Engineering Notes / 2026
Evidence-led engineering notes from aerospace simulation, numerical verification, rocket mission analysis, and scientific machine learning.
The aircraft simulator reuses the same aircraft derivative across integration, trim, and the adverse-yaw study. Keeping one computational path helps analysis stay aligned with the simulation being investigated.
Matching documented FLIGHTv2 cases supports a bounded behavior-preservation claim for those test conditions. It does not, by itself, establish agreement with flight-test data or validate a real aircraft.
The Apogee demonstration uses synthetic motor and sample flight-log inputs. Its simulation and Monte Carlo outputs demonstrate the analysis workflow, not a hardware flight, certified envelope, or flight-readiness verdict.
The PINN framework compares representative cases with known behavior, including analytical solutions where available. Stronger claims need classical numerical baselines, multi-seed comparisons, and clearly scoped error measurements.
These notes summarize the present public record; the linked case studies remain the source of truth for implementation details and verification boundaries.
Follow the questions in Research, then open the reinforcement-learning laboratory for experiments, results, theory, and failure analysis.
Browse Research → Explore the RL Lab →