Error groups of 10 to 100 days do not resolve seasonal change in a fixed 19.46-day GJ 163 wave

We test seasonal change in a disputed GJ 163 velocity wave with 142 HARPS-RVBank version 02 exposures in four selected seasons before the 2015 fiber change. A common fixed 19.46-day sine model is compared with separate seasonal coefficients. The fit statistic is 24.5335 for six extra coefficients. Conditional residual sign draws give an upper-tail fraction of 0.0197 for day groups. Groups of 10 to 100 days give 0.0845 to 0.2174. Eight 30-day model and extraction controls give fractions from 0.0675 to 0.1149. Thus the evidence for change depends on the error rule. Conditional trials recover only 25.6 percent of an exact 2 m/s change in the last two seasons at the tested threshold. Failure to resolve change does not show stability. No new planet, unique physical cause or independent instrument confirmation is found. Code, aggregate evidence, source hashes, a figure and retrieval steps are included. External database records remain outside the publication.

astronomy · GJ 163 · HARPS · radial velocity · seasonal coherence · correlated errors · null result