You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Review 160 reconstruction assessments and separate numerical coefficient
agreement from calibration replay, source benchmarks, data agreement,
author-workflow reproduction and recommendations. Update 88 scoped outcomes:
conditional parameter comparisons, registered-refit replays and a calculated
source-equation check, each with its explicit criterion and qualifications.
Reconstruct or recover missing conditional fitting uncertainties and ordered
covariance, including staged/shared-observation propagation and Farla RbBr/RbI.
Recover the omitted Tange thermal parameter errors from primary Table 2.
Keep verified printed precision separate from statistical uncertainty, retain
missing covariance as unknown, and preserve the existing Frost Pd use hold.
Preserve every published 0.17.0 material source field, coefficient, observation,
calibration reduction, default and use decision. Add complete frozen review
inputs, source-preservation checks and human-readable record-by-record reports.