google-deepmind / google-deepmind/torax
Failure to map sources when loading IMAS core_sources
- Dominant language
- Python
- Stars
- 721
- Forks
- 145
- Avg merge
- 2d 14h
- Merged PRs (30d)
- 49
Description
In TORAX 1.4.1, I thought we supported loading from core_sources.
However, when I load I get:
```
2026-06-22 11:21:20,425 [ INFO] IMAS Source total does not have a corresponding TORAX source model. Skipping this source.
2026-06-22 11:21:20,425 [ INFO] IMAS Source auxiliary does not have a corresponding TORAX source model. Skipping this source.
2026-06-22 11:21:20,425 [ INFO] IMAS Source nbi does not have a corresponding TORAX source model. Skipping this source.
2026-06-22 11:21:20,426 [ INFO] IMAS Source lh does not have a corresponding TORAX source model. Skipping this source.
2026-06-22 11:21:20,428 [ INFO] IMAS Source bootstrap_current does not have a corresponding TORAX source model. Skipping this source.
2026-06-22 11:21:20,428 [ INFO] IMAS Source radiation does not have a corresponding TORAX source model. Skipping this source.
2026-06-22 11:21:20,428 [ INFO] IMAS Source synchrotron_radiation does not have a corresponding TORAX source model. Skipping this source.
2026-06-22 11:21:20,428 [ INFO] IMAS Source cold_neutrals does not have a corresponding TORAX source model. Skipping this source.
2026-06-22 11:21:20,428 [ INFO] IMAS Source charge_exchange does not have a corresponding TORAX source model. Skipping this source.
```
Most of these are expected, except from `charge_exchange` and `radiation` and `synchrotron_radiation` - we do have those in TORAX, so it should be possible to load them. Or am I misunderstanding the choice made in the mapping?
@MateoBell , @Nush395
Contributor guide
Research direction
Start by tracing the IMAS core_sources loading and mapping logic for charge_exchange, radiation, and synchrotron_radiation. Compare those source names with the corresponding TORAX source models and existing mapping choices; done means supported sources load instead of being skipped, while expected unsupported sources retain their current behavior.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- python
- Domain
- backend
- Issue type
- Bug
- Difficulty
- 3/5
- Estimated time
- 1-2 days
- Activity status
- Quiet
- Clarity
- Mostly clear
- Newbie friendliness
- 55/100