{"id":"b4e901f0-6fc5-4643-8761-8caf5cd1699e","arxiv_id":"1909.00070","paper_version":1,"verdict":"UNVERDICTED","confidence":"MODERATE","novelty_score":4.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":4,"one_line_summary":"ATLAS is a proposed 1.5m infrared space telescope that would use digital micromirror arrays to obtain spectra for hundreds of millions of galaxies and thousands of Kuiper Belt objects.","lead":"This paper proposes a NASA space telescope mission, ATLAS, that would take infrared spectra of about 200 million galaxies, plus stars and small solar system bodies. It argues the mission is technically feasible within a probe-class budget and would answer major questions about galaxy evolution, cosmic acceleration, the Milky Way, and the Kuiper Belt.","discovery_kind":"new_method","skeptic_critique":{"model":"deepseek-v4-flash","headline":"Wide survey time budget is the load-bearing assumption: the paper gives no exposure-time calculation, and using its own data-rate and multiplex numbers the 183M-galaxy survey leaves little or no room for the other stated surveys.","rationale":"The proposal's strongest claim is that the complete survey program fits within a 5-year mission. Everything downstream, 200M redshifts, BAO/RSD forecasts, Milky Way structure, and KBO compositions, depends on reaching a line flux of 5e-18 erg/s/cm2 over 2000 deg2 with a 1.5m DMD spectrograph. The white paper does not show this; it defers to Wang et al. (2019a). I agree with the reader's weakest assumption. My review sharpens it by checking the paper's own internal budget: Table 2's 170 samples/day over five years corresponds to about 155 Ms of on-sky time. The 183M Wide targets at 6,000 multiplex require about 30,500 exposures, so the average exposure is capped near 5,000 s. A rough exposure-time estimate for the stated line flux, aperture, throughput, and H4RG dark current gives a required exposure in the 4-7 ks range at 1.2-2 um, i.e., right at the cap, leaving nothing for the Medium/Deep, Galactic Plane, and KBO programs. Thus the central claim is not merely unverified; it is close to the boundary of being internally inconsistent. The concrete test, a full exposure-time calculation at representative wavelengths folded into the mission schedule, would decide. I keep the verdict UNCHANGED because UNVERDICTED is the correct status: the text as written does not permit a positive or negative conclusion. The paper does have real supporting evidence, including DMD environmental testing, a JPL TeamX cost study, the ISCEA pathfinder selection, and companion papers, but none of that evidence addresses the missing sensitivity and survey-time budget.","tokens_in":13742,"tokens_out":19987,"duration_ms":187228,"concrete_test":"Run a standard emission-line exposure-time calculation for the Wide survey using the paper's stated parameters: 1.5m primary with 3.7% central obscuration, R=1000 prism spectroscopy, 0.75 arcsec DMD slitlets, 25-30% end-to-end throughput, H4RG dark current below 0.01 e-/s/pix, 5e read noise with 16 samples, and L2 zodiacal background, to reach S/N=5 on a 5e-18 erg/s/cm2 emission line at 1.2, 2.0, and 3.5 um. Then multiply by 30,500 exposures (183M targets / 6,000 multiplex) and add the stated durations of the other surveys (0.4-year Galactic Plane, 3,000 x 2,500 s KBO pointings, and nominal Medium/Deep exposures). If the total exceeds the roughly 155 Ms available from Table 2's 170 samples/day over 5 years, the paper's central claim is not supported as written. Alternatively, verify that Wang et al. (2019a) provides this exact calculation and that it closes the budget.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The central claim is that ATLAS can complete the three tiered surveys plus the Galactic-plane and KBO programs in a 5-year mission. That claim rests on an unstated exposure-time budget: no integration time, throughput, or completeness calculation appears in this white paper; the reader is referred to Wang et al. (2019a). The mission's own numbers make this the critical assumption. From Table 2, the observing cadence is about 170 samples per day over 5 years, i.e., roughly 155 Ms of on-sky time (85,000 s/day). The Wide survey alone requires 183M targeted spectra at a multiplex of 6,000, i.e., about 30,500 exposures. If a 5e-18 erg/s/cm2 line at 2 um with a 1.5m aperture, 25% throughput, and H4RG dark current below 0.01 e-/s/pix needs roughly 4,000-6,000 s for S/N near 5 (with the H-alpha line at 1.2 um needing more because each photon carries less energy), the Wide survey consumes roughly 130-150 Ms. Adding the explicitly quoted 0.4-year Galactic Plane survey (about 12.6 Ms), the 100 deg2 Medium survey, the 1 deg2 Deep survey, and 3,000 KBO spectra at 2,500 s each leaves no time in the 5-year mission. The feasibility of every headline science goal therefore hinges on a number that is absent from this paper; if the true required exposure time is a factor of two larger, the mission cannot complete.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This white paper proposes ATLAS, a 1.5m probe-class space telescope with a 0.4 deg2 field of view and DMD-based multi-object spectroscopy at R=1000 over 1-4 microns. The mission concept includes three galaxy redshift surveys (2000, 100, and 1 deg2), a Galactic Plane survey, and pointed Kuiper Belt Object spectroscopy, all within a five-year L2 mission. The paper claims roughly 200 million galaxy redshifts out to z~7, with 183 million in the Wide survey, enabling galaxy evolution studies and dark energy measurements via BAO and redshift-space distortions. It also presents an instrument optical design, mission architecture, mass estimates, a schedule, and a cost estimate near the $1B probe-class target. The central feasibility claim is that the survey program can be completed with the stated sensitivity and multiplex factor; however, the paper does not include an exposure-time calculation or a full time budget.","tokens_in":13968,"tokens_out":5576,"duration_ms":46472,"significance":"If the survey completion can be demonstrated, ATLAS would provide a spectroscopic dataset of unprecedented size and density, roughly 12 times the WFIRST GRS number density, over a wide area. This would enable transformative galaxy evolution science and competitive dark energy constraints, as well as opening new windows on the Milky Way and outer solar system. The engineering description is concrete: the design uses mature DMD technology with reported TRL5-6, H4RG detectors, a passive cooling approach, and a detailed cost and schedule model with explicit mass and cost tables. The paper also benefits from a companion detailed publication (Wang et al. 2019a) and related science white papers. The main weakness is that the survey speed and sensitivity, which underpin all science goals, are asserted rather than derived, and the paper's own data-rate and multiplex numbers make the time budget a serious risk.","major_comments":[{"comment":"The central feasibility claim that the three galaxy surveys, the Galactic Plane survey, and the KBO program can be completed in five years is not supported by an exposure-time calculation. From Table 2, the mission has about 170 samples per day of 500 seconds each, i.e., roughly 85,000 s/day, or about 155 Ms over five years. The Wide survey alone requires 183 million spectra; at a multiplex of 6,000 this is about 30,500 exposures. Reaching the stated line flux limit of 5e-18 erg/s/cm2 with a 1.5m aperture at R=1000 plausibly requires several thousand seconds per exposure; if the required exposure time is 4,000 s, the Wide survey consumes about 122 Ms, leaving only about 33 Ms for all other surveys, and if it is 6,000 s, the Wide survey alone exceeds the total available on-sky time. The paper defers this calculation to Wang et al. (2019a) without presenting even a reference sensitivity estimate. This is load-bearing because all four science goals depend on survey completion. Please include an exposure-time calculation, a full time budget including overheads, and a demonstration that the stated line flux limit is reachable with the assumed aperture, throughput, and detector noise.","section":"Section I and Table 2"},{"comment":"The dark-energy and galaxy-evolution forecasts are generated from assumed survey parameters, including a number density about 12 times that of the WFIRST GRS and a redshift error of 1e-4, using Fisher-matrix techniques from the same collaboration. This is a standard forecasting approach, but it means the projected constraints are conditional on exactly the survey completion that is not demonstrated elsewhere in the paper. The manuscript should explicitly state this dependence and provide a sensitivity test showing how the Fig. 3 constraints degrade if the Wide survey achieves, say, 50% or 70% of the assumed number density, or if the effective exposure time is a factor of two larger than assumed.","section":"Section I and Fig. 3"},{"comment":"The claim that the ATLAS Galactic Plane Survey covers 700 deg2 in 0.4 years is stated without a supporting calculation. With a 0.4 deg2 field of view, this requires about 1,750 pointings; even at 1,000 s per pointing that is about 20 days, but the required SNR>30 for AB<18.2 and SNR>5 to AB=21.5, particularly through high extinction regions, likely requires longer integrations. The paper should provide the exposure time per pointing, the assumed source surface density, and show how the 0.4-year figure is derived, including overheads.","section":"Section I, Galactic Plane Survey"}],"minor_comments":[{"comment":"The phrase 'a square 0.75ʺ″ field' is unclear; the field of view is 0.4 deg2 and the slit size is 0.75 arcsec per micro-mirror according to Table 1. Please clarify whether this is a typo for the micro-mirror scale or the field size.","section":"Section II, first paragraph"},{"comment":"The constraint 'Long exposures (up to days)' is not obviously consistent with the stated cadence of 170 samples per day of 500 seconds each; please reconcile these numbers or define what a 'sample' means.","section":"Table 2"},{"comment":"The abstract states '~200M galaxies out to z=7 and beyond', while the text gives 183 million galaxies for the Wide survey; please state whether the 200M figure includes the Medium and Deep surveys and how the quoted totals are obtained.","section":"Abstract and Section I"},{"comment":"In the cost table, the entry 'Optical Instrument Instrument ROT' appears to have a typo, and the relationship between 'Total A-F' and 'Cost Target (incl LV)' should be clarified, since the mode total of $861.2M is lower than the $1,000M target while the 70th percentile is $975.7M.","section":"Table 7"},{"comment":"The Castellano et al. reference title contains a typo: 'Sopport' should be 'Support'.","section":"References"}],"recommendation":"major_revision","confidential_remarks":"The white paper is explicitly a summary of a more detailed companion paper (Wang et al. 2019a). The missing exposure-time and time-budget analysis is the key risk for the mission's feasibility claims; I would advise the editor to request that the authors either include the full calculation in this manuscript or make the companion paper's relevant sections available for review. The cost estimate appears detailed, but the 70th percentile cost is close to the $1B target and the relationship between the mode total and the cost target could be clearer."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nThe short version: this is an Astro2020 white paper, not a research preprint. It pitches ATLAS, a 1.5-m DMD-based multi-object near-IR spectrograph at L2, as the spectroscopic follow-up to WFIRST. The vision is genuinely attractive: 200 million redshifts over 2,000 sq deg, with R=1000 slit spectra in the 1-4 µm window, would transform galaxy-evolution and dark-energy science. The paper is clear, unusually honest about being a concept, and the engineering appendix (mass, cost, schedule, DMD TRL testing) gives it a concreteness that many white papers lack.\n\nThat said, the load-bearing feasibility number is absent. The Wide survey alone needs 183 million targeted spectra; at a multiplex of 6,000 that is about 30,500 pointings. To reach the stated line flux limit of 5e-18 erg/s/cm2, each pointing needs a real integration time, and this white paper gives none. The reader is referred to Wang et al. 2019a. Using the mission's own data-rate numbers (600 Mb per 500 s sample, 170 samples/day, 5 years), if one pointing takes 4,000-6,000 s, the Wide survey eats 130-150 Ms, essentially the whole mission before the Galactic Plane survey (0.4 year), the 100 deg2 Medium survey, the Deep survey, and the 3,000 KBO spectra. The stress-test arithmetic holds up. This does not make the mission impossible, but it means the central claim — that all four science goals fit in a probe-class envelope in five years — is asserted, not demonstrated.\n\nThere is also a circularity concern with the cosmological forecasts: Fig. 3 uses Fisher matrices from the same group (Wang et al. 2013, Samushia et al. 2019) with number densities twelve times WFIRST's as inputs. That is normal practice for mission proposals, but it means the forecasts are not independent predictions.\n\nWhat the paper does well: it is honestly labeled as a concept summary, it gives enough optical design and technology status to show the DMD approach is maturing, and it lays out a compelling four-goal science case. The novelty relative to Wang et al. 2019a is low — this is a condensation, not new work — but the concept itself is new and worth taking seriously.\n\nWho is this for? Anyone tracking Astro2020 mission concepts or planning future large-area spectroscopic facilities. It deserves a serious referee if submitted as a journal concept paper, but the referee should demand an exposure-time calculation and a full survey speed budget before believing the 200M-galaxy number.\n\nMy recommendation: engage with the wider ATLAS program, but do not treat this white paper as the definitive technical reference; get Wang et al. 2019a and ask for the time budget.","headline":"A well-packaged mission concept whose headline numbers hinge on an exposure-time budget the paper never shows; the idea deserves attention but the survey-speed claim remains unverified.","tokens_in":14860,"tokens_out":3348,"would_cite":false,"duration_ms":31839,"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":"ATLAS Probe proposes that a single 1.5-m space telescope with digital micromirror slits can obtain spectra of about 200 million galaxies out to redshift 7 while probing the Milky Way and the Kuiper Belt, within a probe-class cost envelope.","keywords":["ATLAS Probe","digital micromirror devices","multi-object spectroscopy","galaxy redshift survey","dark energy","galaxy evolution","Kuiper Belt objects","infrared space telescope"],"falsifier":"An end-to-end exposure-time model would settle it: using the stated telescope aperture, $R=1000$, 1–4 µm coverage, multiplex factor of about 6,000, DMD throughput, detector noise, zodiacal background, and overheads, compute how many five-year missions it takes to cover 2,000 square degrees to $5\\times10^{-18}\\ \\mathrm{erg\\,s^{-1}\\,cm^{-2}}$; this calculation is absent from the paper, and if the answer clearly exceeds one mission the central claim fails.","tokens_in":13437,"feed_emoji":"🔭","tokens_out":19614,"duration_ms":153508,"temperature":0.7,"pith_summary":"The paper proposes that one probe-class space mission—a 1.5 m infrared telescope with a 0.4 square degree field of view and arrays of tiny tilting mirrors (digital micromirror devices) used as configurable slits—can obtain $R=1000$ slit spectroscopy over 1–4 µm for roughly 200 million galaxies, out to redshift 7 and beyond. ATLAS is positioned as the spectroscopic follow-up to the WFIRST High Latitude Survey, turning its imaging over 2,000 square degrees into a three-dimensional map of the cosmos, with nested wide, medium, and deep surveys plus a Galactic plane survey and a Kuiper Belt program. A sympathetic reader would care because this is a single-instrument design that claims to deliver four science goals—galaxy evolution in the cosmic web, dark energy and tests of gravity, Milky Way structure through dust, and outer Solar System composition—inside a probe-class cost envelope. The viability of that claim rests on the DMD multiplex factor of about 6,000, which gives slit spectra rather than noisier slitless spectra, and on a five-year survey plan whose sensitivity budget is not derived in this white paper.","feed_headline":"One telescope could map 200 million galaxies back to redshift 7","feed_subtitle":"A 1.5m probe with digital mirror slits would deliver spectra for galaxy evolution, dark energy, the Milky Way, and Kuiper Belt objects.","key_machinery":"The load-bearing mechanism is the selectable slit: a digital micromirror device (DMD), a $2048\\times1080$ array of $13.7\\,\\mu\\mathrm{m}$ mirrors that each tilt $\\pm12^\\circ$, sending the light of one chosen target into the spectrograph and rejecting everything else. That is what produces a multiplex factor of roughly 6,000 without moving parts, and it is what lets a 1.5 m telescope do wide-area slit spectroscopy from space rather than slitless grism work with higher background. The instrument is four identical spectrometer modules covering $1$–$2.1\\,\\mu\\mathrm{m}$ and $2.1$–$4\\,\\mu\\mathrm{m}$ at $R=1000$, using H4RG-type detectors, with the optics kept near 50 K so that thermal emission stays below the zodiacal background.","core_discovery":"The paper's central claim is that ATLAS—a 1.5 m telescope with a 0.4 square degree field of view, resolving power $R=1000$, continuous wavelength coverage from 1 to 4 µm, and about 6,000 simultaneous targets selected by digital micromirror devices—can carry out the first wide-area slit spectroscopy from space at high redshift. In this design, each micro-mirror tilts to send the light of one galaxy into the spectrograph while blocking the rest, so a survey can be configured without moving parts. The authors argue that ATLAS and WFIRST together would make a definitive three-dimensional map of 2,000 square degrees, adding spectroscopic redshifts for roughly 200 million galaxies to the WFIRST imaging. The same observatory would then deliver the four headline goals: tracing galaxy evolution against the dark-matter web from cosmic dawn through the peak of galaxy assembly, measuring dark energy and testing general relativity, seeing through dust to the far side of the Milky Way, and obtaining about 3,000 Kuiper Belt spectra at wavelengths where compositional features have not been seen.","pith_inferences":["If the DMD architecture performs as assumed, the same configurable-slit technology could become a reusable platform for other probe-class spectroscopy missions; the small pathfinder mission named in the paper would be a direct test of that.","A near-term laboratory check is whether DMDs switch reliably at the roughly 50 K operating temperature, since the cryogenic tests described in the paper reached only about 80 K; stiction or switching failure there would directly cap the multiplex factor.","The science yield is coupled to WFIRST's High Latitude Survey as planned; if that imaging program changes depth, area, or filters, the roughly 200 million galaxy target list and the clustering forecasts would need to be rederived.","The dark-energy forecasts lean on the galaxy bispectrum, a method the paper acknowledges is still technically challenging; if that method matures more slowly than expected, the claimed gain over two-point statistics would be delayed, not invalidated."],"forward_implications":["A five-year ATLAS mission would yield a 3D galaxy map over 2,000 square degrees with spectroscopic redshifts for about 183 million galaxies, roughly twelve times the galaxy number density of the WFIRST galaxy redshift survey and fifty times Euclid's.","The same data would measure the cosmic expansion history and growth of structure over the redshift range 0.5 to 4 using baryon acoustic oscillations, redshift-space distortions, and the galaxy bispectrum, providing dark-energy constraints that do not depend on a chosen cosmological model.","Emission-line spectra at resolving power 1000 would detect Hα out to redshift 5 and [OIII]+Hβ in the late reionization era to redshift 7, connecting hundreds of millions of galaxies to their dark-matter halos via clustering.","A dedicated Galactic plane survey would produce signal-to-noise ratio above 30 spectra for 95 million stars to AB magnitude 18.2 across 700 square degrees in 0.4 years, probing the inner Milky Way through heavy dust extinction.","The Solar System program would gather about 3,000 Kuiper Belt spectra down to r magnitude 23.2, where the 3 µm region can reveal silicates and water ice not yet confidently identified in small Kuiper Belt objects."],"supporting_citations":[{"why":"Supplies the detailed ATLAS design, survey definitions, and sensitivity assumptions on which this white paper's numbers rest.","marker":"Wang et al. (2019a)"},{"why":"Argues for the very high density wide galaxy redshift survey and forecasts its dark-energy reach.","marker":"Wang et al. (2019b)"},{"why":"Provides the galaxy bispectrum forecasts behind the dark-energy constraints shown in Fig. 3.","marker":"Samushia et al. 2019"},{"why":"Establishes baryon acoustic oscillations in the galaxy power spectrum as a dark-energy ruler.","marker":"Blake & Glazebrook 2003"},{"why":"Forecasts BAO constraints from future large redshift surveys, the method ATLAS's clustering analysis builds on.","marker":"Seo & Eisenstein 2003"},{"why":"Demonstrates redshift-space distortions as a test of cosmic acceleration.","marker":"Guzzo et al. 2008"},{"why":"Provides the redshift-space distortion and growth formalism used to forecast growth-rate constraints.","marker":"Wang 2008"},{"why":"Gives the stellar-mass–halo-mass relation that ATLAS would measure as a function of redshift and galaxy properties.","marker":"Moster et al. 2013"},{"why":"Provides the SMHMR and star-formation-history framework used to interpret dark-matter halo masses.","marker":"Behroozi et al. 2013"},{"why":"Supplies the splashback-radius method ATLAS would use to measure average dark-matter accretion rates out to redshift 5.","marker":"More et al. 2016"}],"fun_headline_variants":["One probe, 200M galaxy spectra, a 3D Universe map","Mirror slits in space map 200M galaxies for dark energy","ATLAS: a 1.5m telescope that sees 200M galaxies","New probe concept: 200M redshifts, Milky Way far side, KBOs","Space telescope to map galaxy evolution and dark matter"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The plan's load-bearing premise is that the ATLAS Wide survey can reach a limiting line flux of $5\\times10^{-18}\\ \\mathrm{erg\\,s^{-1}\\,cm^{-2}}$ over 2,000 square degrees and obtain spectroscopic redshifts for about 183 million galaxies within a five-year mission; this survey-speed budget is asserted by reference to a companion paper, not derived here, so if real sensitivity or multiplexing is lower, the headline returns do not follow.","fun_headline_variants_meta":{"raw":{"variants":["One probe, 200M galaxy spectra, a 3D Universe map","Mirror slits in space map 200M galaxies for dark energy","ATLAS: a 1.5m telescope that sees 200M galaxies","New probe concept: 200M redshifts, Milky Way far side, KBOs","Space telescope to map galaxy evolution and dark matter"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000256,"raw_usage":{"total_tokens":1708,"prompt_tokens":1214,"completion_tokens":494,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":830,"completion_tokens_details":{"reasoning_tokens":397}},"tokens_in":830,"tokens_out":494,"duration_ms":4991,"temperature":1.0,"reasoning_tokens":397,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-14T10:02:36.126628+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"An end-to-end exposure-time model would settle it: using the stated telescope aperture, $R=1000$, 1–4 µm coverage, multiplex factor of about 6,000, DMD throughput, detector noise, zodiacal background, and overheads, compute how many five-year missions it takes to cover 2,000 square degrees to $5\\times10^{-18}\\ \\mathrm{erg\\,s^{-1}\\,cm^{-2}}$; this calculation is absent from the paper, and if the answer clearly exceeds one mission the central claim fails.","supporting_citations":[],"review_version":1}