Scientific Validation¶
Astraeus has 21 registered scientific and engineering validation cases. They compare selected calculations and integrated workflows with published references, independent libraries, external-tool results, analytical calculations, and physical or accounting invariants.
This evidence supports the specific claims and operating envelopes stated below. Product certification, hardware qualification, and purpose-specific assurance require additional evidence for the intended configuration and use.
Evidence Portfolio¶
| Area | Cases | Evidence basis | Representative agreement |
|---|---|---|---|
| Orbit, frames, and ground geometry | 4 | Vallado SGP4, Astropy/IERS, NGA WGS-84, and independent topocentric geometry | Near-Earth SGP4 components within 2e-7 km and 2e-7 km/s; terrestrial-frame position within 0.1 km |
| Sampled access and coverage | 3 | Independent geometry and analytical reductions | Access/revisit expectations exact; weighting and fold invariants within 1e-12 |
| Attitude realization | 2 | Published UVEX maneuver timing and archived Sindri/NASA 42 telemetry | UVEX total duration within 1 s; calibrated nadir acquisition within 5 s |
| RF and link budgets | 3 | ITU-R, JPL, NASA NEN, and independent SI calculations | Free-space loss within 0.02–0.03 dB; published link metrics within 0.15 dB |
| Beams, routing, scheduling, interference, transport, and service | 6 | Analytical models, exhaustive compact topologies, and conservation invariants | Beam/route selections exact; capacity, backlog, delivery, and KPI accounting within declared tolerances |
| Integrated workflows | 3 | Independent end-to-end expectations plus published STK and OMNeT++ studies | All registered workflow acceptance limits pass |
The 18 component cases isolate important calculations. The three integrated cases then exercise normal CLI/RunSpec execution and verify the resulting artifacts and summary metrics.
Integrated Published And End-To-End Cases¶
| Scenario | Scale | Comparison | Headline result |
|---|---|---|---|
| Direct-service workflow | One satellite, two service points, three time samples | Independently calculated states, geometry, service, latency, byte accounting, and artifact contract | 50% sampled service coverage; 3 items delivered and 3 expired; expected artifacts and hashes present |
| Published global coverage | 72 satellites, 2,592-point grid, 24 hours | CBO Figure 10, calculated with Ansys STK | Latitude-curve RMSE 2.812 percentage points; area-weighted global error 1.552 percentage points |
| Published routed latency | 625 satellites, 751 one-second samples | OMNeT++ Figure 6 | 100% route reachability; mean round-trip latency error 0.105 ms; P95 error 0.034 ms |
The two external-tool comparisons reconstruct the published scenarios from the information available. Their agreement supports the recorded scenario and metrics; the claim remains scenario-specific rather than a general equivalence claim for STK or OMNeT++.
Reference Basis¶
The evidence set draws on:
- Vallado's SGP4 verification reference
- Astropy satellite-coordinate guidance and IERS Earth-orientation data
- NGA WGS-84
- the published UVEX slew model and archived Sindri acquisition telemetry
- ITU-R P.525-5, JPL link-budget material, and the NASA NEN X-band study
- the published CBO/STK and OMNeT++ constellation studies summarized above
Supported Claim Envelope¶
The current evidence supports bounded claims for:
- near-Earth SGP4 propagation, TEME/terrestrial transforms, and WGS-84 ellipsoid geometry;
- sampled access, revisit, equal-point coverage, represented-surface coverage, and geometric folds;
- reduced-order finite-slew attitude timing and availability;
- aggregate RF and link budgets, parametric beams, active-set interference, deterministic routing and scheduling, aggregate transport accounting, and sampled service KPIs;
- the three exact integrated workflows described above.
Evidence Boundaries¶
| Area | Outside the current evidence portfolio |
|---|---|
| Orbit and geometry | Deep-space propagation, general orbit-prediction uncertainty, terrain, refraction, local horizons, geoid heights, or higher-fidelity datum transforms |
| Access and service events | Continuous-time roots between samples or statistical guarantees beyond the tested cases |
| Attitude | Controller, actuator, disturbance, sensor, flexible-body, or full flight-dynamics equivalence |
| RF and antennas | Measured/full-wave patterns, detailed atmosphere and rain, polarization, modulation, BER, adjacent-channel leakage, or general interference equivalence |
| Network and scheduler | Packet protocols, stochastic traffic, congestion control, scheduler optimality, or standards conformance |
| Workflow breadth | Every stage combination, Console or AI-skill behavior, and a general runtime, memory, or supported-scale envelope |
The initial independent scientific-evidence portfolio centers on orbit, geometry, attitude, communications, transport, and integrated workflows. Power and propulsion currently rely on software-correctness and engineering-model coverage; their model pages state the applicable assumptions and intended use.
Qualification Use¶
Scientific validation applies to an exact Astraeus revision and its stated claim envelope. A release qualification should confirm that all registered cases pass on the candidate and that their assumptions and exceptions remain acceptable for the intended study.
Security, packaging, performance, licensing, support, and operational readiness are separate release decisions. See Product Boundaries And Support for the current product boundary.