Repositorytess-single-transit-01REPORT.md

Document · REPORT.md

Report: toi-5696-03

Generated report (draft marker intact; pending review). Outcome, provenance, artifact-audit states and caveats.

Project-authored

Drafted by python -m cygnus.campaign report from the runner's outputs; reviewed 2026-09-25. The escalation and reviewer notes below were added by hand; every other number is the runner's.

Escalation (2026-09-25)#

Positive control failed on data that covers the epoch. Sector 72 (tess2023315124025-s0072-...) contains the catalogued epoch (BJD 2460284.7141) with 242 usable in-transit cadences, yet the screen recovered only 820 ± 48 ppm against a catalogue depth of 1204 ppm — not_recovered. docs/AGENT_RUNBOOK.md makes a positive-control failure on covered data an escalation. Per that rule, no threshold was tuned and no state was changed; the failure is reported as-is.

Most likely cause, from the runner's own calibration: at sector 72's k* = 3 the 90 %-completeness depth is 5000 ppm for an 8 h signal, and the catalogued transit is 1204 ppm over 8.09 h — about four times below the detection floor. A shallow-transit-below-sensitivity failure is therefore the first hypothesis, ahead of a wrong epoch or a wrong TIC. It cannot be distinguished further without an independent reduction or a deeper light curve, neither of which is built here.

Two further notes for a reviewer: the TIC is a multi-signal host (TESS_TOI lists TOI-5696.01 PC, .02 CP and .03 PC on the same TIC), so a blend with a neighbouring signal is a live alternative; and the three persistent events all lie in sector 46, which does not cover the epoch.

  • Campaign spec: campaigns/toi-5696-03.yaml
  • Parent queue: tess-mono-01
  • Ledger runs: calibrate_screen #285, fetch_products #284, known_signal_recovery #286, period_aliases #288, prior_art #289, residual_screen #287
  • Runner finished (UTC): 2026-09-25T04:18:11Z

Bottom line#

Positive control failed: BJD 2460284.7141: not recovered, depth 820 ± 48 ppm (catalogue 1204 ppm). Outside the catalogued epoch the screen left 14 threshold entries forming 5 distinct event(s), 3 persistent (SAP and PDCSAP, two or more baselines). None is vetted; see Screen events.

This is a pipeline check on a known object, not a discovery claim. No period is implied by a single transit.

Reviewer notes (2026-09-25, supersede pass)#

Superceded by the multi-archive runner: re-run through python -m cygnus.multi on 2026-09-25, steps fetch_products, calibrate_screen, known_signal_recovery, residual_screen, period_aliases, prior_art (ledger fetch_products #1585, calibrate_screen #1586, known_signal_recovery #1587, residual_screen #1588, period_aliases #1589, prior_art #1590). Same retrieved products, same science outputs: entries, depths, k-star calibration and aliases match the production-runner results recorded here; the multi runner additionally records a screen-suitability census and a red-noise systematics model per light curve. The record is regenerated by the multi runner; nothing in the review changes.

Reviewer notes (2026-09-25)#

Positive control failed on data that covers the epoch, so this is escalated above. Sector 72 contains the catalogued epoch (BJD 2460284.7141) with 242 usable in-transit cadences, but the screen found only 820 ± 48 ppm where the catalogue gives 1204 ppm — not_recovered. The other two retrieved light curves (sectors 45, 46) do not cover the epoch.

The sensitivity table gives the most likely explanation: at sector 72's calibrated threshold (k* = 3) the 90 %-completeness depth is 5000 ppm for an 8 h signal, and the catalogued transit is 1204 ppm over 8.09 h — roughly four times shallower than the detection floor. So the failure is consistent with the signal being below sensitivity rather than with a wrong epoch; it is recorded, not tuned around.

The three persistent screen events (2459563.12910 / .16521 / .21800, sector 46) span ≈ 0.09 d, i.e. one contiguous ~0.25–0.27 % feature, not three transits. Sector 46 does not cover the catalogued epoch. Catalogue cross-match answered by all four services: TESS_TOI lists TOI-5696.01 (PC), .02 (CP) and .03 (PC) on this TIC — a multi-signal TIC where the other dispositions matter for any interpretation.

Target#

FieldValueSource
nameTOI-5696.03NASA Exoplanet Archive TOI table (copied in the spec)
tic393546540NASA Exoplanet Archive TOI table (copied in the spec)
ra_deg161.156925NASA Exoplanet Archive TOI table (copied in the spec)
dec_deg6.333475NASA Exoplanet Archive TOI table (copied in the spec)
t0_bjd2460284.714061NASA Exoplanet Archive TOI table (copied in the spec)
depth_ppm1204.4442717NASA Exoplanet Archive TOI table (copied in the spec)
duration_h8.0912058NASA Exoplanet Archive TOI table (copied in the spec)
tmag9.1352NASA Exoplanet Archive TOI table (copied in the spec)
catalogue_row_updated2024-11-23 12:03:04NASA Exoplanet Archive TOI table (copied in the spec)

Products#

ProductSectorCovers catalogued epochSHA-256 (first 16)Retrieved now
tess2023315124025-s0072-0000000393546540-0267-s_lc.fits72True4a77c6806b7c1057True
tess2021310001228-s0045-0000000393546540-0216-s_lc.fits45False82aeabf2edcb94f7True
tess2021336043614-s0046-0000000393546540-0217-s_lc.fits46False60b33114b9c3d990True

Positive control (catalogued transit)#

ProductEpoch (BJD)StateUsable in-transit cadencesMeasured depth (ppm)Catalogue depth (ppm)Entry offset (h)
tess2023315124025-s0072-0000000393546540-0267-s_lc.fits2460284.71406not_recovered242820 ± 481204—
tess2021310001228-s0045-0000000393546540-0216-s_lc.fits—epoch not in this light curve——1204—
tess2021336043614-s0046-0000000393546540-0217-s_lc.fits—epoch not in this light curve——1204—

Depth: median PDCSAP residual inside ±duration/2 about the catalogued epoch, against a 2-d running median; the error is statistical only. A depth unlike the catalogue's may reflect dilution, detrending or an epoch/duration error; it is reported, not used as a pass mark.

Calibration and sensitivity#

Productk* (sign-flip null)At grid floor90 % completeness depth at k = 5, by durationat k*
tess2023315124025-s0072-0000000393546540-0267-s_lc.fits3False1h: 5000, 2h: 5000, 4h: 5000, 8h: 50001h: 5000, 2h: 2000, 4h: 2000, 8h: 5000
tess2021310001228-s0045-0000000393546540-0216-s_lc.fits3False1h: 5000, 2h: 5000, 4h: 5000, 8h: 50001h: 5000, 2h: 5000, 4h: 2000, 8h: 5000
tess2021336043614-s0046-0000000393546540-0217-s_lc.fits3False1h: 5000, 2h: 5000, 4h: 5000, 8h: 50001h: 5000, 2h: 5000, 4h: 2000, 8h: 5000

The null result (if any) excludes only dips deeper than the 90 %-completeness depth for their duration.

Screen events outside the catalogued epoch#

Entries merged where they overlap in time. Persistent = SAP and PDCSAP at two or more baselines.

ProductMid time (BJD)Deepest median residualMax cadencesFluxBaselines (d)Persistent
tess2021336043614-s0046-0000000393546540-0217-s_lc.fits2459563.12910-0.002692PDCSAP+SAP1, 2, 3yes
tess2021336043614-s0046-0000000393546540-0217-s_lc.fits2459563.16521-0.002592PDCSAP+SAP2, 3yes
tess2021336043614-s0046-0000000393546540-0217-s_lc.fits2459563.21800-0.002502PDCSAP+SAP2, 3yes
tess2021336043614-s0046-0000000393546540-0217-s_lc.fits2459563.04298-0.002362SAP3no
tess2023315124025-s0072-0000000393546540-0267-s_lc.fits2460272.20995-0.002302PDCSAP3no

None has been vetted: centroids, pointing, background, momentum dumps and other reductions are not tested. Most screen events in TESS light curves are systematics; each is at most an unverified lead.

Catalogue cross-match#

TOI-5696.03

  • NASA_Exoplanet_Archive (done, 2026-09-25): no match in NASA_Exoplanet_Archive within 30" as of 2026-09-25T04:18:04Z
  • TESS_TOI (done, 2026-09-25): 3 match(es) in TESS_TOI within 30" as of 2026-09-25T04:18:06Z: TOI-5696.01 (TIC 393546540, disposition PC); TOI-5696.02 (TIC 393546540, disposition CP); TOI-5696.03 (TIC 393546540, disposition PC)
  • VSX (done, 2026-09-25): no match in VSX within 30" as of 2026-09-25T04:18:08Z
  • SIMBAD (done, 2026-09-25): 1 match(es) in SIMBAD within 30" as of 2026-09-25T04:18:09Z: BD+07 2351 (PM*)

Checks#

CheckStateNote
Product integrity (SHA-256)passed3 product(s) checksummed at first retrieval
Known-signal recovery (positive control)failedBJD 2460284.7141: not recovered, depth 820 ± 48 ppm (catalogue 1204 ppm)
Calibrated false-alarm threshold (sign-flip null)passedscreen run at each light curve's own k* (3, 3, 3; ≤ 0 persistent null events outside the veto)
Synthetic signal injection–recoverypassedcompleteness for the reference box (2000ppm_4h) at each light curve's k*: 90%, 90%, 90% (pass mark 90%); 90%-completeness depths are in calibration.json
Catalogue cross-matchpassed1 target(s) × 4 services, radius 30″; 4 answered, 0 errored; results in the ledger prior_art table
Period aliases (repeat events)not_testedcatalogued transit not recovered; no reference depth
Alternative detrendingnot_tested
Difference-image centroids / blend auditnot_tested
Pointing / jitter correlationnot_tested
Literature (ADS) auditnot_tested

Reproduction#

python -m cygnus.campaign run campaigns/toi-5696-03.yaml
python -m cygnus.campaign report campaigns/toi-5696-03.yaml

Addendum 2026-09-26: suite-expansion checks#

Re-run on 2026-09-26 with the steps added by the suite expansion (docs/SUITE_EXPANSION.md §7.3) and the corrected moving-object check (TESS-centred SkyBoT positions, distance-aware). The earlier science outputs (screen, calibration, positive control, aliases, cross-match) were compared leaf by leaf and are unchanged apart from timestamps. States and notes below are copied from sky_record.json; the review above is not changed by them unless a note says so.

CheckStateNote
Target-to-Gaia identification (proper motion propagated)passedTOI-5696.03: Gaia DR3 3859284748342110208 at 0.00" (propagated 2016.0 → J2015.5; 0.03" unpropagated, proper-motion shift 0.03")
Stellar priors (Gaia colour and parallax)passedTOI-5696.03: Teff 5269 K, R* 0.81 ± 0.07, M* 0.88 ± 0.09, ρ* 1.64 ± 0.43 ρ☉ (dwarf sequence, M_G 5.55, no extinction)
Blend and dilution census (Gaia DR3 cone)inconclusiveTOI-5696.03: 1 Gaia neighbour(s) within 52.5", contamination 0.33%; depth 1204 ppm (catalogue depth); 1 could produce it if fully eclipsed (brightest 3859283962363313152, 48.5", ΔG 6.21); a centroid test is needed
Pointing and quality census per eventpassed3 persistent event(s), 3 clean; no artifact quality bit within ±0.25 d and no centroid, pointing or background shift beyond 5σ
Moving objects at screen-event epochsinconclusive3 event epoch(s) queried in SkyBoT (observer C57, r=600"); no known object bright enough within 63" plus its motion at any queried epoch (supports, does not prove, a non-asteroid origin); 1 epoch(s) not answered
Variable-catalogue collision (VSX)passedTOI-5696.03: no VSX entry within 10"
Object-class guard (SIMBAD)passedTOI-5696.03: BD+07 2351 otype PM* (star_or_other) at 1.0"
Independent repetition (other MAST collections)not_testedno repeat candidate with allowed period aliases
Independent-epoch confirmation (ZTF)not_testedno repeat candidate with allowed period aliases

failed on the per-event census means at least one persistent screen event carries an in-event artifact flag or pointing shift (event_census.json); it is a statement about those events, not about the catalogued signal.