Loading retained release reports…
The original validator accepts NaN for inertia, damping and droop parameters. The proposed correction rejects those inputs. The other 21 cases match.
NaN means “not a number.” Both less-than and greater-than comparisons with NaN are false, so a range check alone can miss it.
| Parameter / input | Original build | Proposed correction | Observation |
|---|
These are retained software checks from pinned builds. The proposed correction is a separate source revision. Case comparison excludes recorded elapsed time.
Open your report
Choose JSON generated by the local checker. It stays in this tab; opening it does not rerun the model.
Drop a JSON report here
Report version 2 · up to 2 MB
Paste report JSON
Reading your report…
| Case | Observed return | Reported outcome | Checks |
|---|
The viewer checks the report’s structure and internal counts. Its native observations are supplied by the file’s producer.
Check a model on your machine
The released Python checker calls a trusted compiled library’s metadata and parameter validator. Use an isolated Python environment; Python 3.12 is the recommended starting point.
Replace the example library path with your own trusted OpenIBR build. Use .dylib on macOS, .so on Linux and .dll on Windows. Choose a new output filename, then open that JSON in .
What the checker observes
- Public exports, model metadata, parameter bounds and selected nonfinite policies.
- Validator returns, active diagnostics, same-instance rechecks and protected buffer bytes.
- Library, adapter and checker identities, with explicitly supplied build evidence.
Read the checks and adapter contract
Only metadata and parameter validation are called. The electrical model is not initialized or stepped. Run native libraries whose code and provenance you trust.