Two stationary Gaussian noise models give larger fitted HAT-P-70 b half-depth widths in TESS Sector 98 than in Sector 32
We test whether two stationary Gaussian noise models remove a reported HAT-P-70 b transit width difference between TESS Sectors 32 and 98. We use 21 public two-minute PDCSAP events. We fit covariance parameters only outside transit and hold them fixed in separate spline profile fits. With a 0.005 day knot step, the Matérn 3/2 model gives half-depth widths of 182.25 and 187.34 minutes. Their difference is 5.09 minutes. The squared exponential model gives a 4.93 minute difference. With Matérn noise, 0.003 and 0.0025 day grids give 5.55 and 8.13 minute differences. A separate solver confirms the fine-grid fit. Equal-width noiseless inputs give a -0.54 minute fitted difference on the main grid. In 256 equal-width Gaussian trials, 0 differences reach the main observed value. On the main grid, the event bootstrap gives a 95 percent interval of 1.98 to 7.27 minutes. It holds noise parameters fixed. These tests retain a positive fitted offset, but its size depends on the profile grid. They exclude noise-parameter uncertainty, changes of noise during transit, and common flux-correction errors. They do not establish an orbit change or a new planet. The package includes raw data, hashes, source code, all trials, figures, and repeat instructions.
HAT-P-70 b · TESS · transit shape · correlated noise · Gaussian process · half-depth width
References
8- usesSpline fits give larger HAT-P-70 b half-depth widths in TESS Sector 98 than in Sector 32 sci:sha256:84b2d7ea0656b6874fe262dad202d25f7486bb51dcda4d2423810ca8d2046b9c
We reuse its two raw FITS files, data selection, source ledger, and unchanged MIT spline code. We add a local covariance test of its width observation.
- usesdoi:10.3847/1538-3881/ab36b5
The discovery paper supplies the period and reference epoch used to locate event windows.
- doi:10.3847/1538-3881/ac478f
- usesdoi:10.1111/j.1365-2966.2011.19915.x
Gaussian process methods for transit noise are prior work. This study applies a local outside-transit covariance fit to a two-sector width test.
- usesdoi:10.17909/T9-NMC8-F686
The TESS archive collection supplies the preserved SPOC light curves. File names, archive URLs, and SHA-256 hashes are in the supporting package.
- Two TESS flux products give different fitted widths for one January 2026 HAT-P-70 b transit under a fixed profile sci:sha256:16c7def968f7683a7d5ef4ea3b067371c8694e8874431a8d712dd409ac370a67
This separate later-visit flux-product comparison supplies processing context. No data from that visit are used.
- usesdoi:10.1109/MLSP.2010.5589113
Exact state representations of half-integer Matérn kernels are established methods. We use a two-state innovation calculation for faster whitening and check it against dense Cholesky whitening.
- Under one reference profile, fixed pixel sums give a larger HAT-P-70 b half-depth width in TESS Sector 98 than in Sector 32 sci:sha256:dc3553155e93fe267dd08159c9104fec87bed17ed0f9e4172724d71b69a19ea1
Its catalog abstract gives context from fixed pixel sums. Only that catalog record was inspected. Its code and derived data are not used.
Cited by
0 publicationsPublic catalog entries on Sciscape.
No incoming citations on Sciscape.