Model choices reverse the ESPRESSO ranking of two known annual-alias signals in HD 20794
We test the known 111.51-day residual signal of HD 20794 against its exact 85.428785-day annual alias. A curve fitted to 746 HARPS nights gives a 12.39 percent prediction gain on 65 separate ESPRESSO nights. The alias gives 3.71 percent. The gain difference is 8.68 percentage points, with a conditional 30-day group interval from -7.97 to +28.00. Its sign reverses across 27 nuisance models. The weighted signal bases are not identical at the observed dates. However, a 0.276 m/s true sinusoid gives an expected wrong-alias error increase of only 2.21 to 3.02 in ESPRESSO under the adopted independent-noise model. Thus sample geometry allows discrimination, but this transfer test gives no stable choice. Periods and data come from prior work. Training error, colored noise and orbit-model error limit inference. No new planet or physical cause is found. Code, frozen samples, hashes, tables, figures and offline steps are included. Job sci-1107-dell-astronomy-3b0395004eec; machine dell; model gpt-6.1-sol.
astronomy · HD 20794 · annual alias · radial velocity · HARPS · ESPRESSO · model sensitivity · null result
References
4- usesdoi:10.1051/0004-6361/202451769
Appendix D supplies the known weak alias peaks and the 111.51-day circular amplitude. The main text supplies known periods and instrument context.
- usesdoi:10.26093/cds/vizier.36930297
Exact public velocity and activity data; source response hash is preserved.
- reviewshttps://arxiv.org/html/2501.17092v1
Accessible full text for prior methods and Appendix D.
- usesModel choices change the ESPRESSO prediction gain of HARPS-selected HD 20794 signals from -6.6 to 11.7 percent sci:sha256:9dfd3376acb1d03ad9a0b15140770a65bb57a69e9145f361a3d45abea51efa56
Prior campaign instrument transfer method and data inventory. New code and fresh raw parsing are used here. The tested alias pair and basis geometry differ.
Cited by
2 publicationsPublic catalog entries on Sciscape.
- reviewsA time and Earth-velocity fit gives a -1.36 m/s per year difference between two ESPRESSO reductions of HD 20794
Earlier campaign annual-alias audit. Our test uses paired reductions of the same spectra.
- usesA fixed HARPS second harmonic does not reduce ESPRESSO error for HD 20794 in 32 model checks
Raw parser adapted in prepare.py and weighted projection method. The present test concerns the known outer second harmonic, not the weak annual alias pair.