Validation¶

Contract¶

Validation is the fail-closed set of authoring, lowering, capability, and backend-legality checks between the Python surface and execution.

Governing equations¶

This page introduces no governing equation.

Symbols and SI units¶

Parameter units follow their owning pages; validation checks domain and finite-ness, not unit conversion.

Assumptions and validity¶

Mathematical validation cannot replace capability resolution; a structurally valid problem may still be rejected when its requested lane cannot execute it.

Python API¶

# %% Validation/provenance stage-first probe
import fullmag as fm

study = fm.study("api_contract_probe")
study.engine("fdm")
study.device("cpu", precision="double")
study.mode("strict")
study.objects.mesh.defaults(cell_size=(2e-9, 2e-9, 2e-9))
body = study.geometry(fm.Box(40e-9, 20e-9, 4e-9), name="film")
body.Ms = 800.0e3
body.Aex = 13.0e-12
body.m = fm.texture.uniform(1.0, 0.0, 0.0)
study.stages.add_run(stage_id="probe", until=1.0e-12)

ProblemIR¶

Lowering preserves the validated authored request; the planner resolves requested execution and records capability decisions.

Round-trip and failure semantics¶

Requested intent is the value authored by Python and preserved in ProblemIR; resolved execution is the planner or realization result. Validation errors identify the violated domain rule, and unsupported combinations are rejected explicitly rather than silently substituted.

Each rejection carries the failing constraint; nothing is silently converted to another interaction, backend, device, or precision.

Discrete realization¶

Validation spans the Python constructors, Problem.to_ir, and the planner capability contract.

Implementation mapping¶

Anchor: packages/fullmag-py/src/fullmag/model/problem.py (class Problem) plus packages/fullmag-py/src/fullmag/_validation.py for shared checks.

Validation¶

Validation behavior itself is covered by rejection-focused tests.

Limitations¶

Validation does not replace executed-device qualification; source presence and skipped tests are not parity proof.

Scientific bibliography¶

No physical model is introduced.

Control Room crosswalk¶

Status: The Control Room authors a study and lowers it to ProblemIR; direct Problem/IR editing is not exposed.

not implemented: frontend support for standalone Problem/ProblemIR authoring. See Control Room capability register for the support matrix and not implemented policy. Frontend source owner: apps/control-room/src/modules/inspector/panels/StudyInspectorPanel.tsx (StudyInspectorPanel).

Source-code index¶