Review script for paper "2023A&A...675A.102K" made with predictor "DataLinkerOptimizedRefine"

[abuonomo]

Created by Paper Data Linking Bot

Please review the download script and instrument details for the paper.

bibcode: 2023A&A...675A.102K predictor: DataLinkerOptimizedRefine (gpt-4o-mini) annotated pdf: view here (requires login)

Checklist

Download Script

• [x] Correct time ranges specified

• [x] Correct VSO instruments specified

• [x] Query returns good results or contains warning with reasoning if not

• [x] Physical observables omitted or correct

• [x] Wavelengths omitted or correct

• [x] Cadences omitted or correct

Instrument Details

• [x] Consistent with download_script.py

Metadata

• [x] Contains the bibcode for the relevant paper
• [x] VSO sources consistent with download_script.py

For more details, refer to the Contributing Guide.

[abuonomo]

Physobs are not correct. They are no adhering to the Phyobs in the VSO interface information. Possible reasons: 1) we are using gpt-4o-mini. Maybe resolved by just using gpt-4o 2) could maybe restructure how VSO interface info is included. For instance, we could do it the the instrument level and/or have pre-constructed instrument-level sentences about allowable values. "LASCO has Physobs X, Y, Z and wavelengths 1, 2, 3..." etc.

[abuonomo]

Full instrument coverage which is great. Maybe gpt-4o (including mini) partially resolves the long context issues with information being lost in the middle. More examples will hopefully elucidate.

[abuonomo]

CALLISTO not available in the VSO.

[abuonomo]

EDIT: actually the model got it correct and I got it wrong. Maybe all is fine as-is for solar cycle times, even with gpt-4o-mini

solar cycle start and end times are not correct. Maybe need an external module for correct solar cycle times and carrington rotation times.

[abuonomo]

EDIT: I was being silly and looking the max time on wikipedia, rather than just the start and end. My mistake. Orignal time provided by gpt-4o-mini was actually correct.

time_range_cycle_23 = (datetime.datetime(1996, 1, 1), datetime.datetime(2008, 12, 31))
time_range_cycle_24 = (datetime.datetime(2009, 1, 1), datetime.datetime(2019, 12, 31))

These time ranges appear to be correct given the paper, though they do not perfectly align with the solar cycles perfectly as listed on wikipedia.

from paper

from wikipedia