Report: WASP-12 residual screen

Run date: 2026-09-24 UTC Bottom line: Bounded null. No candidate and no dossier. Evidence level: No unverified lead retained. The expected WASP-12 b transit signature is known and excluded from any novelty interpretation.

Ranked search outcomes#

Coordinates are copied from TESS headers: RA 97.6366528°, Dec +29.6722962°; coordinate frame/epoch not independently checked. This table is the complete ranked outcome for this finite campaign, not a survey-wide ranking.

Hypothesis and falsification#

Hypothesis: at least one isolated negative excursion in either of two specified 120-s TESS SPOC light curves is both ≥5 robust-MAD below a centered local median, spans ≥2 consecutive cadences, and falls outside ±0.08 orbital phase of the known WASP-12 b transit. Persistence under both SAP and PDCSAP and all three baseline windows would motivate a lead, not establish one.

Observed: no detected screen entry outside the broad phase veto in any product/reduction/window combination. This falsifies this narrowly defined screen outcome for the two files; it does not rule out smaller, shorter, in-transit, other-band, other-sector, or non-photometric phenomena.

NASA Exoplanet Archive pscomppars query on 2026-09-24 returned WASP-12 b, period 1.09141890100 d, transit midpoint 2457607.51930500, duration 3.001 h, and discovery reference Hebb et al. 2009. The phase window is intentionally broad and provisional: updated timing/period evolution was not fit. Archive query URL and literal ADQL are in SEARCH_LOG.md.

Provenance, footprint, and selection#

The two public MAST TESS SPOC files were transferred one-at-a-time from the existing [private Drive store] Drive tree to nonsynced scratch. TESS pipeline headers give S20 PROCVER=spoc-5.0.96-20230729/DATA_REL=73 and S43 PROCVER=spoc-5.0.45-20211006/DATA_REL=62. Both byte sizes and SHA-256 hashes exactly match docs/tier1_pack/MASTER_MANIFEST.csv after retrieval. No original was changed. The underlying archive URL, stored product IDs, and manifest provenance are documented in the search log. S20 spans header dates 2019-12-25–2020-01-20; S43 spans 2021-09-16–2021-10-11. FITS table time is TDB with BJD reference 2457000; code preserves stored TIME and separately records BJD-like values.

Finite footprint: one target, one sector-20 light curve and one temporally independent sector-43 light curve, single optical TESS passband, 120-s cadence. The latter is a second epoch from the same mission/instrument/pipeline, not independent instrumentation. All 18,954 and 17,804 rows were screened; 2,403 and 2,227 rows respectively were excluded by combined quality/nonfinite/nonzero-flux rules (per-reason counts not generated). All exclusions and limitations are in SEARCH_LOG.md.

Analysis and caveats#

The checksum-gated script uses only local FITS products, Astropy FITS I/O, NumPy and SciPy. It compares SAP and PDCSAP and centered running-median windows of 1, 2, and 3 days. It requires QUALITY == 0, finite TIME/flux, and nonzero flux. A robust scale is estimated as 1.4826×MAD of fractional residuals. Because cadence noise is time-correlated and the search scans thousands of samples and multiple recipes, the -5 scale threshold is not a p-value, sigma-calibrated detection, or false-alarm probability. Repeated transit-phase snippets and non-independent recipes are not counted as independent evidence.

Off-transit phase controls supply same-product descriptive robust scatter only (SAP/PDCSAP: S20 0.002145/0.001957; S43 0.001935/0.001911 fractional). No synthetic signal injection was run. Centroid measurements exist in the light curves but were not correlated with event residuals. There was no pixel-level background/blend inspection, cosmic-ray/hot-pixel audit, or independent reduction. These checks remain not tested, not passed.

Catalog/literature status#

The NASA Exoplanet Archive live pscomppars snapshot queried 2026-09-24 establishes the already-known WASP-12 b context; no separately versioned release identifier was exposed by that query. Generic web searches for current literature were attempted but the configured search service returned HTTP 402 (insufficient balance); ADS, ExoFOP, additional variable-star catalogs, and publication searches were not completed. Accordingly no claim about absence of prior reports is made. There is no candidate for a novelty claim.

Reproduction and artifacts#

• Configuration: campaigns/tess-wasp12-residual-01.yaml
• Code: tools/analyze_tess_residual.py
• Full search log and exact queries/checksums: SEARCH_LOG.md
• Machine-readable output: sector20/screen.json, sector20/normalized_series.csv, sector43/screen.json, sector43/normalized_series.csv (derived outputs; not source data)
• Environment: Python 3.13.3, NumPy 2.5.3, SciPy 1.18.1, Astropy 8.0.1.
• Reproduction commands, manifest gates, package/test context: see SEARCH_LOG.md.

The general Cygnus analysis suite is a triage utility collection validated on synthetic inputs; this run used a purpose-written, checksum-pinned screen rather than treating an uncalibrated utility score as a detection. The prior-art gate is incomplete. No data or report was sent externally.

Addendum (2026-09-24, later): calibration and ledgered re-run#

The campaign was re-run through the campaign runner (python -m cygnus.campaign run campaigns/tess-wasp12-residual-01.yaml; ledger runs #31–#34). The re-run reproduced the original outputs exactly: byte-identical normalized_series.csv in both sectors and identical screen entries (835 and 464, 0 outside the phase window). Two checks listed above as not tested were then run on these same light curves:

• Calibrated false-alarm threshold (sign-flip null). The same screen was applied to brightenings instead of dips (which a transit cannot produce), requiring SAP and PDCSAP agreement at the 2-day baseline, outside the phase window. Sector 20: 9 null events at k = 2.5, 2 at 3.0, none from k = 3.5; Sector 43: none at any k on the grid (so its k* is ≤ 2.5, an upper bound). The declared k = 5 is therefore conservative: no noise-driven crossings are expected at it in these two light curves. State: passed (for the false-alarm question only).
• Injection–recovery. Box dips of 0.1–2 % depth and 0.5–4 h duration, 10 per cell (seed 20260924), injected into SAP and PDCSAP outside the phase window. At the declared k = 5 the screen recovers ≥ 90 % of dips only from 1.5 % depth in both sectors (1 % dips: 0–40 %; ≤ 0.75 % dips: 0 %). At the calibrated thresholds the 90 % depth falls to 1.0 % (Sector 20) and 0.5–0.75 % (Sector 43). State: inconclusive against the reference signal (0.5 %, 2 h: 0 % recovered at k = 5).

Consequence for the bottom line. The bounded null stands, but it is weaker than the text above might suggest: it excludes isolated dips of roughly ≥ 1.5 % depth outside the WASP-12 b transit window in these two sectors, and says nothing about shallower ones. A screen meant to find shallower dips should use the calibrated threshold, with the persistence rule, and its completeness quoted.

Catalogue cross-match (new adapters; ledger prior_art). Within 30″ of the Sesame position: NASA Exoplanet Archive — WASP-12 b; TESS TOI table — TOI-1725.01 (TIC 86396382, disposition KP); VSX — WASP-12 (type EP, P 1.091422 d); SIMBAD — WASP-12, WASP-12b, the WASP-12BC pair and two 2MASS field stars. All consistent with the known system; no ADS literature query was made.

Details: sector20/calibration.json, sector43/calibration.json, runner/*.json, and the regenerated sky_record.json.