Kuiper Belt / TNO Optical Anomaly Hunt
An exploratory comparison of brightness models for 121 object/filter cases. The reported ranking selects 582301 (2015 RM306) for follow-up, but its model preference is inconsistent across time subsets. Archival checks provide no independent confirmation of unusual emission.
Premise and method
This screen compares two simplified descriptions of optical brightness as an object moves relative to the Sun and observer. A preference for one description does not establish the physical source of the light.
Reflected-light: ln F = c − 2 ln r − 2 ln Δ + ε
Intrinsic-like: ln F = k − 2 ln Δ + ε
Here r is heliocentric distance and Δ is observer distance. “Intrinsic-like” names the fitted distance dependence; it does not establish intrinsic emission or artificiality. Surface properties, phase behaviour, rotation, calibration, and sampling can affect this comparison.
The reported pipeline uses nightly flux binning, separate filters, minimum epoch and geometric-span cuts (N ≥ 20, span(ln r·Δ) ≥ 0.35), robust fits, bootstrap checks, bright/faint-point trims, and observatory and temporal subsets. ΔBIC = BIC(reflected) − BIC(intrinsic) ranks preference between these models; positive values favour the intrinsic-like fit.
Reported screening results
The run assigns one case to P, twenty-five to P-weak, and ninety-five to N. These are internal screening labels. P identifies a follow-up target, P-weak denotes weaker or check-dependent preference, and N denotes no qualifying preference under this procedure. The counts alone do not establish the screen's accuracy or detection sensitivity.
582301 (2015 RM306) — follow-up target
The full G-band sample gives ΔBIC = 16.588 across fifty nightly-binned epochs, a geometric span of ln(r·Δ) = 0.995, and a cross-validation advantage of 0.064. The report records passing trim and observatory-subset checks. Adding the tested phase correction to the reflected model gives ΔBIC = 17.689. The reflected-bootstrap test reports p = 0.0025; this is conditional on that null model and is not a probability of artificiality or an established search-wide significance.
Splitting the sample into time thirds gives ΔBIC values of 3.906, −1.405, and −0.174. The full-sample model preference therefore does not persist across those subsets. Its physical origin remains unresolved, and model misspecification or archival systematics remain plausible.
Archival follow-up
The reported GALEX, AllWISE, and NEOWISE catalogue checks found no time- and position-consistent detections. NEOWISER track validation returned 0 / 13 matches within an arcsecond, with a minimum separation of 1,949″. Coverage near a projected path alone does not establish that a survey could detect the target.
Checks of NVSS, TGSS, and VLASS found no track-coincident radio counterpart. The nearest cited NVSS source, NVSS 081927+153638 (6.2 ± 0.5 mJy), lies 30.93″ away, outside the 15″ matching threshold, and was treated as likely unrelated. These checks supply no corroboration; without adequate coverage and sensitivity, they also cannot exclude faint emission.
Secondary cases
523676 (2013 UL10), in G-band, has ΔBIC = 4.659 across twenty-seven epochs and fails the bright/faint-point trimming check. 44594 (1999 OX3), in R-band, has ΔBIC = 1.213 across seventy-nine epochs and a slightly negative cross-validation advantage. Both remain inconclusive.
Limitations
The sample is heterogeneous and the natural-light model is simplified. Reusing archival subsets tests consistency but is not a new observation. Selection effects, repeated comparisons, and sensitivity to alternative natural models need further assessment. The P label should be read alongside the failed temporal consistency check.
The working console
The interactive console summarizes the reported category counts, three named cases, and follow-up checks. Its orbital view is schematic. The downloadable PDF preserves the earlier report; the qualifications on this page apply to its stronger terminology.