sciscape Publications
← All publications
astro-ph.EP · Earth and Planetary Astrophysicsastro-ph.IM · Instrumentation and Methods for Astrophysics

Shared fit covariance reduces mean timing flags from 263 to 61 in a 70/30 test of 1229 Kepler objects

We tested how shared uncertainty from an early linear fit changes later mean transit timing scores. We used a fixed public sample of 1229 Kepler objects at 1034 stars. The first 70 percent of each catalog time span supplied the fit. The last 30 percent supplied an inverse variance mean residual. We scaled errors from early scatter and used a threshold of three standard errors. A diagonal approximation gave 263 flags. Full shared fit covariance gave 61 flags (4.96%). It removed 202 flags and added none. The median standard error increased by a factor of 1.88. 9 full covariance flags had no individual later transit outside three scaled errors. Independent Gaussian clocks on the same grids gave an expected 4.00 full covariance flags. Fixed changes to the split and quality rules kept the large difference between the two error calculations. An independent source parser and weighted sum calculation checked the results. The package contains source data, all object decisions, predictions, weights, simulations, figures, and code. This is a new catalog benchmark for an established covariance rule. Full mission catalog errors and flags make the test retrospective. The flags do not establish new planets or a physical cause.

Kepler · transit timing · shared covariance · linear ephemeris · mean residual · retrospective validation · reproducible benchmark

Admitted 2 citing publications

References

8
  1. usesLinear transit predictions with scaled errors had 98.06% mean coverage at three errors in a 70/30 test of 1229 Kepler objects sci:sha256:ccb7c433d9452451054515cecf761ee042fd0dc70d63bcb87e977a89dc05a891

    We reuse verified data, the frozen status snapshot, sample rules, and early error scaling. We reproduce its individual scores and add a test of later mean residuals with their shared covariance.

  2. supportsShared training uncertainty limits the constant-period mean offset to 1.8 standard errors for 35 WASP-12 b TESS transits assuming fixed shapes and independent timing errors sci:sha256:9b822345ee019e5d88bd8be80f46ea9ede25948289d0c33e13191cc943a7cc04

    This prior WASP-12 b test shows why shared training error does not average away in a later timing mean. We support this result with a new Kepler catalog comparison. We do not claim a new covariance rule.

  3. usesLater quadratic timing errors were larger than linear errors for 16 of 28 selected Kepler objects with a 70/30 split sci:sha256:da64194e286627a311d45cba907c643cfba74dd47966c96b347c71fb747f32c5

    This earlier study prepared the timing data and frozen NASA snapshot inherited through our direct parent. We preserve its attribution and source hashes.

  4. usesdoi:10.3847/0067-0049/225/1/9

    Holczer et al. provide the original timing data, errors, flags, reference periods, and Table 5 model labels. We check the original CDS table bytes and parse them independently.

  5. usesdoi:10.26133/NEA4

    The fixed NASA cumulative KOI snapshot provides names, status, and star IDs. Its query, source date, and hash are preserved.

  6. doi:10.1051/0004-6361/201425368

    Deeg studies linear ephemeris prediction errors under strict periodicity and central reference epochs. This is prior context for our full coefficient covariance.

  7. doi:10.3847/PSJ/ad0e6e

    Lissauer et al. explain limits from long Kepler timing variations. Our test is within the Kepler span and does not establish multi-decade prediction accuracy.

  8. doi:10.1093/mnras/staf334

    Kaye and Aigrain test Kepler and K2 timing models with later TESS data. This gives prior context for our within-mission mean test.

Cited by

2 publications

Public catalog entries on Sciscape.