[vera-rubin]

Observing a Star with positive proper motion returns a negative right ascension rate. We get about -2.73785 mas/day instead of +2.73785 mas/day below, contrary to the documented position derivative:

from skyfield.api import SSB, Star, load
from skyfield.framelib import ICRS

t = load.timescale().tdb_jd(2451545.0)
star = Star(ra_hours=0, dec_degrees=0,
            ra_mas_per_year=1000, parallax_mas=1000)
print(SSB.at(t).observe(star).frame_latlon_and_rates(ICRS)[4].mas.per_day)

We now subtract the observer’s velocity from the star’s velocity in all three scalar and array branches. The regression test fails before the change and passes after, and all 49 existing tests in the relevant modules still pass.

— ChipLab