{"id":"54f32d56-12a5-4e66-8ca2-9a0896ad98c2","arxiv_id":"2412.13239","paper_version":1,"verdict":"ACCEPT","confidence":"HIGH","novelty_score":4.0,"correctness_risk":"low","formal_verification":"none","parameter_count":0,"one_line_summary":"Gemini now automatically reduces GHOST spectra with DRAGONS and delivers them to PIs, with 97% of science frames reduced successfully between March and November 2024.","lead":"This research note describes an automated service that reduces spectra from the GHOST spectrograph at Gemini South and delivers them to astronomers. It reports that over 4,500 spectra were successfully reduced and made available to researchers between March and November 2024.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Abstract's 'over 4500 spectra' does not match §3's reported 2095 science FITS files and 1711 calibration files; the headline count needs a clear definition or correction.","rationale":"The paper is a useful, well-scoped service announcement: it describes a reproducible pipeline, publishes the reduction scripts, gives transparent success rates, and explicitly warns PIs to inspect outputs for non-standard reduction needs. The reader's weakest assumption, default DRAGONS reduction quality, is indeed acknowledged in the text and is not, by itself, a fatal issue. I do not find a problem with the pipeline or the service's value proposition. However, the single most concrete factual assertion, 'over 4500 spectra have been reduced and made available to PIs,' is numerically inconsistent with the detailed accounting in Section 3. The paper reports 2095 science FITS files successfully reduced and 1711 calibration files, and the described product structure does not straightforwardly produce over 4500 spectra unless calibration products are counted as spectra, which is never stated. The '9028 files ingested into GOA' similarly cannot be reconciled with the reported product counts from the text alone. This is an internally checkable inconsistency in the headline claim, not a vague concern about quality. It does not undermine the existence or usefulness of the service, but it should be clarified or corrected before the central quantitative claim is accepted. I therefore recommend a CONDITIONAL verdict: accept the paper once the abstract count is reconciled with the data section or explicitly redefined.","tokens_in":2997,"tokens_out":8324,"duration_ms":79903,"concrete_test":"Query the Gemini Observatory Archive for GHOST reduced products (calibrated.fits and dragons.fits) ingested between 2024-03-20 and 2024-11-30, and count (a) combined spectra (dragons.fits), (b) all science reduced files, and (c) total science+calibration products. Compare each count with 'over 4500'; if (a) or (b) is below 4500 and calibration frames are not documented as 'spectra', the abstract overstates the headline number.","verdict_should_be":"CONDITIONAL","load_bearing_attack":"The central quantitative claim is the abstract's statement that 'over 4500 spectra have been reduced and made available to PIs.' Section 3 reports 2095/2165 science FITS files and 1711/1744 calibration files successfully reduced, and states that each raw science file yields two reduced products (calibrated.fits and dragons.fits) per exposure per arm. Depending on whether a raw FITS file is a single-arm frame or a two-arm exposure, the number of combined spectra (dragons.fits) is at most 2095 or 4190, both under 4500; counting the 1711 calibration products still gives at most 5901 total products, and calling calibration frames 'spectra' is not stated. The '9028 files ingested into GOA' also does not reconcile with any straightforward product count (e.g., 2095×2 + 1711 = 5901, or 2095×4 + 1711 = 10091). Thus the abstract's specific number is either undefined, inflated, or counting a different set than the data section. Because this is the paper's most concrete factual assertion and is externally checkable, it should be corrected or explicitly qualified before the central claim is accepted at face value.","agreement_with_reader":"disagree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This research note describes a service run by the US National Gemini Office with the Gemini Science User Support Department and the GHOST instrument team, which automatically reduces GHOST science and calibration frames using DRAGONS default options, performs a basic visual quality assessment, and ingests the reduced files into the Gemini Observatory Archive for access by principal investigators. The paper reports operational statistics: between March 20 and November 30, 2024, 97% of science (2095/2165) and 98% of calibration (1711/1744) FITS files were successfully reduced, with median reduction times under 45 minutes for calibrations and roughly 92 minutes per night for science data. The paper also lists the product types generated (calibrated.fits and dragons.fits) and notes that flux calibration and telluric corrections are not applied.","tokens_in":3223,"tokens_out":6900,"duration_ms":61584,"significance":"If the reported statistics are reliable, this service is a valuable community resource: it lowers the barrier for GHOST users to inspect their data, provides homogeneous reduced products, and creates long-term archive value once proprietary periods expire. The paper is appropriately cautious in advising PIs to verify whether default reductions meet their scientific needs, and it explicitly states that non-standard reduction steps may be required. The concrete success rates are a strength, as is the transparent description of the pipeline and its limits. The absence of any external validation of spectral quality, such as comparison with independent reductions or standard-star measurements, means the 'science-ready' phrasing in the introduction is stronger than the evidence presented, but this is acknowledged by the authors' own caveats.","major_comments":[{"comment":"The abstract's headline claim that 'over 4500 spectra have been reduced and made available to PIs' is not reconciled with the numbers in Section 3. The body reports 2095 successful science FITS files and 1711 successful calibration FITS files, and states that each raw science file yields two reduced files (calibrated.fits and dragons.fits) per exposure per arm. Depending on whether a raw FITS file corresponds to one arm or both, the resulting count of reduced science products is at most 4190, which is still below 4500 unless calibration products are counted as 'spectra,' a definition that is not stated. The reported total of 9028 files ingested into the GOA also does not match any straightforward sum of raw and reduced files. Please define precisely what quantity 'spectra' refers to (which product types, whether both arms are counted separately, whether calibration frames are included) and correct the numbers in the abstract and Section 3 to be mutually consistent.","section":"Abstract and Section 3"}],"minor_comments":[{"comment":"The sentence 'For each raw science file, there are two reduced files for each exposure in each arm' is ambiguous and should be rewritten to clarify whether a raw FITS file contains data from one arm or from both arms, as this directly affects the interpretation of the product counts.","section":"Section 3"},{"comment":"The author name 'William D. V acca' contains a stray space and should be spelled 'Vacca'.","section":"Author list"},{"comment":"The quality keywords 'QA=Undefined/Usable/Pass' are used without definition; please explain or cite documentation for these values so that the reader knows which frames are eligible for reduction.","section":"Section 2"}],"recommendation":"major_revision","confidential_remarks":"The paper is a straightforward operational report, and the core service appears to function as described. The main obstacle is the abstract's unsupported and internally inconsistent 'over 4500 spectra' claim, which is the paper's most visible quantitative assertion. If the authors provide a clear definition of what is counted and correct the numbers, I would be willing to accept the revised version."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Quick take: this is a clean, well-scoped research note about an operational data reduction service. The new content is the automated GHOST reduction pipeline and its throughput statistics, not a new reduction technique. It does what it says, with appropriate caveats about default options and the need for PIs to check their data.\n\nThe paper's real strength is its candor. It states plainly that reductions use default options, that flux calibration and telluric corrections are not applied, that QA is visual inspection, and that PIs should assess whether non-standard steps are needed. The success rates (97% science, 98% calibration) are concrete and plausible. This is the kind of operational detail the community needs.\n\nThe soft spot is the abstract's 'over 4500 spectra.' Section 3 reports 2095 science and 1711 calibration FITS files successfully reduced, and says each raw science file yields two reduced products per arm. Depending on how you count arms and products, 4500 doesn't obviously match 2095, 4190, or any straightforward product count. The '9028 files ingested into GOA' also doesn't reconcile cleanly. The number needs a definition (what counts as a 'spectrum') or a correction before the central claim is taken at face value. This is a minor but real fix, not a load-bearing flaw.\n\nThe other limitation is the lack of external validation of spectral quality—no comparison with manually reduced data, no quantitative metrics. But the paper doesn't oversell; it advises PIs to inspect outputs. For a research note, that's acceptable.\n\nI'd send this to a referee—it's a legitimate service announcement with operational data that high-resolution spectroscopists will want to know about. With a clarified count, I'd accept. The citation pattern is appropriate (instrument paper, DRAGONS, GOA, etc.).\n\nRecommendation: accept after minor revision.","headline":"A useful, honest service note on GHOST automated reduction; the headline '4500 spectra' count needs a definitional fix, but the paper is otherwise solid and worth publishing.","tokens_in":3745,"tokens_out":2615,"would_cite":false,"duration_ms":23043,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"An automated nightly pipeline has delivered more than 4,500 reduced GHOST spectra to principal investigators since March 2024, with 97% of science frames reduced successfully.","keywords":["GHOST spectrograph","automated data reduction","DRAGONS pipeline","Gemini South","observatory archive","high-resolution spectroscopy","queue observing","reduced spectra"],"falsifier":"Take a representative sample of raw GHOST frames already reduced by the service and re-reduce them with fully manual, program-specific calibrations; if a substantial fraction of the delivered spectra differ in measured line strengths, radial velocities, or continuum shape by more than the formal uncertainties, the delivered products would not qualify as science-ready for those programs.","tokens_in":2857,"feed_emoji":"🔭","tokens_out":9135,"duration_ms":81499,"temperature":0.7,"pith_summary":"This research note reports an operational result: a dedicated server at the observatory now reduces every GHOST calibration and science frame automatically, using the DRAGONS data-reduction pipeline with default options, and delivers the reduced spectra to the program's principal investigator. Between 20 March and 30 November 2024 the service reduced 97% of science frames (2095 of 2165) and 98% of calibration frames (1711 of 1744), and more than 4500 reduced spectra became available. The point of the effort is to let astronomers inspect and use their high-resolution spectra without installing reduction software, to speed up publication, and to build a uniform archive of reduced data for the community once proprietary periods end. The authors are careful to say the reduced files are not stacked or flux-calibrated, and that PIs should judge whether non-standard reductions are needed.","feed_headline":"Nightly pipeline delivers 4,500+ reduced GHOST spectra","feed_subtitle":"97% of science frames reduced automatically and available to principal investigators without any reduction work.","key_machinery":"The mechanism is a pair of autonomous reduction scripts, calibration_reduce.sh and science_reduce.sh, that run on a dedicated server and call the DRAGONS command-line interface, version 3.2.2. DRAGONS is the Gemini data-reduction software; in this service it performs bias subtraction, flat correction, wavelength calibration, sky subtraction, and barycentric correction. For each science exposure in each arm, the pipeline writes calibrated.fits (individual orders) and dragons.fits (orders combined, resampled onto a log-linear wavelength scale), and a basic visual inspection precedes ingestion into the observatory archive. Keeping reduced calibrations on the server allows science frames to be re-reduced with different calibrations if needed.","core_discovery":"The paper's central claim is that a fully automated, centrally run reduction service can keep pace with a queue-scheduled high-resolution spectrograph and hand every partner program usable spectra. The evidence is the reduction statistics: nightly runs process calibrations first and science frames after, with median processing times under 45 minutes for calibrations and about 92 minutes for science; the delivered products are bias-subtracted, flat-corrected, wavelength-calibrated, sky-subtracted, barycentric-corrected blue and red spectra, with orders combined by variance weighting onto a log-linear wavelength scale. The authors present this as a sustainable way to increase demand for the instrument and to create a legacy value in the archive, because the reduced files follow the same proprietary-period rules as the raw data and will eventually be community-accessible.","pith_inferences":["The paper's success statistics count frames that reached the archive; they do not by themselves demonstrate that the default-option spectra are scientifically equivalent to careful manual reductions, so a direct comparison would be the natural next test.","The model should transfer to other queue-scheduled spectrographs, but the occasional 80-hour processing nights show that the automation, not spare staff time, is what makes such a service sustainable.","Since flux calibration and telluric corrections are not applied, 'science-ready' in practice covers line-position and line-ratio work; programs needing absolute fluxes or telluric-sensitive measurements will still need their own reductions.","A community-visible log of reduction failures and their causes would turn the remaining 3% of failed science frames into a diagnostic tool for instrument health."],"forward_implications":["Principal investigators can open reduced spectra immediately, without installing DRAGONS or learning reduction details, lowering the barrier to first science results.","Because the same default options are used for every frame, the archive accumulates a homogeneous set of reduced GHOST spectra that will be community-accessible once proprietary periods expire.","Raw files and stored calibrations remain available, so any program can still be re-reduced with non-standard settings, and special cases such as zero-flux frames have a synthetic slit-view workaround.","The reported median reduction times (under 45 minutes for calibrations, about 92 minutes per night for science) indicate the service can keep up with normal queue operations."],"supporting_citations":[{"why":"Describes the GHOST instrument whose data the reduction service processes.","marker":"Kalari et al. 2024"},{"why":"Introduces DRAGONS, the pipeline that performs the nightly reductions.","marker":"Labrie et al. 2023"},{"why":"Supplies the specific DRAGONS v3.2.2 release used in the scripts.","marker":"Simpson et al. 2024"},{"why":"Documents the observatory archive into which the reduced products are ingested.","marker":"Hirst & Cardenes 2017"}],"fun_headline_variants":["GHOST pipeline delivers 4,500+ spectra automatically","Automated GHOST reduction serves 4,500+ spectra","GHOST's nightly pipeline hands PIs spectra with zero effort","Open-access GHOST spectra: 4,500+ reduced automatically","GHOST data pipeline automates 4,500+ spectrum deliveries"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The entire service depends on the assumption that DRAGONS default reductions, confirmed only by a basic visual inspection, are good enough for the variety of GHOST science programs; the authors themselves tell PIs to check whether non-standard steps are required.","fun_headline_variants_meta":{"raw":{"variants":["GHOST pipeline delivers 4,500+ spectra automatically","Automated GHOST reduction serves 4,500+ spectra","GHOST's nightly pipeline hands PIs spectra with zero effort","Open-access GHOST spectra: 4,500+ reduced automatically","GHOST data pipeline automates 4,500+ spectrum deliveries"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000128,"raw_usage":{"total_tokens":1036,"prompt_tokens":784,"completion_tokens":252,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":400,"completion_tokens_details":{"reasoning_tokens":165}},"tokens_in":400,"tokens_out":252,"duration_ms":2439,"temperature":1.0,"reasoning_tokens":165,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-11T13:23:56.610376+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Take a representative sample of raw GHOST frames already reduced by the service and re-reduce them with fully manual, program-specific calibrations; if a substantial fraction of the delivered spectra differ in measured line strengths, radial velocities, or continuum shape by more than the formal uncertainties, the delivered products would not qualify as science-ready for those programs.","supporting_citations":[{"cited_title":"2024, DRAGONS, 3.2.2, Zenodo, 10.5281/zenodo.13821517","cited_arxiv_id":null,"evidence_quote":"Supplies the specific DRAGONS v3.2.2 release used in the scripts."},{"cited_title":"2017, in Astronomical Society of the Pacific Conference Series, Vol","cited_arxiv_id":null,"evidence_quote":"Documents the observatory archive into which the reduced products are ingested."}],"review_version":1}