A team writes an orbital-drag code to predict the peak deceleration of a small…
PHYSICS · Multi-part · Fermi University Physics · Computational Physics · Research software practice, verification and validation · Research software practice, verification and validation
Scientific Computing for Physicists, Ch Are they the right equations? Solution verification, benchmarks and validation
Problem
A team writes an orbital-drag code to predict the peak deceleration of a small satellite during atmospheric reentry. The code has been checked only by confirming it runs without crashing and gives plausible-looking numbers; no order-of-accuracy test, no comparison against an analytic two-body or drag-free orbit, and no manufactured-solution check has been carried out. The code predicts a peak deceleration of 8.42g; telemetry from an actual reentry gives 8.40±0.15g (one standard deviation). The team writes in their report: "since our prediction lies inside the telemetry error bar, our code is verified as well as validated."
((a)) Classify what kind of evidence the 8.42g versus 8.40±0.15g comparison actually is.
((b)) Explain, with reference to a concrete failure mode, why this comparison cannot by itself establish that the code solves its equations correctly, and state what would be needed to close that gap.