I

iowarp/scientific-debugging

Copy page

Diagnoses stalled or cross-system failures and wrong, NaN, nondeterministic, or unexpectedly slow scientific results through falsifiable hypotheses across distinct fault classes, with evidence-cited verdicts before any fix. Not for designing benchmarks or pre-registered experiments; use experiment-protocol.