A decreasing period improves 2023–2025 ZTF predictions for J0651 with fixed light-curve models

We test whether a decreasing orbital period improves predictions of later public measurements of the known double white dwarf J0651. Four fixed ZTF light curves provide 1,180 accepted measurements from 2018–2025. The sample uses DR24 for the main g object and DR23 for three other objects. We fit shared linear and quadratic clocks with separate Fourier profiles for each filter and constants for each archive object. The main model uses six harmonics. Fits to 940 measurements before 2023 predict 240 later measurements without changing any coefficient. The later weighted squared error decreases from 635.21 for the constant clock to 522.39 for the shrinking clock, a 17.8% decrease. The improvement remains positive for harmonic orders 2, 4, 8, and 10. It also remains positive for 30-second exposures and when only later object means are fitted. The full-sample period derivative is −8.50 × 10⁻¹² seconds per second. A residual bootstrap with 30-day blocks gives a conditional 95% interval of (−11.4 to −6.0) × 10⁻¹² seconds per second. The red filter alone does not resolve decay. The inference assumes stable filter profiles and a common clock. This study provides an independent later-data check and a reproducible prediction test. Earlier work measured the decay with much better precision. The package preserves raw data, source URLs and hashes, code, predictions, simulations, figures, and reproduction commands.

J0651 · double white dwarf · orbital decay · Zwicky Transient Facility · public data · astronomical time series · prediction test