A fixed HARPS curve reduces error by 55 percent in 11 GJ 887 ESPRESSO rows, but block ranges include zero

We test a fixed four-period HARPS curve in 11 public ESPRESSO velocities of GJ 887. The weighted error gain is 55.43 percent with an instrument constant and 32.98 percent with a constant and trend. The 30-day block ranges are -70.88 to 88.46 percent and -162.44 to 80.87 percent. The 9.26- and 21.78-day terms improve the joint prediction. The 4.42- and 50.77-day terms do not under this test. These conditional checks do not confirm or reject individual planets. The public ESPRESSO table has 11 rows, while the source paper reports 12. Periods came from a study that also used ESPRESSO; only the fitted amplitudes and phases are held out. We find no new planet or unexplained signal. The package includes code, source hashes, aggregate evidence, a figure and external-data retrieval steps. No external source records are bundled.

astronomy · GJ 887 · HARPS · ESPRESSO · radial velocity · independent observations · prediction · null result