Pith. sign in

REVIEW 3 major objections 5 minor 17 references

By combining split wavelength-calibration exposures, a hybrid lifetime-position architecture, ground-system LP patching, and a revised gain-sag flagging threshold, the COS2035 strategy aims to keep the COS far-ultraviolet channel fully prod

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

COS2035 combines four new operational techniques and two usage caps to keep Hubble's COS far-ultraviolet spectrograph productive through the 2030s despite accumulating detector damage.

T0 review reviewed 2026-08-02 challenge →

load-bearing objection COS2035 is a competent, genuinely new instrument-operations strategy that deserves referee time; the LP12 gain-sag extrapolation is the one load-bearing assumption to pin down. the 3 major comments →

arxiv 2607.14241 v1 pith:3BDB4B7X submitted 2026-07-15 astro-ph.IM

COS2035: Extending COS/FUV Operations Through the 2030s

classification astro-ph.IM
keywords COS2035gain saglifetime positionfar-ultraviolet spectroscopyHubble Space Telescopewavelength calibrationdetector operationssignal-to-noise cap
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper argues that the far-ultraviolet detector of the Cosmic Origins Spectrograph would, without mitigation, lose its most-used modes to gain sag by the 2030s. It presents COS2035, a four-part technical strategy plus two usage policies that together are claimed to keep the channel scientifically productive through the mid-2030s with no flight-software or hardware changes. The key moves are separating wavelength calibration from science exposures, letting different gratings operate at different detector positions at the same time, removing the eight-lifetime-position software limit through ground patching, and flagging gain-sagged pixels only when their integrated count loss exceeds a mode's fixed-pattern-noise limit. A sympathetic reader cares because COS is a unique UV capability, and the strategy's success determines whether high-resolution FUV spectroscopy remains available into the mid-2030s.

Core claim

The central claim is operational: the COS2035 strategy extends the useful life of the COS/FUV detector through the 2030s by exploiting previously unusable detector area and by changing when and how gain-sagged pixels are excluded. Specifically, SPLIT-wavecals remove the light leak as a constraint and open detector real estate above it; the hybrid-LP architecture places each grating/cenwave at the lifetime position best suited to its cross-dispersion profile and usage; LP-infinity patches the flight-software's eighth lifetime-position slot with ground-system tables so new positions require no software update; and the revised gain-sag flagging method flags a pixel only when its weighted contri

What carries the argument

Four mechanisms carry the argument: (1) SPLIT-wavecals, which take wavelength-calibration lamp flashes at a separate aperture-block position from the science exposure, avoiding the count-rate violation above the Pt-Ne light leak and opening the upper detector region, at about 15% per-orbit overhead for two-exposure orbits thanks to a simulated 600-s flash; (2) hybrid-LP, where different gratings occupy different lifetime positions simultaneously, allowing narrow profiles like G160M to fit between old positions and reducing each position's cross-dispersion footprint; (3) LP-infinity, which reserves the flight-software's eighth table entry as patchable scratch space loaded from ground tables,

Load-bearing premise

Continuum gain sag that lowers a column's integrated counts by less than the mode's 50th-percentile fixed-pattern-noise S/N limit has no measurable effect on extracted spectra, and this remains true as sag accumulates at LP12.

What would settle it

Take archival LP1/LP2-era spectra of a bright continuum source, simulate the LP12 cross-dispersion profile with accumulated sag, and extract with the gain-sag flagging threshold at the 50th and 90th percentiles of the tabulated fixed-pattern-noise values; if the two extractions differ by more than the roughly 3.8% fixed-pattern-noise floor for G130M/1291, or if integrated column loss exceeds the threshold in any column at projected LP12 sag levels, the claim that residual sag is harmless fails.

Watch this falsifier. Get emailed when new claim-graph text bears on it.

If this is right

  • The most-used FUV modes, notably G130M/1291, remain usable through the mid-2030s with no changes to flight software or detector hardware.
  • New lifetime positions become routine table updates: commissioning effort drops from weeks of FSW and rules work to days of table changes, with placeholder support for LP10-LP17.
  • G130M/1291 returns to a high-resolution location (LP12, between LP1 and LP2) with concurrent wavecals and no SPLIT-wavecal overhead, while the blue G130M cenwaves stay above the light leak at LP7.
  • The two usage policies—maximum achievable S/N per target and 2% per-program lifetime cap—spread the detector's remaining lifetime across many programs instead of letting a few exhaust local gain.
  • The strategy is projected to extend FUV operations into the 2040s, bridging the gap to future UV observatories and enabling concurrent far-UV spectroscopy with other flagship missions through the next decade.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The 50th-percentile threshold is a policy choice, not a physical boundary; the same data could be reduced with more conservative percentiles, and a side-by-side comparison at LP12's expected sag would reveal how much safety margin exists.
  • The hybrid-LP principle generalizes beyond COS: any detector suffering localized damage can be modeled as a multi-resource allocation problem, and the paper's mode-to-position matching suggests a template for other aging UV instruments.
  • The table-based LP selection exposed to users could eventually enable dynamic per-exposure LP assignment, letting schedulers route each exposure to the healthiest region rather than fixing placements by cycle.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit.

Referee Report

3 major / 5 minor

Summary. This manuscript, an Instrument Science Report, presents the COS2035 strategy for extending the operational lifetime of the Cosmic Origins Spectrograph far-ultraviolet (FUV) channel through the 2030s. The strategy comprises four technical developments—SPLIT-wavecals (decoupling wavelength calibration from science exposures and opening new detector real estate), a hybrid lifetime-position (LP) architecture, the LP-infinity ground-system framework (removing the eight-LP flight-software limit), and a revised gain-sag flagging method that evaluates integrated column count loss against a mode's maximum achievable S/N—plus two usage policies: a per-target S/N cap and a 2% per-program lifetime-usage cap. The paper reports that LP7 and LP10 are operational as of Cycle 33, with LP11 and LP12 in commissioning for Cycles 34 and 35. The central claim is that these changes keep the most-used FUV modes, notably G130M/1291 (about 40% of FUV exposure count), scientifically usable through the mid-2030s despite continuing gain sag, without hardware or flight-software changes.

Significance. If correct, the COS2035 strategy has substantial practical value: it would preserve Hubble's unique UV spectroscopy capability for the community through the 2030s and bridge to future missions. The paper is strong in its operational grounding: it documents concrete implementations (LP7/LP10 commissioning, LP-infinity patching, table-based APT/TRANS rules) with references to prior ISRs, and it introduces a thoughtful philosophy of flagging based on scientific impact rather than raw gain thresholds. The inclusion of quantitative usage caps (Table 2) is a useful service to observers. However, the load-bearing evidence for the most novel element—the revised gain-sag flagging method that justifies LP12—is incomplete: the stated per-pixel criterion does not, by itself, bound the column-integrated count loss, and this is exactly the quantity that determines extracted flux accuracy. The simulated 600-s flash success claim also lacks supporting distributional evidence. These gaps do not invalidate the overall engineering plan, but they affect the paper's central assertion of LP12 viability.

major comments (3)
  1. [§5 and §7] The revised flagging rule is stated per-pixel: 'A pixel with modal gain below 3 is added to the GSAGTAB only when its contribution to the integrated count loss along its column would exceed the maximum achievable S/N' (§5). This does not imply that the total integrated count loss across all sub-threshold pixels in a column remains below the fixed-pattern-noise S/N limit. LP12 sits between LP1 and LP2 where residual sag is concentrated in the outer wings of the cross-dispersion profile—pixels with small individual weights, so many such pixels could jointly push a column above the ~3.8% threshold. Figure 4 shows an integrated column-loss curve, but the text does not explicitly state that column totals were compared against the threshold, nor does it report the column-total distribution for the LP12 region. Please provide a direct analysis of the total integrated count loss per column for L
  2. [§5 and §6.1] The maximum-S/N threshold is taken from the 50th-percentile values in Roman-Duval et al. (2023) with no sensitivity analysis. The fixed-pattern-noise floor varies across modes and exposure patterns; a more conservative percentile (e.g., 90th) would lower the threshold, changing both the flagging decisions and the S/N cap. Since LP12 viability and the per-target S/N policy rest on this choice, please justify the percentile selection or show that results are robust to using a more conservative value.
  3. [§2.1] The simulated 600-s flash is claimed to recover the wavelength solution to within roughly 0.5 pixels for 85–90% of exposures, but no error distribution, sample size, or validation details are given in this paper (only references to ISRs 2024-08 and 2025-13). This simulation is one of the four technical breakthroughs and is central to the SPLIT-wavecal overhead saving. Please include the residual distribution (e.g., histograms of wavelength error) and a description of the validation sample, or provide a publicly accessible ISR with these details.
minor comments (5)
  1. [§6.2] The 27,000-count pixel lifetime is stated as a fact without a reference. Please add a citation to the relevant detector characterization study.
  2. [Figure 2] The figure is dense and the labels for LP10–LP12 are small; consider enlarging or using distinct markers for not-yet-commissioned positions.
  3. [§1] The text says 'above +5.5″' but later offsets are quoted with mixed signs (e.g., LP5 at +5.4″); please standardize the sign convention for cross-dispersion offsets.
  4. [Title page / Change History] The title page date (13 July 2026) and the Change History date (17 July 2026) are inconsistent; please verify the correct version date.
  5. [Table 2] The LP12 and LP11 entries are marked 'preliminary pending ETC support.' Please clarify whether these numbers are model predictions or measured count rates, and state the caveat in the table caption.

Circularity Check

0 steps flagged

No circularity found: the gain-sag threshold is externally tabulated and archival-tested; LP12 viability is an extrapolation, not a prediction built from its own input.

full rationale

The COS2035 report is an operational engineering document, not a derivation in which outputs are forced by inputs. The load-bearing quantitative claim—that residual gain sag at LP12 will not significantly affect G130M/1291 spectra—is supported by the revised flagging method presented in Section 5, but the method's threshold is not fitted to LP12. It is taken from the empirically tabulated fixed-pattern-noise maximum-S/N values of Roman-Duval et al. (2023), and Section 5 reports an archival X1DSUM comparison as an independent check that sub-threshold continuum sag does not alter extracted spectra. The LP12 conclusion is therefore an application of a pre-existing, externally measured threshold to a planned detector region plus an extrapolation to future sag accumulation; the paper explicitly identifies LP12 as in active commissioning and labels its ETC-based values as preliminary. The legitimate concerns raised about the method—per-pixel vs per-column summed loss, and the conservative-ness of the 50th-percentile threshold—are correctness or calibration risks, not circularity: they question whether the chosen threshold is too permissive, not whether the result was written into the definition. Use of the team's own prior ISRs is normal institutional continuity and none of those citations functions as a uniqueness theorem or as a substitute for the in-paper derivation. No step reduces, by construction or definition, to its own input.

Axiom & Free-Parameter Ledger

4 free parameters · 7 axioms · 0 invented entities

The strategy's quantitative load rests on the team's own calibration fits (600 s drift model, WCPTAB), the 27,000-count lifetime constant, and the 50th-percentile S/N thresholds from a self-cited ISR. These are operational inputs rather than free physics parameters, and most are reasonable for an engineering report; the median-percentile choice and the drift-fit success tail are the two that deserve scrutiny.

free parameters (4)
  • Piecewise OSM-drift model parameters (simulated 600 s flash) = grating-, exposure-time-, and drift-dependent; values in WCPTAB (not stated)
    Section 2.1: drift at 600 s is simulated from end-of-exposure drift via a piecewise function fit to archival COS data. Load-bearing for the SPLIT-wavecal overhead reduction (~15% vs ~30%).
  • 50th-percentile maximum-S/N thresholds (fixed-pattern noise) = per grating/FP-POS, from Roman-Duval et al. 2023
    Sections 5 and 6.1: used both as the gain-sag flagging threshold and the per-target S/N cap. The median-percentile choice is a modeling decision not justified against more conservative percentiles.
  • 27,000-count pixel lifetime = 27,000 counts in brightest pixel
    Section 6.2: basis for the 2% (540-count) per-program lifetime cap. An empirical detector-aging constant treated as fixed.
  • LP placement offsets = LP5 +5.4", LP6 +6.5", LP7 +8.3", LP10 -3.7", LP11 -6.7", LP12 +1.7"
    Sections 2.2, 3, 7: offsets chosen by gain-modeling trade studies balancing spectral resolution against neighbor-LP gain interactions. Hand-selected operational design parameters.
axioms (7)
  • domain assumption Aperture block repositioning is stable to sub-pixel accuracy under repeated moves
    Section 2.1, citing Fox et al. 2020. SPLIT-wavecals require the wavecal solution taken at a separate aperture position to transfer to the science exposure.
  • domain assumption Concurrent Pt-Ne wavecals above +5.5" would trigger a count-rate violation (light leak)
    Section 2, citing Oliveira et al. 2013. Defines the constraint that SPLIT-wavecals remove.
  • domain assumption External white dwarf (WD-0308-565) gain maps are equivalent in quality to internal FCA-illuminated maps above the FCA limit
    Section 2.3. Gain monitoring above the internal-lamp range depends on this.
  • domain assumption Modal gain < 3 corresponds to ~5% per-pixel count loss; gain sag is the dominant FUV degradation mode
    Section 5. Background detector physics the flagging revision builds on.
  • ad hoc to paper Early-exposure OSM drift is recoverable from end-of-exposure drift via a piecewise fit
    Section 2.1: the simulated 600 s flash is the paper's own calibration model; validity is asserted from archival fits with an 85-90% success claim.
  • domain assumption Fixed-pattern noise caps achievable S/N, so exposure beyond the cap wastes lifetime without adding signal
    Section 6.1, from Roman-Duval et al. 2023. Justifies the per-target S/N cap.
  • ad hoc to paper Continuum sag below the fixed-pattern-noise threshold leaves integrated flux accuracy unaffected, now and as sag grows
    Section 5: the revised flagging method's central premise; tested on archival data at current sag levels, extrapolated to LP12's accumulated sag.

reviewed 2026-08-02 · how reviews work

0 comments
Cite this review

Pith. "Pith review of COS2035: Extending COS/FUV Operations Through the 2030s." pith.science (2026). https://pith.science/paper/3BDB4B7X

@misc{pith2026260714241,
  author       = {Pith},
  title        = {Pith review of: COS2035: Extending COS/FUV Operations Through the 2030s},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/3BDB4B7X}},
  note         = {Machine review of arXiv:2607.14241}
}
Share X Bluesky LinkedIn Reddit HN
read the original abstract

The far-ultraviolet (FUV) detector of the Cosmic Origins Spectrograph (COS) accumulates gain sag where photons land, and without continued mitigation this degradation would render the most used modes unusable. To extend COS FUV operations through the 2030s, the COS team developed the COS2035 strategy, which builds on the existing COS2025 rules with four technical breakthroughs and two new usage policies. First, SPLIT-wavecals decouple wavelength calibration from science exposures and open detector real estate above the Pt-Ne lamp light leak. Second, a hybrid lifetime position (LP) architecture allows different gratings to operate at different LPs simultaneously. Third, the LP-infinity framework removes the dependence on the eight-LP limit in the COS flight software, supported by a new table-based APT and TRANS rules architecture. Fourth, a revised gain-sag flagging method evaluates integrated column count loss against the maximum achievable signal-to-noise (S/N) per mode. The two new usage policies cap per-target S/N at the maximum achievable value set by fixed-pattern noise, and limit any single program to 2\% of the lifetime at any single LP. With LP7 and LP10 enabled in Cycle 33 and LP11 and LP12 in active commissioning for Cycles 34 and 35, the COS2035 strategy positions the FUV channel for continued high productivity through the 2030s.

Figures

Figures reproduced from arXiv: 2607.14241 by Alan Welty, Anna Payne, Bethan James, Beverly Serrano, Christian I. Johnson, David French, David Sahnow, Debopam Som, Diego Mundo, Elaine Frazer, Jacqueline Hernandez, John Debes, Joshua Goldberg, Julia Roman-Duval, Karla Peterson, Kate Davis, Kate Rowlands, Lauren Miller, Leonardo Dos Santos, Marc Rafelski, Mark Giuliano, Mike Kelly, Nick Indriolo, Olivia Lupie, Rachel Plesha, Ravi Sankrit, Scott Swain, Serge Dieterich, Sten Hasselquist, Svea Hernandez, Travis Fischer, Van Dixon.

Figure 1
Figure 1. Figure 1: Cross-dispersion locations of the COS FUV lifetime positions across the two detector segments, shown as the accumulated two-dimensional spectral traces as a function of XCORR pixel and cross-dispersion offset from LP1. The dashed lines mark the PSA active-area limit at the top of the usable region and the PSA soft stop at the bottom. LP1 through LP7 and LP10 are commissioned positions; LP11 and LP12 (blue)… view at source ↗
Figure 2
Figure 2. Figure 2: The COS FUV detector real estate showing the locations of the PSA, BOA, and WCA at each lifetime position, in relation to the operational boundaries (light leak, aperture mechanism soft stops, and the range over which geometric distortion corrections are available). For lifetime positions that use concurrent TAGFLASH wavecals, the WCA shares the same dispersion (X) position as the PSA and is offset only in… view at source ↗
Figure 3
Figure 3. Figure 3: LIFETIME-POS optional parameter selection in APT. Beginning with APT 2026.3 (Phase II for Cycle 34), users specify the LP value at the exposure level for FUV science and target acquisition exposures. 5. Revised gain-sag flagging method The COS FUV detector is subject to gain sag, the gradual decline in microchannel plate response as a function of accumulated charge at each pixel location. As the modal gain… view at source ↗
Figure 4
Figure 4. Figure 4: Top: fractional count map for G130M/1291 on the FUVB segment, showing the contribution of each pixel to the integrated flux along its column, with pixels reaching modal gain 3 highlighted. These highlighted pixels include both the Lyα airglow holes and the continuum-driven gain sag on the short-wavelength side of the segment. Bottom: integrated count loss along each XCORR pixel within the extraction region… view at source ↗
Figure 5
Figure 5. Figure 5: Projected assignment of FUV gratings and cenwaves to lifetime positions through the early 2030s. Each colored bar shows the operational period of a given grating/cenwave at the indicated lifetime position. LP11 and LP12 are in commissioning for Cycles 34 and 35 respectively. Placements beyond LP12 (Cycle 36 and later), including the G160M migration and any subsequent lifetime positions, are notional: they … view at source ↗

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

17 extracted references

  1. [1]

    A., et al

    Dos Santos, L. A., et al. 2025, COS Instrument Science Report 2025-13

  2. [2]

    J., et al

    Fox, A. J., et al. 2020, COS Instrument Science Report 2020-05

  3. [3]

    2025, COS Instrument Science Report 2025-07

    Indriolo, N., et al. 2025, COS Instrument Science Report 2025-07

  4. [4]

    L., et al

    James, B. L., et al. 2023a, COS Instrument Science Report 2023-10

  5. [5]

    L., et al

    James, B. L., et al. 2023b, COS Instrument Science Report 2023-15

  6. [6]

    1999, COS Aperture Plate Baseline Design, COS-11-0009

    Morse, J. 1999, COS Aperture Plate Baseline Design, COS-11-0009

  7. [7]

    2013, COS Instrument Science Report 2013-02

    Oliveira, C., et al. 2013, COS Instrument Science Report 2013-02

  8. [8]

    2018, COS Instrument Science Report 2018-16

    Oliveira, C., et al. 2018, COS Instrument Science Report 2018-16

  9. [9]

    2013, COS Instrument Science Report 2013-16

    Osten, R., et al. 2013, COS Instrument Science Report 2013-16

  10. [10]

    2026, COS Instrument Handbook, Version 18.0

    Payne, A., & Dixon, W., eds. 2026, COS Instrument Handbook, Version 18.0

  11. [11]

    2018, COS Instrument Science Report 2018-23

    Plesha, R., et al. 2018, COS Instrument Science Report 2018-23

  12. [12]

    2015, COS Instrument Science Report 2015-03

    Proffitt, C. 2015, COS Instrument Science Report 2015-03

  13. [13]

    2016, COS Instrument Science Report 2016-01

    Roman-Duval, J., et al. 2016, COS Instrument Science Report 2016-01

  14. [14]

    2018, COS Instrument Science Report 2018-14

    Roman-Duval, J., et al. 2018, COS Instrument Science Report 2018-14

  15. [15]

    2023, COS Instrument Science Report 2023-11

    Roman-Duval, J., et al. 2023, COS Instrument Science Report 2023-11

  16. [16]

    2024, COS Instrument Science Report 2024-08

    Rowlands, K., et al. 2024, COS Instrument Science Report 2024-08

  17. [17]

    2021, Proc

    Sahnow, D., et al. 2021, Proc. SPIE, 11821, 118211H STScI 2024, COS Space Telescope Analysis Newsletter, July 2024 STScI 2025, COS Space Telescope Analysis Newsletter, March 2025 Instrument Science Report COS 2026-04(v1) Page 23

This paper was first reviewed by deepseek-v4-flash on August 2, 2026.