04:00
cf836d095c
Move divmod responsibility from Time to Timescale
2020-06-07 13:10:03 -04:00
a6fa806131
Now that pyflakes runs in CI, fix its complaints
2020-06-07 12:57:47 -04:00
352f0c6122
Better approach for keeping TDB fraction precision
2020-06-07 12:12:51 -04:00
4f1bd9a64d
Fix how time equality calls time subtraction
2020-06-07 11:55:34 -04:00
47c0de9446
Complete a few timelib TODOs related to precision
2020-06-07 11:47:06 -04:00
ccdfdba649
Bring Time attribute documentation up to date
...
Also, add methods by which users can ask for the nutation and precession
matrices in case they are doing low-level operations.
2020-06-07 11:12:12 -04:00
b3cfd26d0d
Stop caching the t.P and t.N matrices
...
These 3×3 matrices consume 72 bytes of memory for each element in a time
vector, yet are never used by Skyfield once it has built the matrix M.
2020-06-07 06:25:43 -04:00
57f5ba6a1b
Move nutation angle attributes to correct section
2020-06-06 12:17:45 -04:00
82b6eefe4c
Remove old wonky _earth_tilt attribute
2020-06-06 12:12:30 -04:00
6be8e0937e
Switch how users select IAU 2000B nutation angles
...
But also provide support for the old pattern.
2020-06-06 12:03:08 -04:00
21175f3ad9
Another round of simplification thanks to radians
2020-06-06 11:06:26 -04:00
b892bc007a
Switch new build_nutation_matrix() to radian args
...
This eliminates conversions in both the caller and the function itself.
2020-06-06 10:45:44 -04:00
20f0af7b36
Switch to caching nutation angles as radians
2020-06-06 10:43:17 -04:00
82a60a95e1
Cache mean obliquity in radians, not arcseconds
...
This is the first step towards caching all nutation-related angles in
radians, to minimize the amount of post-processing necessary when they
are used.
2020-06-06 10:38:50 -04:00
7d2aa41341
Try new t.nutation_angles_arcseconds angle pair
2020-06-05 22:16:57 -04:00
240789d1cc
Break compute_nutation() free of _earth_tilt
2020-06-05 21:54:50 -04:00
d89ab7d22b
Break ecliptic_xyz() free of _earth_tilt
2020-06-05 21:25:51 -04:00
9eecf711ef
Bypass _earth_tilt for computing t.gast
2020-06-05 21:12:26 -04:00
f77fc0beaa
Push improved time precision into .observe()
...
Now that we represent time internally as two floats instead of one, our
light-time correction routine can be improved to generate far less noise
down below the micro-arcsecond level. That might not sound like much,
but — for example — over the ten seconds around the maximum elongation
of Venus, the elongation varies by only about 5e-5 mas! Jitter down in
the tens of nano-arcseconds is enough to make a solver think it’s found
several different seconds in which a maximum elongation occurred.
2020-06-04 15:27:31 -04:00
2d1e8b3761
Use builtin=True more consistently for timescales
...
It looks like http://maia.usno.navy.mil/ is down this morning, which
reminds me that documentation should be more consistent in loading
timescales from internal files unless illustrating explicitly how to
download up-to-date files.
2020-06-04 05:59:32 -04:00
4fe491bbfb
Add attributes year, month, etc to `t.utc` objects
2020-05-24 15:05:32 -04:00
0bab9296c7
Further whitespace normalization
2020-05-22 07:59:41 -04:00
ebb4901110
Try rescuing CI with a more stabilized repr()
...
Apparently the versions of NumPy on Travis CI have different ideas
about whether to put a space after the opening bracket.
2020-05-22 07:56:07 -04:00
1f8e66a42f
Switch to keeping only one whole number in a Time
...
Instead of splitting every time scale into two floats, let’s keep only a
single master `whole` number and then a series of fractions, each that
when combined with the whole gives the time in a particular scale. This
saves memory, keeps the attribute list smaller, is conceptually simpler,
and avoids the problem I just encountered with wanting a high precision
two-float version of `ut1` — `ut1_1` or `ut11` would look confusing to
others reading the code, and given that the UT1 timescale at least
includes a number, folks reading `tt1` and `tt2` might have thought
those were different timescales instead of simply two parts of a single
number. So I will be using `whole` and `fraction` consistently in the
code from now on.
2020-05-12 10:09:52 -04:00
aabb29b57a
Use two floats for the TDB Julian date as well
2020-05-10 18:45:32 -04:00
2baa5d0b06
Fix #293 by keeping two TT floats inside each Time
...
More work will be necessary to carry through the extra precision
everywhere, but this is a start that addresses the most visible
manifestation of the previously low precision.
2020-05-10 13:19:47 -04:00
982bc79867
Remove obsolete UTC year experimental method
2020-05-10 12:15:51 -04:00
688de2e0fd
Avoid repeatedly recomputing leap_reverse_dates
2020-05-10 10:03:26 -04:00
ec3c0f3d7f
Auto-round to minutes and seconds for strftime()
...
This eliminates an awkward `+=` maneuver that was preventing me from
improving the internal representation of time objects. Hopefully this
is not one of those magic conveniences I come to deeply regret later.
2020-05-10 08:50:57 -04:00
36e279ada8
Always use _mxm and _mxmxm for matrix multiplies
...
This gives the code greater consistency, and as a side effect also makes
it easier to search and find all matric multiplies.
2020-05-08 13:43:38 -04:00
f65450d783
Bring Time.utc() back into agreement with docs
2020-02-02 07:48:43 -05:00
5f867b9b25
Fix #335 by providing an actual weekday value
2020-02-02 06:32:27 -05:00
c8af2b2859
Document millisecond precision of datetimes
...
Minimally fix #293 by documenting the behavior that a user found
surprising, and make plans for improving it in the future.
2019-11-07 19:52:39 -05:00
348f03caa3
Constellation lookup function, which fixes #228
2019-10-09 15:36:36 -04:00
911360fb9f
Fix Python 2.6
2018-09-23 17:01:39 -04:00
db52f44a3a
Like Python datetime, allow another ISO delimiter
...
ISO date-times are so terrible to try to read using the default 'T'
delimiter. Offer a way to override it.
2018-09-23 16:48:40 -04:00
9bcb9d8323
Start explaining “almanac”, and fix a bug
2018-09-11 06:16:51 -04:00
6c2be20d31
More Almanac tweaks; add phase and phase angle
2018-09-09 21:39:26 -04:00
d2be5672e4
Remove redundant array from IAU2000B
2018-09-09 17:35:42 -04:00
141d996175
Revert experiment in slicing away nutation terms
...
It made things a bit faster, but it’s overly arbitrary; it will make
more sense to actually implement IAU2000B for real.
2018-09-09 16:33:45 -04:00
5d4ae0987e
Experiment: what if Time can choose nutation cost?
2018-09-09 16:23:48 -04:00
Brandon Rhodes