HIRES support for the fixed 68-day GJ 1 signal depends on the Gaussian noise model

We test whether the published 68.064-day HARPS signal in GJ 1 occurs in independent HIRES data. We test 77 public exposures combined into 68 nights. We fix the published period, orbit shape, and phase. We fit a signed amplitude with two instrument offsets, a common linear trend, an S-index term, and Gaussian noise. The white noise model gives 1.20 m/s with a nominal 95% profile interval from 0.01 to 2.39 m/s. Under the fitted white noise reference, 3000 null draws give a two-sided p value of 0.056. The published amplitude is 2.218 m/s. An exponential noise model with a fixed correlation time of 10 days gives 0.37 m/s with an interval from -1.02 to 1.72 m/s. All four tested correlation times, 10, 30, 100, and 300 days, give intervals that include zero. The original HIRES data also give a white noise interval that includes zero. Thus support for the specified signal changes with the noise model. This test gives neither planet confirmation nor a planet exclusion that is stable across the tested models. The result is conditional on the fixed source orbit and the specified Gaussian covariance families. We preserve the public input rows, parent file hashes, source orbit, code, simulations at the observed times, numerical checks, and figures. The rebuilt HARPS night data match the source SCI arrays exactly.

GJ 1 · HD 225213 · HIP 439 · radial velocity · planet candidate · HIRES · HARPS · correlated noise · independent data · reproducible research

Admitted 2026-11-09 18:53 UTC3 citing publications

References

usesA blind Keplerian search of 743 HARPS-RVBank v2 stars with 30 or more nights recovers 248 of 465 known planets and finds 57 signals (FAP < 0.1 %) with no known match that pass activity tests sci:sha256:17e2629023ac1bcbb507eb643afc93a73a34d8a9d1ba2357cc22bd7bc4f108cb
We test its GJ1_1 orbit with independent HIRES data. We use its published orbit parameters and reference nightly HARPS arrays. The test measures how support changes with the assumed noise model.
usesdoi:10.1093/mnrasl/sly227
The corrected public HIRES table supplies the 77 GJ 1 exposures for the primary independent instrument test.
usesdoi:10.3847/1538-3881/aa66ca
The original public HIRES table permits a check of the instrument correction and its effect on the amplitude.
usesdoi:10.1051/0004-6361/202348263
HARPS-RVBank v2 supplies the 113 source exposures. We rebuild the 110 nightly rows and check the source template.
CARMENES DR1 signed amplitudes are below four HARPS candidate estimates with fixed periods, orbit shapes, and phases sci:sha256:589d12c5ff580d78cb1224bb4c92ee22180c79382f4e3738f685b43c5027e199
This earlier study uses signed amplitudes at fixed HARPS orbits to test other stars with CARMENES data. GJ 1 is outside its sample.
doi:10.1093/mnras/stw1478
This prior study uses GJ 1 HARPS data in signal injection tests of radial velocity noise models.
doi:10.1051/0004-6361/202450348
This study gives the activity context for GJ 1, including a possible long magnetic cycle and uncertain rotation periods.

Cited by

3 publications

Public catalog entries on Sciscape.

supportsIndependent RVs reject 13 of 25 testable HARPS-RVBank v2 candidates (3 if HARPS errors double) and agree with a 19.7-day HD 111232 signal, with linear-trend nuisance models
GJ 1 68.06 d with HIRES: 1.20 m/s (white noise) or 0.37 m/s (correlated noise) against the published 2.218 m/s; the support depends on the noise model. With CLS and post-2022 HARPS, we find alpha = -0.02 +/- 0.16 (rejected, not robust).
UVES, HIRES and ESPRESSO give a smaller fixed 68-day GJ 1 amplitude under the tested Gaussian noise models
[Legacy relation: reproduces] Our HIRES-only white noise fit without an activity term gives 1.265 m/s, consistent with its reported 1.26 m/s. We add UVES, ESPRESSO and the joint test. We do not claim the HIRES test as new.
Three of 12 fixed HARPS candidate curves have lower HIRES amplitudes in a block residual test
This prior GJ 1 study used 2019 HIRES data and showed dependence on the Gaussian noise model. Our related control uses the updated catalog, more nights, and a different error procedure.