domains / families / research / jse-tju-robustness-reproducibility

jse-tju-robustness-reproducibility

separate

Use when stress-testing and packaging the reproducibility evidence for 《系统工程学报》 (Journal of Systems Engineering, Tianjin University), including parameter sensitivity, initial conditions, alternative models and metrics, extreme scenarios, Monte Carlo repetition, random seeds, software versions, data and code statements, and explicit failure boundaries.

When to use

• stress-testing reproducibility evidence
• parameter sensitivity analysis
• initial conditions and extreme scenarios
• alternative models and metrics
• Monte Carlo repetition and random seeds
• software versions and environment fixation
• data and code statements
• explicit failure boundaries

When not to use

• masking missing main validation
• mechanically floating all parameters by same percentage
• keeping only seeds that support main conclusion
• reporting only average improvement without failure rates
• using local absolute paths or unrecorded data cleaning
• claiming data available on request without verification path
• writing conclusion reversal as numerical anomaly

Skill page

Source