{"id":"d15435d7-3ced-49cb-bd34-3ae2ad76db70","arxiv_id":"2508.07824","paper_version":1,"verdict":"UNVERDICTED","confidence":"LOW","novelty_score":6.0,"correctness_risk":"unknown","formal_verification":"none","parameter_count":0,"one_line_summary":"tilepy now schedules ground and space telescopes to observe the highest-probability sky regions of gravitational-wave, gamma-ray burst, and neutrino events, while handling occultations and varied field-of-view shapes.","lead":"This paper announces new capabilities in tilepy, a software tool that plans follow-up observations of poorly localized cosmic events by deciding which patches of sky each telescope should observe. The updates let the tool schedule both ground- and space-based observatories while accounting for Earth, Sun, Moon, and radiation-belt constraints.","discovery_kind":"extension","skeptic_critique":{"model":"deepseek-v4-flash","headline":"No significant objection identified; only a verification gap remains.","rationale":"The reader's weakest assumption captures the same point: the central claim depends on the fidelity of the observability constraints and on validation against real operations. Since only the abstract is available, there is no evidence to confirm or refute this assumption. I find no specific technical error or internal inconsistency to raise as a load-bearing objection. The appropriate verdict remains UNVERDICTED, exactly as the reader concluded. The proposed concrete test is a reasonable next step that would move the verdict toward conditional acceptance if tilepy matches or beats a baseline scheduler on historical events.","tokens_in":764,"tokens_out":1083,"duration_ms":15410,"concrete_test":"Obtain a set of historical GW alerts with published skymaps and the actual observation logs of the follow-up facilities that used tilepy (or similar tools). Run the current tilepy version on those skymaps with its default constraints, and compare the generated schedules against (a) the actual pointings executed, and (b) a Monte Carlo injection of the true source location to compute the probability enclosed per unit observing time. Also compare against a greedy rank-by-probability baseline. If tilepy schedules do not cover at least as much probability as the baseline or violate the stated slew/occultation constraints, the central capability claim weakens.","verdict_should_be":"UNCHANGED","load_bearing_attack":"The abstract-only submission provides no equations, code, or test results, so the central claim that tilepy produces executable schedules for space-based observatories cannot be checked. No internal inconsistency is apparent from the abstract; the claimed capabilities are plausible and consistent with known tilepy functionality. The load-bearing assumption is that the modeled observability constraints (Earth/Sun/Moon occultation, SAA passage, slew limits, field-of-view geometry) are faithful enough that the generated schedules are actually executable by real observatories. If the constraint models omit important effects (e.g., stray light, thermal constraints, attitude control margins), the claimed 'optimized' schedules could be unusable. However, this is a generic and unsubstantiated concern, not a demonstrated flaw. The paper is best treated as unverified rather than wrong.","agreement_with_reader":"agree"},"referee_report":{"model":"deepseek-v4-flash","summary":"This abstract-only submission presents tilepy, a Python library for scheduling electromagnetic follow-up of poorly localized transients such as gravitational-wave events, Fermi-GBM gamma-ray bursts, and IceCube neutrino candidates. The claimed novelties are support for space-based observatories with Earth/Sun/Moon occultation and South Atlantic Anomaly passage constraints, handling of diverse field-of-view shapes, and the use of artificial-intelligence-based scheduling. The text gives examples of use on transient events but provides no equations, algorithmic details, quantitative performance results, or validation against real observatory operations. This review is based solely on the abstract because the full manuscript text was not available.","tokens_in":912,"tokens_out":2084,"duration_ms":27398,"significance":"If the capabilities claimed in the abstract are realized and validated, tilepy would be a practically useful tool for multi-messenger astronomy, particularly for the emerging need to coordinate ground- and space-based follow-up of poorly localized alerts. The breadth of considered constraints—occultation, SAA passage, variable FOV shapes, and AI scheduling—addresses real operational problems. However, the abstract alone offers no evidence that the optimization is correct, that the constraint models are faithful, or that the AI scheduler generalizes beyond training examples. The significance therefore cannot be assessed from the submitted text; the claims are plausible but unverified.","major_comments":[{"comment":"The central claim is that tilepy 'optimizes' follow-up scheduling, but the abstract gives no definition of the objective function, the constraint set, or the optimization method. There is no quantitative evaluation—e.g., probability enclosed, time-to-observation, comparison with greedy scheduling or current practice—so the reader cannot verify that the schedules are actually optimal or even improved. This is load-bearing for a software paper.","section":"Abstract"},{"comment":"The mention of 'optimizing observation scheduling with artificial intelligence tools' is not accompanied by any description of the architecture, training data, loss function, or validation methodology. A learned scheduler risks overfitting to the specific event set used for training; without evidence of generalization, this claimed innovation cannot be assessed. The abstract should at least state how the AI component is trained and evaluated.","section":"Abstract"},{"comment":"The claim that tilepy can operate with space-based observatories while accounting for Earth, Sun, and Moon occultation and SAA passage implies that the generated schedules are executable by real spacecraft. The abstract provides no validation against actual operations, telemetry, or simulated attitude-control constraints. If the modeled constraints omit important effects (e.g., stray light, thermal limits, attitude margins), the 'optimized' schedules may not be flyable. This verification gap directly affects the paper's main new capability.","section":"Abstract"}],"minor_comments":[{"comment":"No URL, repository link, version number, or software license is mentioned, which is essential for a software paper.","section":"Abstract"},{"comment":"No comparison with existing tools (including prior versions of tilepy) is given, making it difficult to identify the incremental contribution.","section":"Abstract"},{"comment":"'South Atlantic Anomaly' is abbreviated only implicitly; the abstract should spell out SAA at first use.","section":"Abstract"}],"recommendation":"uncertain","confidential_remarks":"The review package contains only the abstract, not the full manuscript. The recommendation 'uncertain' reflects that the central claims cannot be checked without the full text. I would advise the editor to obtain the complete paper—particularly the sections describing the optimization formulation, the AI scheduler, and the validation on real or simulated events—before a final decision. This is not a rejection of the work; it is a request for the evidence that a software paper of this type normally requires."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"This is an abstract-only submission, so the whole review hinges on what the abstract does and does not claim. What it does claim is coherent and plausible: tilepy now handles space-based observatories, treats Earth/Sun/Moon occultation and South Atlantic Anomaly passage, supports arbitrary field-of-view shapes, and adds an AI-based scheduling mode. Those are real extensions to a tool that already existed for ground-based follow-up of poorly localized transients. The writing is honest in one important respect: it says \"examples of its use,\" not \"validated on observations\" or \"outperforms baseline.\" That restraint is a point in its favor.\n\nThe soft spots are mostly a verification gap. There are no numbers, no comparisons, no description of how the AI scheduler was trained or what objective it optimizes, and no indication that the constraint models were checked against real telescope operations. The load-bearing assumption is that the occultation, SAA, and slew models are faithful enough that the optimized schedules are actually executable. That concern is real but generic; nothing in the abstract makes it a demonstrated flaw. A second, smaller soft spot is that the abstract cites no prior work, so we cannot tell how much of this is new relative to the authors' own previous tilepy papers. That is probably fine in an abstract, but it makes novelty hard to judge.\n\nI would not treat this as wrong. I would treat it as unverified. The central claim—that tilepy can produce optimized follow-up schedules for space-based observatories—is the kind of thing that can be checked by running the code on recorded events and comparing against simpler heuristics. If the full paper contains that kind of validation, it is a solid software contribution. If it does not, then the AI scheduler in particular will be hard to trust, because without a baseline comparison \"optimized\" is just a word.\n\nMy recommendation: send this to peer review, but only if the full manuscript includes quantitative tests (execution time, schedule quality, comparison to greedy or human planning) and a clear description of the AI method, including how the training set was generated and whether any test events were part of training. If the authors can show that, the paper is useful to anyone planning EM follow-up with space telescopes. If they cannot, the paper should be revised before acceptance. Either way, the abstract alone is not enough to desk reject an otherwise plausible tool paper.","headline":"Plausible engineering extension of tilepy to space-based and AI scheduling, but the abstract alone provides no evidence that the new modes actually execute on real telescopes.","tokens_in":1359,"tokens_out":1359,"would_cite":false,"duration_ms":19242,"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":"tilepy now plans telescope follow-ups of poorly localized transients for ground and space observatories, folding in Earth, Sun, Moon, and South Atlantic Anomaly constraints.","keywords":["multi-messenger astronomy","telescope scheduling","gravitational-wave follow-up","localization skymaps","space-based observatories","South Atlantic Anomaly","field-of-view tiling","AI scheduling"],"falsifier":"Run tilepy on a historical poorly localized event whose electromagnetic counterpart is known, such as a gravitational-wave event with an observed kilonova, and check whether the event's actual counterpart falls inside the top-ranked scheduled tiles; if the counterpart is systematically missed in the highest-probability tiles, the scheduling optimization and its constraint modeling fail.","tokens_in":1099,"feed_emoji":"🔭","tokens_out":1532,"duration_ms":38117,"temperature":0.7,"pith_summary":"This paper presents the latest developments of tilepy, a Python library that schedules telescope observations to follow up astrophysical transients with poor sky localization, such as gravitational-wave events, Fermi-GBM gamma-ray bursts, and IceCube neutrino candidates. The central claim is that tilepy now extends from ground-based to space-based observatories, explicitly modeling Earth, Sun, and Moon occultation plus South Atlantic Anomaly passage, while supporting arbitrary field-of-view shapes and AI-based scheduling. A sympathetic reader would care because poorly localized events can span thousands of square degrees, and automated optimization saves scarce telescope time for time-domain programs. The paper argues tilepy provides a common platform for diverse observatories, working individually or simultaneously, and demonstrates its use on real transient events.","feed_headline":"tilepy plans telescope follow-ups from ground and orbit","feed_subtitle":"New scheduler folds in Moon, Earth, Sun occultation and SAA passage for gravitational-wave, burst, and neutrino alerts.","key_machinery":"The scheduling optimizer that tiles candidate fields over the localization probability map, ranking pointings by scientific return while respecting telescope-specific constraints, including visibility, slew limits, field-of-view geometry, occultation by Earth, Sun, and Moon, and South Atlantic Anomaly passage for space observatories.","core_discovery":"tilepy is presented as a mature tool that turns a localization probability map of a poorly localized transient into an optimized observing schedule, and this contribution adds three capabilities: space-based observatory support with explicit modeling of Earth, Sun, and Moon occultation and South Atlantic Anomaly passage; handling of varied field-of-view shapes beyond simple circular footprints; and the possibility of optimizing schedules with artificial-intelligence tools. The paper's claim is that these extensions make tilepy a ready platform for electromagnetic follow-up across the multi-messenger landscape, including gravitational waves, gamma-ray bursts, and neutrinos, for observatories","pith_inferences":["The same tiling machinery could apply to other poorly localized alerts, such as fast radio bursts or orphan afterglow searches, with only a change of the input probability map and constraint set.","A natural, testable extension is a historical replay: run tilepy on past events with known electromagnetic counterparts and check whether the top-ranked tiles contained the counterpart, which the abstract does not yet report.","The AI-based scheduling component, if it learns from executed schedules, could in principle convert modeled constraints into empirical ones, correcting for discrepancies between simulated visibility and real observatory behavior.","The paper leaves open quantitative comparisons of tilepy's schedules against actual executed observations or against other schedulers; such benchmarks would sharpen the claim that the modeled constraints are faithful enough to execute."],"forward_implications":["Events localized to tens or thousands of square degrees can be followed up automatically with a ranked, constraint-aware tiling rather than ad-hoc manual pointing.","A single scheduling platform can coordinate ground- and space-based observatories simultaneously, including mixed fleets with different field-of-view shapes.","Space-based observatories gain usable schedules that avoid instrument-damaging South Atlantic Anomaly passages and occulted regions, extending the lifetime of follow-up campaigns.","AI-based scheduling opens the door to adaptive re-planning when new localizations arrive mid-campaign.","Observatory time is saved for other time-domain science because the optimizer concentrates exposures on the highest-probability regions first."],"supporting_citations":[],"fun_headline_variants":["tilepy scheduler now handles space telescopes with occultation","From ground to orbit: tilepy optimizes transient follow-ups","tilepy adds space-based observatories to transient follow-up planning","Orbit-ready tilepy: scheduling for gravitational wave and neutrino alerts","tilepy brings AI to multi-messenger follow-up scheduling"],"cache_read_input_tokens":3456,"weakest_assumption_plain":"The whole optimization rests on the assumption that the input localization probability maps are accurate and that the modeled observability constraints (Earth, Sun, Moon occultation, South Atlantic Anomaly passage, slew limits, field-of-view geometry) are faithful enough that the computed schedules can actually be executed by real observatories.","fun_headline_variants_meta":{"raw":{"variants":["tilepy scheduler now handles space telescopes with occultation","From ground to orbit: tilepy optimizes transient follow-ups","tilepy adds space-based observatories to transient follow-up planning","Orbit-ready tilepy: scheduling for gravitational wave and neutrino alerts","tilepy brings AI to multi-messenger follow-up scheduling"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.001163,"raw_usage":{"total_tokens":4658,"prompt_tokens":760,"completion_tokens":3898,"prompt_tokens_details":{"cached_tokens":256},"prompt_cache_hit_tokens":256,"prompt_cache_miss_tokens":504,"completion_tokens_details":{"reasoning_tokens":3810}},"tokens_in":504,"tokens_out":3898,"duration_ms":32843,"temperature":1.0,"reasoning_tokens":3810,"cache_read_input_tokens":256,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-05T21:49:08.071733+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"Run tilepy on a historical poorly localized event whose electromagnetic counterpart is known, such as a gravitational-wave event with an observed kilonova, and check whether the event's actual counterpart falls inside the top-ranked scheduled tiles; if the counterpart is systematically missed in the highest-probability tiles, the scheduling optimization and its constraint modeling fail.","supporting_citations":[],"review_version":1}