Skip to content

026-WP05-G05 V&V and Robustness Evidence

Status

026-WP05-G05 = PASS_WITH_LIMITATIONS

This report records the bounded execution batch for G05. READY means that a case has a controlled executable model; it does not mean that the capability is qualified. The official foundation gate 026-G05 is closed as PASS_WITH_LIMITATIONS, with bounded external coverage and no broader claim. continues to mean Modal / dynamic / harmonic maturity.

Controlled provenance

Item Value
Source SHA 2a27b4423965f79ae2f3359dae6b7d93d3564eb8
Source worktree clean
Captured at 2026-08-28T19:42:56.565693Z
Solver qf_solver 0.2.6a0
Registry digest ba2028115e08e1b02e5d773199ad08c2f1081b37e648d3c8216db59728b6d571
Result manifest SHA-256 431d393ecf63b67144e956ab06183fcd20e7d386bc635142676da30f57c7531f
Environment Windows AMD64, CPython 3.13.1, NumPy 2.2.6, SciPy 1.15.2

Execution summary

The batch converted 50 planned definitions into executable cases, increasing the registry from 10 to 60 READY cases and leaving 120 definitions PLANNED. All 50 G05 cases were executed by the controlled runner:

Result Count
PASS 47
EXPECTED_FAILURE 3
FAIL 0
BLOCKED 0

The three expected failures are deliberately invalid inverted-TET4 geometry cases and all returned the INVALID_ELEMENT contract. The batch completed in 3.338 seconds of summed case wall time.

Coverage of existing routes

Family PASS Expected failure
Linear solids (LIN) 8 0
Shell / beam / discrete (SHL) 5 0
Modal (MOD) 5 0
Transient dynamics (DYN) 6 0
Harmonic (HAR) 5 0
Small-strain J2 (J2) 6 0
Geometric nonlinear (GNL) 3 0
Linear buckling (BUC) 3 0
Frictionless contact (CON) 4 0
Adversarial (ADV) 0 3
Scaling routes (SCL) 2 0

The execution exercised 22 linear-static, 5 modal, 6 transient, 5 harmonic, 6 nonlinear-static, 3 geometric-nonlinear and 3 linear-buckling routes. Variants use only maintained public example models plus declared deterministic load, analysis or material overrides recorded in case_registry.json.

Verification boundary

The evidence demonstrates executable coverage, deterministic result manifests and expected-failure handling. It does not, by case count alone, establish analytical accuracy, mesh convergence, external correlation or a new maturity claim. Those remain subject to the applicable campaign oracle and official gate. No FEM formulation, public solver API or tolerance was weakened.

Checks

  • Framework contracts: 5 passed.
  • Ruff: PASS.
  • compileall: PASS.
  • git diff --check: PASS.

The reproducible raw result directory is intentionally excluded from normal source history; the compact manifest digest and this summary are the archived evidence.