Work

Release03 reproducibility archive: Physics-Guided Neural–Numerical Identification of Reactivity and Thermal-Feedback Parameters in Coupled Nuclear Reactor Transients

Year: 2026
Type: dataset
Abstract: Reproducibility archive for the manuscript Physics-Guided Neural–Numerical Identification of Reactivity and Thermal-Feedback Parameters in Coupled Nuclear Reactor Transients (Journal of Computational and Cognitive Engineering). Contains the source code (numba, C and pure-Python solver back-ends), seeded configurations (master seed 20260828), row-level results and figure scripts that regenerate every table and figure: the 480-fit benchmark (30 Latin-hypercube truth vectors × 4 noise realizations × 4 power-noise levels), cluster bootstrap, discretization budget, filtered inverse-kinetics comparator with Savitzky–Golay sensitivity, neural/cold/multi-start initialization comparison, temperature-sensor bias, drift and availability study, observation-cadence study, AR(1) and known-constant stress tests, and inhour verification. A verification script checks all 411 numbers printed in the revised manuscript. All data are synthetic (six-group point kinetics coupled to a two-node thermal model). Code: MIT licence. Data and figures: CC BY 4.0.
Language: English
Field: Energy
Open Access status: green
OpenAlex ID: W7224112210
Cites:
Cited by:
Related to: 10