Verification & Validation

This section only includes cases that are currently treated as scientific verification evidence in this repository. Examples that are merely smoke tests, qualitative regressions, or surrogate setups are intentionally excluded from the tables below even if related scripts still exist elsewhere in the tree.

The repository's capability matrix determines which solver families are currently allowed to support publication-grade claims. Stable claim-bearing solver features must have automated evidence in this section; provisional features may still appear elsewhere in the repository, but they are not elevated to the same claim status until their evidence ladder is complete.

Evidence cases in the validation manifest may also declare an explicit ladder stage, canonical entrypoint, and machine-readable summary requirement. The scientific evidence runner executes those entrypoints directly and records the resulting metadata for release reporting.

Verification vs. Validation

Verification asks whether the discretisation and implementation solve the intended equations correctly. These cases use manufactured solutions, exact solutions, or discrete invariants.

Validation asks whether the chosen equations reproduce external physical measurements. No experimental validation claims are currently made in this section, because the available cavity/torus surrogate examples are not yet rigorous literature-backed reproductions.

Code Verification

ExampleEvidence BasisMetricAcceptance
MMS ConvergenceManufactured solutionObserved convergence rates in multiple normsMeets script tolerance for second-order behaviour
Decoupled MMS ConvergenceManufactured solutionSeparate spatial and temporal convergence ratesMeets script tolerances for each refinement study
Euler MMS ConvergenceManufactured solutionL1 density error, pressure error, GCI ratioDensity rates and GCI remain within script tolerances
Poisson ConvergenceExact solutionL2 error and observed convergence rateMeets script convergence tolerance
Smooth AdvectionExact solutionL1 transport error and observed convergence rateMeets scheme-specific tolerance
Source Term ConvergenceManufactured solutionConvergence with source forcingMeets script tolerance
Flux BalanceDiscrete invariantResidual in assembled flux balanceMachine-precision imbalance
Conservation VerificationDiscrete invariantDrift in conserved totalsRemains within script tolerance
Species ConservationDiscrete invariantDrift in total and per-species quantitiesRemains within script tolerance
Coupling Null-Source IdentityCanonical non-coupled reference datasetMaximum conserved-state difference vs the SciML-backed hyperbolic solveRemains below the stated machine-precision threshold
Coupled Mass ConservationDiscrete invariantRelative drift in total mass and total momentum under cooling-source couplingRemains below the stated machine-precision threshold
Passive Scalar ConvergenceExact solutionL1 passive-scalar error and observed rateMeets script tolerance
GRMHD Flat-Space ReductionAsymptotic reductionGRMHD/SRMHD flux mismatch and round-trip errorRemains below stated thresholds
GRMHD Newtonian LimitAsymptotic reductionPrimitive-recovery error and static-atmosphere driftRemains below stated thresholds
MHD div(B) PreservationDiscrete invariantMaximum cellwise $\nabla\cdot B$Remains below stated threshold
AMR Smooth-Pulse ConvergenceExact solutionL1 density error and observed refinement rateError decreases monotonically and rates remain above the stated threshold
AMR Regridding ConservationDiscrete invariantMaximum relative drift in mass, momentum, and energyDynamic AMR reaches level 2 and all reported drifts remain below the stated threshold

Analytical Benchmarks

ExampleEvidence BasisMetricAcceptance
Sod Grid ConvergenceExact Riemann solutionL1 density error and convergence trendErrors decrease monotonically
Toro Riemann TestsPublished star-state valuesStar-region pressure errorEach case meets the script tolerance
MHD ConvergenceExact Alfvén-wave solutionL1 magnetic-field error and observed rateRates exceed the script threshold
Navier-Stokes ConvergenceExact Taylor-Green vortexLinf velocity error and observed rateRates exceed the script threshold
Taylor-Green KE DecayExact kinetic-energy decay lawRelative decay errorMeets the script tolerance
Porous Medium (Barenblatt)Exact self-similar solutionL2 error trend and minimum observed rateErrors decrease monotonically
Coupled Cooling ReferenceTrusted fine-grid coupled reference datasetL1 pressure error and observed refinement rateErrors decrease monotonically and rates exceed the stated threshold
AMR Reference TrackingTrusted structured-grid reference datasetL1 density error versus a uniform baseline plus compression ratioError decreases with AMR budget while active-cell compression stays below the stated threshold
SRMHD ConvergenceExact smooth-wave solutionL1 density error and observed rateRates exceed the script threshold
SRMHD Eigenmode ConvergenceLinearised exact eigenmodesSelf-convergence rate for each modeAll mode-wise rates exceed the script threshold
GRMHD ConvergenceExact smooth-wave solution in Minkowski spacetimeL1 density error and observed rateRates exceed the script threshold

Scope Notes

  • Tutorials and smoke tests are validated separately as executable documentation, not as scientific evidence.
  • Regression-style scripts that compare one numerical method to another without an external truth model are not listed here.
  • The AMR evidence listed here is deliberately scoped to smooth transported-pulse and conservative regridding cases; shock-dominated AMR tutorials remain outside the automated evidence ladder for now.
  • Experimental validation will be added back only when the implementation reproduces the referenced benchmark physics and compares against published data with explicit quantitative acceptance criteria.
  • Demoted or manually reviewed cases are tracked in the validation manifest and release report rather than being silently left in the tree as implied evidence.

References

  • P. J. Roache, Verification and Validation in Computational Science and Engineering, 1998.
  • ASME V&V 20-2009, Standard for Verification and Validation in Computational Fluid Dynamics and Heat Transfer, 2009.
  • W. L. Oberkampf and T. G. Trucano, "Verification and validation in computational fluid dynamics," Progress in Aerospace Sciences, 2002.
  • C. J. Roy, "Review of code and solution verification procedures for computational simulation," Journal of Computational Physics, 2005.
  • E. F. Toro, Riemann Solvers and Numerical Methods for Fluid Dynamics, 3rd ed., 2009.