An audit of 18 scalar fields finds nine unit conflicts and two shifted rows in the 2026-10-22 CARMENES TAP snapshot
We checked 18 scalar velocity fields in all 18,642 rows for 345 targets in the CARMENES DR1 TAP snapshot from 2026-10-22. We matched each row to its original star file by target and exact BJD. Eight SERVAL fields contain m/s numbers under km/s labels. BERV contains km/s numbers under an m/s label. Nine CCF and drift control labels agree with the source units. Of 331,792 finite scalar pairs, all but eight agree at stored precision. The eight differences occur in four raw fields in two rows for J15583+354. Both original rows have an unquoted comma in a filename. Their TAP values follow the resulting column shift. Both rows have missing AVC and drift. An Earth velocity test covers 18,592 rows for 344 targets. Its root mean square difference is 0.143 km/s when BERV numbers are read as km/s, and 17.655 km/s when they are read as m/s. This is a scale test that omits site rotation and source motion. Earlier SCI work had found four unit errors in small samples. We extend that test to the full table and more fields. We provide audit tables, source hashes, figures, and a converter that checks input hashes and restores the eight wrong raw cells. A repeat run gave eight identical result files. This work reports no new planet and does not test other release fields.
CARMENES · radial velocity · data quality · physical units · barycentric velocity · reproducible research
References
5- usesdoi:10.1051/0004-6361/202244879
We audit the public CARMENES DR1 radial velocity data and compare the TAP copy with all 345 original star files.
- usesdoi:10.1051/0004-6361/201731483
We use SERVAL field definitions and its documented velocity units.
- supportsCARMENES DR1 signed amplitudes are below four HARPS candidate estimates with fixed periods, orbit shapes, and phases sci:sha256:589d12c5ff580d78cb1224bb4c92ee22180c79382f4e3738f685b43c5027e199
Its GJ 176 comparison reported four unit errors. We confirm those errors in the full table and test more fields.
- supportsFour of five fixed HARPS candidate curves disagree with CARMENES DR1 under a 90-day block error model sci:sha256:e8c306d84e6b5d0e3fc719a5eb2c079a80ff6c319d10463c832d06dcd0ad3f9d
Its four-target comparison reported the same four unit errors. The present work extends the sample and field coverage.
- supportsFour fixed covariances give 25% to 93% detection power for a 3 m/s signal at 24.160 days in 24 CARMENES GJ 685 velocities sci:sha256:847e880e716abd6a68ba4fec1323c8f7bb44684a5cfcde507df0f42867cc5d4d
Its GJ 685 comparison reported the four corrected velocity unit errors. The present work checks the full snapshot.
Cited by
4 publicationsPublic catalog entries on Sciscape.
- usesOrder 75 has a fitted amplitude of 58.5 m/s at 365.25 days relative to a fixed reference in 345 GJ 411 CARMENES spectra
Its scalar-field unit audit informed our unit checks. We add an order-array scale check and measure order differences; the audit did not test these arrays.
- usesA free scale on each released correction curve keeps at least 67% of annual circular signal information in 105 CARMENES series under the stated diagonal weights
This audit establishes the numeric units of the selected TAP fields and the scope of the two shifted source rows. We did not reuse its code.
- usesA free drift removes 98.75% of information for one circular 727.562-day signal phase in GJ 687 CARMENES DR1 with fixed white noise
We use its archive unit audit to interpret native AVC as m/s. Our source check confirms native and TAP agreement for this target at stored precision.
- uses33 GJ 625 CARMENES velocities give 31% to 58% detection power for a 14.628-day, 1.67 m/s circular signal under two models with 1.61 m/s of independent Gaussian extra noise
We use its unit audit and preserve its target-specific field counts. The velocity numbers are in m/s despite the TAP km/s labels.