A free drift removes 98.75% of information for one circular 727.562-day signal phase in GJ 687 CARMENES DR1 with fixed white noise

We test how a free linear velocity drift changes sensitivity to the reported outer planet GJ 687 c. The 41 public CARMENES DR1 measurements span 353.24 days. We hold the outer period at 727.562 days and fit a circular inner signal at 38.142 days. With the same fitted white-noise covariance, a drift removes 98.75% of information in one circular signal phase direction. The loss is 98.66% for the 37 measurements with zero quality flags. At amplitude 2.44 m/s and a 1% false alarm rate, fixed-noise detection probability ranges from 13.9% to 92.3% without drift. It ranges from 1.6% to 14.0% with drift. The fitted outer amplitude changes from 2.63 to 13.53 m/s when drift is added. We check the information matrix by independent QR projection and test power with simulated records. Fixed-period null simulations that refit white noise give tail estimates of 0.159 without drift and 0.006 with drift. These tests compare against zero outer amplitude. They do not test the published orbit. We also test flags, period errors, activity proxies, time deletions, and eccentric orbit shapes. The package supplies source snapshots, provenance, code, matrices, power grids, and figures. The result gives a numerical sampling limit for fixed-period circular fits. It does not confirm or reject the planet. It is conditional on the tested mean and Gaussian noise models.

GJ 687 · CARMENES DR1 · radial velocity · planet detection · linear drift · sampling information · detection probability · reproducible research