Ephemeris Generation for Services¶

Problem¶

You need a repeatable service endpoint that transforms object identifiers and observation requests into ephemerides with uncertainty context.

Implementation Options and Tradeoffs¶

Propagate first, then generate ephemerides: Better for reuse when many observer requests share the same propagated states.
Direct ephemeris generation from initial orbits: Simpler request path, potentially more repeated propagation work.

Runnable Example¶

import numpy as np

from adam_core.orbits import Orbits
from adam_core.observers import Observers
from adam_core.orbits.query import query_horizons
from adam_core.time import Timestamp
from adam_assist import ASSISTPropagator

request_object_ids: list[str] = ["Apophis", "Bennu", "Eros"]
t0: Timestamp = Timestamp.from_mjd([60000.0], scale="tdb")
request_times: Timestamp = Timestamp.from_mjd(
    t0.mjd().to_numpy() + np.arange(0, 20), scale="tdb"
)

orbits: Orbits = query_horizons(request_object_ids, t0)
observers: Observers = Observers.from_code("I11", request_times)

propagator = ASSISTPropagator()

# Pattern A: direct output for one request.
ephemeris = propagator.generate_ephemeris(
    orbits,
    observers,
    predict_magnitudes=True,
    predict_phase_angle=True,
    chunk_size=100,
    max_processes=4,
)

# Pattern B: reusable propagated cache for multiple observer sets.
propagated = propagator.propagate_orbits(orbits, request_times, max_processes=4)

print(ephemeris.to_dataframe().head())

When to Use This Pattern¶

Use this when building observatory planning APIs, nightly scheduling pipelines, or batch prediction systems that need deterministic behavior.

Notes¶

This pattern treats these objects as propagated targets. If your use case is about perturbing-body states (for example Earth, Moon, planets), use adam_core.utils.get_perturber_state and origin codes instead.