Import, export & HDF5
Choose a format for model input, result inspection or bounded large-result storage.
Source and release provenance · v0.2.11
docs/reference/qf_solver_api.md ↗Model inputs
The .inp importer rejects unsupported content rather than implying it has been interpreted correctly.
| Format / entry point | Boundary |
|---|---|
| JSON / load_model | Stable model workflow; supported JSON v1 schema |
| Gmsh / import_gmsh_model | Stable MSH 4.1 plus setup JSON entry point |
| Abaqus/CalculiX .inp / read_inp | Provisional bounded subset; not full-format compatibility |
Result exports
Use save_result for JSON, save_result_csv for result tables, save_result_vtu for ASCII VTU and save_audit_markdown for the audit. These are documented stable facade functions.
from qf_solver import save_result_vtu, save_result_csv
# result and model come from the validated calculation
save_result_vtu(result, model, "results/model.vtu")
save_result_csv(result, "results/tables", model)Family-aware HDF5
The opt-in schema-v1.0 mixed TET4/WEDGE6/HEX8 result path is EXPERIMENTAL_BOUNDED and requires h5py. The provisional reader can select families, fields and region_id. The writer requires source_sha, and a semantic digest helper is available. This is result storage, with no general restart, parallel-HDF5 or scalability claim.
Preserve the evidence
save_evidence produces a reproducible evidence bundle; verify_evidence checks fingerprints. The package ships lightweight traceability records. Heavy raw arrays and complete historical campaign output remain repository evidence.