baram2584/FESADev
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity
Files
26939140dae4fc83746d05d4fcf48085462ef83b
FESADev/phases/shell-surface-pressure-load/step6.md
T

8.6 KiB
Raw Blame History

Step 6: compare-pressure-reference

과거 개발 계획이다. 아래 작업·상태·예시는 현재 실행 지시가 아니다. 현재 구현과 검증 한계는 기능 요약을 따른다.

실행 범위와 상태 소유권

이 Step's single responsibility is to add deterministic comparison support between FESA schema-1 factor-one HDF5 and the exact stored shell-pressure displacement CSV using the unchanged common tolerance. It does not generate reference data and does not write final Gate 6 reports; Step 7 owns closure reports.

Complete RED, expected failure, minimal GREEN and focused VERIFY here, then end so automatic Stop owns exactly one whole-project full verification. Update only Step 6's status and one-line Codex-owned payload. Executor owns all timestamps, commits, retry/advancement, top-level status and output JSON.

읽어야 할 파일

  • AGENTS.md
  • ../../docs/solver-core/README.md
  • ../../docs/solver-core/README.md
  • ../../docs/solver-core/README.md
  • ../../docs/solver-core/README.md
  • ../../docs/solver-core/README.md
  • .codex/hooks.json
  • ../../docs/solver-core/README.md
  • ../../docs/shell-surface-pressure-load/README.md
  • ../../docs/shell-surface-pressure-load/README.md §§6.2–6.3
  • ../../docs/shell-surface-pressure-load/README.md
  • ../../docs/shell-surface-pressure-load/README.md §8
  • ../../docs/shell-surface-pressure-load/README.md
  • All prior implemented/test files from Steps 1–5, especially tests/integration/analysis/shell_surface_pressure_nonlinear_test.cpp
  • tests/reference/reference_tolerance_policy.h
  • tests/reference/reference_tolerance_policy.cpp
  • tests/reference/reference_tolerance_policy_test.cpp
  • tests/reference/nonlinear_static_reference_comparison.h
  • tests/reference/nonlinear_static_reference_comparison.cpp
  • tests/reference/nonlinear_static_reference_cases_test.cpp
  • tests/reference/mitc4_reference_comparison.h
  • tests/reference/mitc4_reference_comparison.cpp
  • tests/reference/mitc4_reference_cases_test.cpp
  • tests/CMakeLists.txt
  • Read-only exact artifacts listed below

Exact read-only artifact contract

  • Required input: reference/shellNL-meshfine-pressure/shellNL-meshfine-pressure.inp, 56,665 bytes, SHA-256 53718ccc9b41c23fcc1b509a51d66f1347e87aca43a74f1d6cec4d2085f4b90e.
  • Required comparison CSV: reference/shellNL-meshfine-pressure/shellNL-meshfine-pressure displacements.csv, 111,182 bytes, SHA-256 2fe53f98eac9f04779100721c3edf8049eb1dbfe733ebcbd819aeeed03b81198.
  • Optional review-only reaction CSV: reference/shellNL-meshfine-pressure/shellNL-meshfine-pressure reactions.csv, never verdict-changing.
  • Excluded stress CSV: reference/shellNL-meshfine-pressure/shellNL-meshfine-pressure stresses.csv, no comparison family.

Never rename, rewrite, normalize, repair, regenerate or reorder any artifact. Never run Abaqus or another reference solver. Run FESA only on a byte-identical copy under 과거 검증 자료(정리됨).

소유 파일과 interface boundary

  • Create test first: tests/reference/shell_surface_pressure_reference_test.cpp
  • Modify: tests/reference/nonlinear_static_reference_comparison.h
  • Modify: tests/reference/nonlinear_static_reference_comparison.cpp
  • Modify: tests/CMakeLists.txt

Preserve existing public operations and extend their per-case value contract minimally:

Result<std::vector<ArtifactSnapshot>> ArtifactCheck(
    const NonlinearReferenceCase& reference_case);
Result<std::filesystem::path> CopyInputToExecutionDirectory(
    const NonlinearReferenceCase& reference_case);
Result<GeneratedRunArtifacts> RunFesaApplicationOnCopiedInput(
    const NonlinearReferenceCase& reference_case,
    const std::filesystem::path& copied_input_path);
Result<NonlinearReferenceComparisonReport> CompareHdf5ToReferenceCsv(
    const NonlinearReferenceCase& reference_case,
    const std::filesystem::path& hdf5_results_path);
NonlinearReferenceClassification ClassifyReferenceComparison(
    const NonlinearReferenceComparisonReport& report);

Represent required input/displacement, optional reaction and excluded stress disposition as explicit per-case value fields. Do not change reference_tolerance_policy.*.

RED

Write/register first:

  • ShellSurfacePressureReference.AuditsExactRequiredArtifactsAndIgnoresOptionalReviewArtifacts
  • ShellSurfacePressureReference.SelectsUniqueSchemaOneFactorOneAcceptedFrame
  • ShellSurfacePressureReference.Projects729WideRowsTo2187LogicalRowsPerFamily
  • ShellSurfacePressureReference.MatchesByInstanceNodeLabelAndComponentNotRowOrder
  • ShellSurfacePressureReference.RejectsSchemaIdentityCardinalityMissingExtraDuplicateAndNonfiniteBeforeTolerance
  • ShellSurfacePressureReference.ClassifiesTranslationBlockingAndRotationWarningOnly
  • ShellSurfacePressureReference.ExcludesReactionAndStressFamiliesFromVerdict
  • ShellSurfacePressureReference.UsesUnchangedCommonScaleNearZeroRelativeAndRmsPolicy
  • ShellSurfacePressureReference.WritesDeterministicComparisonReport

Run before comparator implementation:

과거 도구 실행·출력 지시는 정리했으며, 현재 검증 방법은 기능 요약을 따른다.

Expected RED: current nonlinear registry has no pressure case/disposition and assumes a legacy artifact portfolio, so the exact two-required-file contract, schema/frame projection or 729/2187 cardinality fails. Record command, exit, duration and tail. Unexpected pass is evidence-gap; unrelated failure is regression.

Minimal GREEN

  • Snapshot exact required artifacts before run, byte-copy only the input into the build tree, and run FESA application only.
  • Require HDF5 schema 1, exact Step-1, a unique accepted frame with load_factor==1/is_factor_one, /model/nodes, and existing nodal displacement.
  • Join HDF5 nodes and CSV by (instance_name="Part-1-1", source node label, component), never row order/internal index. Expand 729 wide rows to 2,187 U1/U2/U3 logical rows and 2,187 UR1/UR2/UR3 rows.
  • Reject schema/frame/identity/cardinality, missing, extra, duplicate and nonfinite rows before tolerance.
  • After prechecks only, call unchanged EvaluateReferenceTolerance: S=max(abs(reference)); near-zero band 0.01*S; near-zero absolute row error <=0.01*S; otherwise relative <=0.05; RMS(error)/S<=0.01; zero-scale requires FESA exact zero.
  • Translation is blocking; rotation is warning-only. Reaction creates no numerical comparator family and is review-only; stress is excluded.
  • Produce deterministic row, family metric, warning and classification order. HDF5-derived CSV, if any, is debug-only and not official output.

VERIFY and Acceptance Criteria

Use 과거 검증 자료(정리됨) first if present; otherwise:

과거 도구 실행·출력 지시는 정리했으며, 현재 검증 방법은 기능 요약을 따른다.

위 targeted build, focused pressure/reference regressions와 nonzero discovery evidence를 기록한 뒤 Step status/summary를 갱신하고 agent turn을 종료한다. Implementation Agent는 target 없는 whole-project build 또는 unfiltered full CTest를 실행하지 않는다. Automatic Stop hook이 exactly one whole-project MSVC x64 Debug build/full CTest를 수행한다.

Acceptance: exact input/CSV size+hash unchanged before/after; 729 unique node keys; 2,187 rows in each U/UR family; all prechecks precede tolerance; common policy unchanged; U blocking and UR warning-only; reaction/stress dispositions exact; deterministic report; focused tests pass with nonzero discovery and no new warning/runtime staging failure; automatic Stop's exactly one whole-project MSVC x64 Debug build/unfiltered CTest passes.

불변식, 실패 처리와 증거

  • Artifact/precheck failure returns a failed classification without tolerance evaluation or artifact mutation.
  • Copied input/generated HDF5/log live only under build tree.
  • Record pre/post snapshots, copy/generated paths, selected frame identity, cardinalities, family scale/row/RMS/worst row, warnings, classification and RED/GREEN/VERIFY command evidence.
  • Failure classification: configure | compile | link | test | reference-comparison | artifact | harness | environment | upstream-contract | regression | evidence-gap.

Trace: FESA-REQ-SHELL-PRESSURE-017–019, 021–024; io §8; reference-model §§3–8; numerical-review external acceptance.

금지사항

  • Do not generate/edit/rename reference files, change tolerance, promote reaction/stress, add metadata requirements or run Abaqus.
  • Do not modify solver production unless failure is returned through Coordinator to the owning implementation Step.
  • Do not write final reports yet, create output JSON, invoke Executor, advance Step 7, commit or push.
Reference in New Issue View Git Blame Copy Permalink