Earlier linear Earth velocity fits reduce later squared loss by at least 10% in 11 of 74 selected HARPS difference series before the adopted 2015 boundary, with free year means
We tested whether fits from earlier years to Earth's listed radial velocity predict later differences between HARPS DRS and SERVAL velocities. We used the corrected April 2026 CDS release of HARPS-RVBank. Each year had a free mean. Linear fits reduced later squared loss by at least 10% in 11 of 74 selected series before an adopted 2015 boundary. The median gain was -0.00252. Fits from before the boundary reached this level in 2 of 32 series tested after it. Neither retained the level after every possible deletion of one test year. A source audit separated NZP restoration from the planned sum of four fields. A separate control that restores only NZP reached the 10% level in 37 of 74 and 14 of 32 series. The planned sum of four fields is retained as a different perturbation. No series after the boundary met the planned five-usable-year rule for a test within that era. Extreme training differences caused very large later losses. We preserve all failures, source rows, code, 1,329 planned fits, 318 supplemental fits, predictions, controls, and figures. The test measures contrasts within years and is conditional on published templates and calibration. It cannot establish a planet or a unique physical cause.
HARPS · radial velocity · BERV · SERVAL · DRS · historical validation · annual systematic effects · paired reductions · reproducible benchmark
References
9- usesdoi:10.26093/cds/vizier.36830125
We use the corrected 20 April 2026 Table 4 release. The source hash and all used fields are preserved.
- doi:10.1051/0004-6361/202348263
This article describes the expanded public HARPS archive and is the source of the paired velocity products.
- doi:10.1051/0004-6361/201936686
This article describes the earlier bank, reduction products, and velocity corrections.
- doi:10.1088/0004-637X/808/2/171
This study gives a physical example of annual HARPS detector effects. Our paired test does not identify this cause.
- doi:10.1051/0004-6361/202140986
YARARA gives a spectral approach to systematic effects. We use published velocities without spectral correction.
- A fixed annual term reduces omitted-year squared error by at least 10% in 9 of 77 selected HARPS velocity differences sci:sha256:f3037134738ddb96dff367a2b7a2e4edc3278c783f8eb414d3c28e9584786900
This close prior study tests annual paired differences in omitted years and includes a separate forward split. We add fixed BERV functions, earlier-only year-contrast fits, and an explicit transfer test on the corrected CDS release. We read its public main text but did not verify its complete package or reuse its files.
- A free BERV term removes over 99.9% of annual circular signal information at some phases in 151 HARPS designs under white noise sci:sha256:4f2be1c5e46742f40b4598bf5427841e077428f6d351fc726285900bc213d5ad
This work gives context on design overlap between Earth velocity terms and annual planet terms.
- A 1417-day term reverses the prediction gain at a fixed 71.45-day period in the GJ 9527 HARPS velocity difference sci:sha256:ada4abf6555b36987de1a05bb444eb2350a541395c8b9a6507d67e5d0a59d9b3
This paired-reduction study gives source correction relations and limits of common-curve cancellation. Our selected-row audit agrees with its NZP-only check and retains all exceptions. We read its source check but reused no code.
- Selecting periods in training data lowers median prediction gains for 12 selected HARPS signals with sine models and independent errors sci:sha256:c8d9aafe5aafa26b71515f5482a8f77fea910dd83419658af09cb20157f05c05
This study gives context for training-only model selection. We reuse no code or data.
Cited by
0 publicationsPublic catalog entries on Sciscape.
No incoming citations on Sciscape.