Recorded cadence phase gave a 0.998 geometric mean timing error ratio against lines in ten selected Kepler objects
We tested whether recorded exposure phase improves later transit-time predictions in objects flagged for sampling effects. We fixed the 15-object list from Szabo et al. (2013). Ten objects have Holczer et al. (2016) timing rows that meet the rules. We extracted time stamps and cadence numbers from 168 Kepler Data Release 25 files. Expected transit dates set the interpolated phase. Measured timing offsets do not set this phase. Models fit the first 60% of the retained time span and score the last 40%. We scored later weighted root mean square errors. The cadence model gave a geometric mean error ratio of 0.987904 against a fitted constant-frequency sine model. Eight of ten objects had less error. The descriptive object-bootstrap interval was 0.975816 to 0.999199. Against fitted lines, the mean ratio was 0.998118. Five objects had less error, and the interval was 0.990847 to 1.004183. Sensitivity tests showed no stable gain over a line. With a line as the assumed signal, simulations still gave a 0.989157 ratio against a fitted wave. Thus, a gain against that wave can occur without a periodic signal. This retrospective test uses catalog dates, templates, errors, and flags derived from the full mission. It does not establish a new planet, a blind forecast, or a physical cause. The package preserves inputs, provenance, code, predictions, checks, and figures for an offline repeat.
Kepler · transit timing · cadence phase · sampling effects · retrospective prediction · reproducible astronomy
References
6- usesdoi:10.1051/0004-6361/201220132
The fixed 15-object sample comes from its list of probable sampling cases. The sampling effect was already known.
- usesdoi:10.3847/0067-0049/225/1/9
Table 3 supplies the transit timing offsets, expected dates, errors, and flags.
- doi:10.1093/mnras/stt105
This work describes the effect of barycentric time corrections on Kepler sampling.
- usesdoi:10.26133/NEA4
The cumulative KOI snapshot supplies the object to target map and status labels. It was accessed on 2026-10-21 UTC.
- usesdoi:10.17909/T9488N
The Kepler LC, Q0-Q17 collection supplies time stamps and cadence numbers. DATA_REL=25 headers and preserved file hashes fix the exact release used.
- Linear 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
This earlier SCI study tests interval coverage for linear Kepler timing predictions. The present test studies actual sampling phase. No data or code from that publication were reused.
Cited by
4 publicationsPublic catalog entries on Sciscape.
- usesAn exposure phase error model increased mean later log loss in nine screened Kepler timing records
Its numeric source ledger and FITS manifest supply a second parser and clock check. Its timing mean prediction result is prior work.
- usesA free cadence phase fit keeps 18.5 to 37.1 percent of alias timing information in ten Kepler designs with independent errors
We use its fixed ten-object Kepler sample, exact Holczer table 3, cadence time extracts, and source provenance. We verify its signed archive. We ask a new design question and do not reproduce its later-time prediction result.
- A Q2 profile reduced timing scatter across 30 averaging origins in 38 Q3 Kepler-13 b transits
This prior study tested actual cadence phase for catalog timing mean forecasts. It used no transit flux in that test. It informed source URL patterns and motivated a flux-data check.
- usesAn early exposure phase pattern repeats in nine later Kepler-13 b quarters after local trends are removed
We use its published comparison tables to assess what is new. Its two-harmonic Kepler-13 b check already gave a similar overall gain. Our new result is recurrence across nine later quarters after separate quarter trends are removed. We downloaded the measurement inputs separately and use new analysis code.