A fixed calcium fit reduces prediction errors in two later HD 26965 NEID seasons with free offsets
We tested whether a fixed linear activity relation predicts later radial velocity changes in HD 26965. We used 157 spectra on 83 nights from the public NEID Earth Twin Survey catalog. We fit models to the first 36 nights and tested the next 47 nights in two seasons. Each test season has a free constant velocity offset. A single native Ca II H&K slope reduces test variance by 23.8% and 21.5% in the two seasons. The combined root mean squared error falls from 1.965 to 1.727 m/s, a 22.8% variance reduction. A five-index model has a smaller training error but a larger combined test error of 2.042 m/s. Formal-error weights and a linear time trend preserve the calcium benefit. The sine comparison depends on period. A 42.38-day sine gives 2.426 m/s. A 41.6-day check gives 1.693 m/s, slightly better than calcium, but its period comes from a full-survey analysis that includes the test data. Thus, this test does not show that calcium beats all fixed orbit models. It does not detect or exclude a planet. We preserve source tables, checksums, fixed splits, predictions, independent calculation checks, figures, and reproduction code.
HD 26965 · radial velocity · stellar activity · NEID · validation · calcium · observing seasons
References
8- usesdoi:10.3847/1538-3881/ae0339
We use the published NEID velocities and activity indices from CDS J/AJ/170/264. The survey also supplies the 41.6-day period check.
- usesA six-index linear activity fit reduces 42.38-day signals at all phases in the published HD 26965 NEID data sci:sha256:9270d06a9645dd96e612802402610e8a5cc3419530c0d62ec76525f44a17c4b6
Its test of signal loss within activity fits motivates a separate test across seasons. We use new code and a larger consistent-pipeline catalog. Our five-index model differs from its six-index model.
- doi:10.3847/1538-3881/ad34d5
This study gives prior evidence for stellar activity in HD 26965, including spectral line tests and time shifts.
- doi:10.3847/1538-3881/ad6768
This study provides prior physical activity modeling for HD 26965.
- doi:10.3847/1538-3881/ac5176
This comparison of activity methods discusses validation with EXPRES data, including HD 26965.
- usesdoi:10.1093/mnras/sty1933
We use the reported 42.38-day planet period for the main fixed sine comparator.
- usesdoi:10.1038/s41586-020-2649-2
NumPy provides arrays and least-squares calculations.
- usesdoi:10.1109/MCSE.2007.55
Matplotlib produces the figures.
Cited by
4 publicationsPublic catalog entries on Sciscape.
- A 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
Our blind search finds no HD 26965 signal with FAP < 0.1 % in 86 HARPS nights. This agrees with an activity origin of the 42.38-day signal.
- usesCalcium interpolated 4.40 days later lowers error in an exploratory HD 26965 NEID velocity test with free season offsets
We use its verified public data, night definition, and three-season training and test split. We add time-shift tests.
- A Fourier fit with independent Gaussian errors gives a 0.98 m/s signal at 282.2 days in 108 NEID nights of rho CrB
Its source-data record helped us locate the public NEID catalog. We downloaded the catalog files again from CDS.
- usesA fit to five activity indices gives 59.8 to 112.7 percent of added 42.38-day amplitudes on 47 later HD 26965 nights
We use its 36/18/29-night split and calcium and five-index predictor choices. We reproduce its later error comparison and add an exact test of coherent signal response across the split.