The corrected EC reference produces native heating and current profiles
We need freshly generated electron-cyclotron (EC) profiles to inspect the reference’s heating and current conventions before coupling or training. The Primer explains source coupling; terminology is in the Glossary.
The owner approved one run with the corrected plotting library. The original ITER numerical deck and GENRAY source remain unchanged. The run now reaches the normal end of the solver and produces readable native output: the corrected EC reference completes in 83.16 s with readable native profilesC-156.
We retained separate electron, ion and collisional heating channels and each signed current definition. Heating volume integrals and the toroidal current area integral agree with their matching stored totals to floating-point precision. Other current definitions retain their upstream totals separately. The single ray stops when its remaining power reaches the deck’s cutoff. Input hashes are unchanged, stderr is empty, and raw outputs, snapshots and accounting records have bound provenance.
The paired saved output uses an older GENRAY producer. Native coordinates and bin volumes/areas match, allowing a direct descriptive comparison. The heating and signed-current curves closely overlap at the displayed scale; the numerical differences are retained without imposing a parity threshold. Positive toroidal and negative parallel current remain distinct.
The saved file comes from upstream independently of our runner. Comparing it uses the same reader for both files, so this is a descriptive regression check and numerical accounting exercise. It is not independent physical validation.
Figure provenance
Tier A; generated 2026-10-08. Package 0.0.1, Python 3.11.7, JAX 0.10.2. Commit a152933; working tree dirty: true (draft documentation and figure generation pending commit). Configuration SHA-256: ef05555c511db86e44114583fef28ab74993394e0718bbe58e61c962356a5dd9. SVG SHA-256: 94ef577c3b934df1348de6a4180892d4a971ceb8d9d1221d7ac59e4d057c8538. Fresh and saved input SHA-256 prefixes: b61e2bb3295d93c5 and 77b5d9f85cced382. The reference used a single CPU; regenerating the figure reads its retained files. Full profile CSV; generator blog/figures/make_m3_genray.py.
The PostScript device now opens and closes without errors. Its output renders to a blank page, so this run does not establish plotted content or plotting parity. A separate comparison figure is generated directly from the native NetCDF profiles; it is not substituted for GENRAY’s plotting output.
The bounded numerical pilot is complete. Production launch, heating/current conversion, conservative radial mapping, fallback criteria and grouped sampling still need to be settled before labelling or training. The ITER pedestal boundary and matched transport comparison remain independent workstreams. Historical suite reports are retained for unchanged tests. The repository checkpoint has 56 focused checks pass in full and Tier-A-only modesC-157; these fresh results are combined with retained reports when generating the aggregate status. That aggregate is not a fresh full-suite run. M3 remains in progress.
Technical details → corrected EC numerical pilot. Decision → OQ-21.
Technical details → Surrogates › genray corrected pilot
Decisions → OQ-21