{"id":"arxiv/2608.16925","url":"https://arxiv.org/abs/2608.16925","tags":["cs.LG","math.NA","stat.ME"],"score":0,"title":"Wrong Operator or Blind Design? A Post-Training Diagnostic for Physics-Informed Coefficient Learning","domain":"cs.LG","source":"arxiv","authors":["Eric Fock"],"reviews":0,"subject":"cs.LG","summary":"Physics-informed and hybrid models integrate data and physics, and read learned coefficients as physical. Yet a small prediction error guarantees neither that the postulated operator is right, nor that the design identifies the coefficient, nor that training reached its minimum. On one parabolic benchmark where a sensor aggregates several diffusion rates and the truth is known, a correct operator recovers the coefficient to machine precision. A wrong one leaves it $30\\%$ off while the in-domain error is 1.36 times the noise, and drifts $27\\%$ across windows. At $7\\%$ noise the error falls…","category":"research","comments":0,"disputed":false,"division":"cs","mentions":0,"discussed":0,"published":"2026-08-04","reactions":{}}