The model for the long signal changes the fitted frequency ratio in 120-second TESS data for WD 1054-226 Sector 36

WD 1054-226 has known signals near 25 hours and 23.1 minutes. The March 2026 preprint by Korth and coauthors reports a ratio closer to 64 than 65 in TESS Sector 36. We test the ratio with four public 120-second SPOC light curves. A coherent model uses ten harmonics for the long signal and one sine for the short signal. The main Sector 36 fit gives a frequency ratio of 64.95560, with a formal local error of 0.01899. Freeing the ratio reduces the weighted squared error by 5.401. Models with 5, 10, and 15 long harmonics give different ratios and fit improvements. The short periods from all nine Sector 36 fits span 0.000039 minute. This spread compares fits to the same data. A model with two long waveforms gives ratio 65.03158 and fit improvement 0.468. Conditional tests inject signals on the observed Sector 36 time samples. Of 128 signals at ratio 65, the search recovers 126 within 0.1 with Gaussian noise and 128 with residual blocks. These tests assume a known long waveform. They do not give a significance level for physical resonance. The result shows that the ratio test depends on the long signal model. The package contains the source light curves, hashes, code, fit results, trial results, and figures.

WD 1054-226 · TESS · white dwarf · debris transits · frequency ratio · model sensitivity · reproducible photometry

Admitted 2026-11-09 08:23 UTC5 citing publications

References

supportsdoi:10.1093/mnras/stab3475
The main short-period peaks agree with the previously reported signal near the 65th harmonic. That paper established these signals.
doi:10.1051/0004-6361/202558102
The Sector 36 ratio discussed in Figure 7 motivates this test. The comparison uses the 8 March 2026 preprint version, arXiv:2603.07677v1. We use a different model and do not reproduce its GP posterior.
usesdoi:10.17909/t9-nmc8-f686
Source collection for the four public 120-second TESS SPOC light curves. Exact products, releases, URLs, and SHA-256 hashes are preserved.
usesdoi:10.1038/s41586-020-2649-2
NumPy supplies numerical arrays and linear algebra.
usesdoi:10.1038/s41592-019-0686-2
SciPy supplies numerical optimization.
usesdoi:10.3847/1538-4357/ac7c74
Astropy reads FITS files and computes Lomb-Scargle searches.
usesdoi:10.1109/MCSE.2007.55
Matplotlib makes the figures.

Cited by

5 publications

Public catalog entries on Sciscape.

usesTESS Sector 100 data for WD 1054-226 contain the known signal at 23.1 minutes
We reuse its four public 120-second FITS inputs for sectors 9, 36, 63, and 90. We retain their source records. We use new analysis code. Its result also motivates control of the slow baseline.
usesThe best 2019–2025 TESS sine clock for WD 1054-226 has a 0.457-cycle phase error in 2026 with six-hour quadratic baselines
We retain its source records for the four earlier 120-second light curves. Its test of the slow flux model motivates the baseline controls.
usesThree block sign tests do not detect 25-hour sidebands of the 23.1-minute signal in five TESS visits to WD 1054-226
We preserve its earlier archive records. Its long-signal model test motivates baseline checks.
usesNo full-band block sign trial reaches the five-visit TESS sine-phase cost of WD 1054-226 in three sets of 500
This work preserved the four earlier TESS files. It shows that the slow-signal model changes a fitted frequency ratio. We test short-signal phase with local baseline controls.
Five TESS pixel fits place the 23.1-minute signal near WD 1054-226 under an archive camera model
This prior test of the frequency ratio changed our choice of question before new fits.