Evaluation

Use the evaluation pages after training has produced an extrapolation_result.nf2.

• Exporting writes field cubes and derived quantities to VTK, NPZ, HDF5, or FITS.

• Quality metrics computes divergence, force-free, current-alignment, and magnetic-energy diagnostics.

Choose the output based on the next tool in your workflow:

Goal	Use	Notes
Inspect field lines or volumes in ParaView	nf2-export --format vtk	Best for visual quality checks and presentations.
Continue analysis in Python	nf2-export --format npz	Keeps arrays easy to load with NumPy.
Share structured data with other tools	nf2-export --format hdf5	Useful for larger runs and series.
Compare physical quality across runs	nf2-metrics	Reports divergence, force-free alignment, and energy diagnostics.
Inspect learned multi-height surfaces	nf2-export --format height	Requires a checkpoint with height mappings.

Evaluation

• Exporting
  • Export A Cartesian Result
  • Export A Spherical Result
  • Export A Series
  • Export Quantities
• Quality Metrics
  • Cartesian Quality Metrics
  • Spherical Quality Metrics