I downloaded the airless_Moon_dt.txt file from JPL Horizons so that
I could reverse engineer their DeltaT function. But keeping it in
the horizons directory caused unit test breakage. Moved to delta_t
directory so dtplot can still use it to make graphs.
I had to increase certain error tolerances in the unit tests.
Reworked the unit tests to make more sense by waiting until
each language step is done to check against each other.
That way I can run a single language step independently.
This bug has been there a long time, but I didn't notice it
until I started experimenting with larger Delta-T values.
Fortunately, the bug was in the unit test, not Astronomy Engine
release code. Calling NOVAS place(), I need to pass in Delta-T
expressed in seconds, not days.
Write eclipse/c_le_stats.csv that contains all the error
values for each lunar eclipse calculation. This helped
me figure out the main source of lunar eclipse error was
discrepancy with Delta-T values in the future.
Using some trial and error, I found that using 85 km instead of 65.4 km
for the thickness of the Earth's atmosphere results in better overall
fit with the test data.
Adding support for lunar eclipse calculations in the C code
caused me to tweak the values of the Sun and Moon radii.
This in turn caused unit test failures.
Made slight changes to the unit tests to get things passing again.
I found that lunar eclipse data is available for many centuries.
I downloaded the data for the years 1701..2200.
Wrote norm.py to extract and convert the parts I care about
into a format that will be much easier to parse in the unit
tests for all four languages.
Regenerate the normalized data from the 'run' script.
This way, I have documentation for where the data came from.
Using geocentric Moon instead of heliocentric Moon
gives more floating point precision for determining
the distance between the Moon and the Earth's shadow ray.
I figured out a formula that determines how far away
the Moon is from the center of the Earth's shadow.
This confirms the formula makes sense for a known
total lunar eclipse on May 26, 2021.
Increase type safety by making the enumerated type Body
derive from Enum rather than IntEnum, as recommended by
https://www.python.org/dev/peps/pep-0435/
Fixed places where I was treating Body values as integers.
Now when a Time object is evaluated and represented in
the Python interpreter, it results in a string of the form:
astronomy.Time(ut)
where ut is the numeric representation of the ut field.
This mimics the exact way such a Time value could be constructed.
That is, eval(repr(t)) results in a time value equal to t.
It turns out I was off by nearly 18 hours in the B1875 epoch.
This has a tiny effect on the orientation of the Earth's axis.
Instead of: ut = 1875-01-01T12:00:00.000Z
the correct epoch is: ut = 1874-12-31T18:12.21.950Z
See the comments in the Constellation functions in
each of the source files for more info.
Changed the units of the angles stored in the constellation
borders data so that the numbers can be represented more compactly.
This requires converting the numbers back at runtime, but this is
well worth the smaller size.
Instead of using decimal hours/degrees rounded to 4 decimal places,
I went back to the original constel.c and modified it to represent
both RA and DEC in degrees, and to round all values to the nearest
quarter arcminute. This seems closer to the original intent of the
constellation boundaries.
I'm using the HYG star database v3 from:
https://github.com/astronexus/HYG-Database
I compare the star constellations it reports against
what I calculate from the star RA/DEC it lists.
When I try this against all stars in the database, I
find 25 disagreements about which constellation contains
the star. Another person found 3 disagreements. See:
https://github.com/astronexus/HYG-Database/issues/21
For now, I'm testing only the stars brighter than mag 4.890,
which eliminates all the disagreements, and still gets me
over 1000 test cases.
Also, now I'm verifying ephemeris file and star database
checksums whether or not they have just been downloaded.
The idea is to catch corruption or unexpected changes
each time I run the unit test.
This unit test only exercises 8 different points.
I want to add a more thorough unit test soon, before
moving on to implementing the constellation finder in
the other supported programming languages.
