Implement IAU1980 nutation #23
Conversation
| @@ -35,7 +34,7 @@ pub fn mean_moon_sun_elongation(t: TDBJulianCenturiesSinceJ2000) -> Radians { | |||
| -0.00003169, | |||
| ], | |||
| ); | |||
| arcsec_to_rad_wrapping(arcsec) | |||
| arcsec_to_rad_two_pi(arcsec) | |||
There was a problem hiding this comment.
Renamed for consistency with other modular conversions I found.
| @@ -162,20 +161,6 @@ impl MeanLongitude for Uranus { | |||
| } | |||
| } | |||
|
|
|||
| const ARCSECONDS_IN_CIRCLE: f64 = 360.0 * 60.0 * 60.0; | |||
There was a problem hiding this comment.
Extracted to common module.
| // match ERFA outputs rather than any target threshold. | ||
| const THRESHOLD: f64 = 1e-11; | ||
| // match ERFA outputs rather than any target TOLERANCE. | ||
| const TOLERANCE: f64 = 1e-11; |
There was a problem hiding this comment.
Renamed for consistency with other tests.
There was a problem hiding this comment.
A less arbitrary definition based on machine epsilon would be:
const TOLERANCE = f64::EPSILON.sqrt(); # ≈ 1e-8
There was a problem hiding this comment.
The square root is probably a bit aggressive for this particular case. Some of the test outputs are on the order of e-6, and I'd hope to test for more than 3 decimal places of precision.
There was a problem hiding this comment.
From reading the Random ASCII post you sent me, it seems like the tolerance does generally have to be tailored to the use case.
| #[derive(Debug, Default, Clone, PartialEq)] | ||
| pub struct Nutation { | ||
| /// δψ | ||
| pub longitude: Radians, |
There was a problem hiding this comment.
For all names in this PR, I'll let me know whether the plain English names or the Greek shorthand is clearer to you as an astrodynamicist. (The plain English is clearest for when you don't have that experience, let me tell you! But we shouldn't adopt a naming convention that will annoy the target audience)
There was a problem hiding this comment.
My rule of thumb: Greek letters and one-letter variable names inside the code to stay closer to the source material (e.g. formula in the paper) but plain names in the API.
| @@ -113,10 +114,6 @@ pub fn cartesian_to_keplerian(grav_param: f64, pos: DVec3, vel: DVec3) -> Kepler | |||
| ) | |||
| } | |||
|
|
|||
| fn normalize_two_pi(a: f64, center: f64) -> f64 { | |||
There was a problem hiding this comment.
Moved to common module.
| #[derive(Debug, Default, Clone, PartialEq)] | ||
| pub struct Nutation { | ||
| /// δψ | ||
| pub longitude: Radians, |
There was a problem hiding this comment.
My rule of thumb: Greek letters and one-letter variable names inside the code to stay closer to the source material (e.g. formula in the paper) but plain names in the API.
| pub fn from_datetime(scale: TimeScale, dt: DateTime) -> Self { | ||
| Self::from_date_and_time(scale, dt.date(), dt.time()) | ||
| } | ||
|
|
||
| /// Returns the J2000 epoch in the given timescale. | ||
| pub fn j2000(scale: TimeScale) -> Self { | ||
| Self::from_raw(scale, RawEpoch::default()) | ||
| } | ||
|
|
||
| /// Returns, as an epoch in the given timescale, midday on the first day of the proleptic Julian | ||
| /// calendar. | ||
| pub fn jd0(scale: TimeScale) -> Self { | ||
| // This represents 4713 BC, since there is no year 0 separating BC and AD. | ||
| let first_proleptic_day = Date::new_unchecked(ProlepticJulian, -4712, 1, 1); | ||
| let midday = Time::new(12, 0, 0).expect("midday should be a valid time"); | ||
| Self::from_date_and_time(scale, first_proleptic_day, midday) | ||
| } | ||
|
|
||
| fn from_raw(scale: TimeScale, raw: RawEpoch) -> Self { |
There was a problem hiding this comment.
I am not entirely convinced about these API changes but we should revisit this later.
There was a problem hiding this comment.
I'm content for these to be part of the crate API. I can't speak to their usefulness to the outside world, but they make testing more convenient.
Implements nutation calculation for the IAU 1980 model, and provides the skeleton to calculate nutation for any given model, pending their implementation.
I had to extend and refactor a few parts of the time crate to support conversions to the inputs expected by the nutation algorithm. My expectation is that this will make implementing future models much easier.
I additionally extracted a few newly shared constants, types and functions to common modules.