Repository navigation
Docs: explain the shot-number calibration behind spectral ranges; move the ARTS2D benchmark guide to NOTES.md - #163
Conversation
…e the ARTS2D benchmark guide to NOTES.md Getting Started now explains that fit_rng values are compared against the calibrated wavelength axis (pixel * dispersion + offset), what happens when a range is partly or entirely off that axis, and that the calibration is chosen by shot number. Shots missing from tsadar/data/calibration.py fall back to other.detector_specs in the defaults deck, which may not be correct for the user's shot. The ARTS2D recovery benchmark page was development notes rather than user documentation. It is removed from the docs site and its content is kept unchanged (converted to Markdown) as a dated entry in NOTES.md. The README now points there. Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
![ergodic-reviewer[bot]](https://avatars.githubusercontent.com/u/74554135?s=60&v=4){kind=small}
There was a problem hiding this comment.
One non-blocking documentation correction: angular data do not honor the documented instrument-response fallback. The benchmark guide relocation preserves its instructions and artifact descriptions; this documentation-only change needs no new postprocessing path. Static review only; tests and the documentation build were not run.
|
|
||
| **These fallback values are generic and are probably not the correct calibration for your shot.** Spectral ranges chosen against the wrong axis can select the wrong part of the spectrum or fall off the axis entirely, and fitted parameters such as density and temperature depend directly on the wavelength calibration. If the fallback values are missing from the deck, a ``KeyError`` is raised instead. | ||
|
|
||
| If you see this warning, either add a calibration entry for your shot to ``tsadar/data/calibration.py``, or set the correct dispersion, offset, and instrument-response values for your shot in ``other.detector_specs`` (see :doc:`defaults`). Then re-check your ``fit_rng`` values against the new axis. |
There was a problem hiding this comment.
Clarify the angular-data exception to this remedy (non-blocking). The preceding paragraph includes angular data, but setting other.detector_specs.widIRF does not change the instrument response when fitting an angular shot. In tsadar/data/calibration.py, the tstype == "angular" fallback reads only EPWDispersion and EPWoffset; it unconditionally sets spect_FWHM_ele = 0.9 and ang_FWHM_ele = 1. Then prepare_data() in tsadar/data/prepare.py replaces the deck's widIRF with that returned dictionary. A user following this advice can therefore supply measured widths yet still fit with the hard-coded ones. Please scope the deck-based IRF remedy to temporal/imaging data and explain that angular IRF changes currently require updating get_calibrations().
There was a problem hiding this comment.
ARTS is still not really supported for users so this will be left as is and fixed once ARTS code paths settle into a more final form.
Summary
fit_rngvalues are compared against the calibrated wavelength axis (pixel × dispersion + offset), not raw pixels.tsadar/data/calibration.py. Shots not in that table fall back toother.detector_specsin the defaults deck, which is generic and may not match the user's shot. The new text says how to fix this.NOTES.md: the page was development notes (ISS140 design and preflight results) rather than user documentation.docs/source/arts2d_recovery_benchmark.rstand its table-of-contents entry.NOTES.md.NOTES.md.🤖 Generated with Claude Code