simulation-validation-and-interpretation

name	simulation-validation-and-interpretation
description	Validation, verification, and interpretation workflow for simulation results in physics and engineering. Use when assessing convergence, uncertainty, physical plausibility, and decision relevance for FDTD, FEM, Monte Carlo, radiation transport, and medical-physics simulations.

Simulation validation and interpretation

Scope

Use this skill after runs complete to determine whether simulation results are trustworthy, explainable, and actionable.

Use for

• Verification of numerical correctness and convergence
• Validation against references, benchmarks, or experimental ranges
• Uncertainty decomposition and sensitivity analysis
• Physical interpretation of outputs and limitations
• Communicating whether conclusions are supported by evidence

V and V protocol

1. Verification (did we solve the equations right?)

• Check residual behavior and solver convergence.
• Run mesh or timestep refinement studies.
• Confirm conservation properties and invariants where applicable.

2. Validation (did we solve the right equations?)

• Compare against analytic limits, benchmark cases, or measured data.
• Quantify discrepancy metrics, not just qualitative overlap.
• Report where the model fails and plausible causes.

3. Uncertainty analysis

• Separate numerical, parametric, and model-form uncertainty.
• For stochastic methods, report confidence intervals and effective sample quality.
• For deterministic solvers, report discretization and solver tolerance effects.

4. Sensitivity and robustness

• Identify high-impact parameters and interactions.
• Test robustness to plausible perturbations in assumptions.
• Flag brittle conclusions that change under minor parameter shifts.

5. Interpretation and decisions

• State what is strongly supported, weakly supported, and unknown.
• Tie conclusions directly to validated metrics and uncertainty bounds.
• Provide next experiments or simulations needed to reduce key uncertainty.

Domain notes

• FDTD: include numerical dispersion and boundary absorber quality checks.
• FEM: include element-quality checks, conditioning concerns, and BC sensitivity.
• Monte Carlo and transport: include variance convergence, bias controls, and seed reproducibility.
• Medical physics: separate research interpretation from clinical decision support unless externally validated.

Reporting template

• Objective and modeled quantity.
• Verification evidence.
• Validation evidence.
• Uncertainty table with dominant contributors.
• Interpretation with confidence level.
• Actionable next steps.

Red flags

• No mesh or timestep study but strong quantitative claims.
• Benchmark mismatch explained only qualitatively.
• Uncertainty omitted from key decision plots/tables.
• Conclusions presented outside validated domain of the model.