Rapid Orbital Decay in the Ultracompact Double-degenerate Binary eRASSU J060839.5 - 704014

We present timing and spectral analysis of the recently identified ultracompact double-degenerate (DD) white dwarf binary eRASSU J060839.5–704014 using observations from NICER and Einstein Probe (EP), together with archival XMM-Newton data. By phase-connecting the long-term XMM-Newton, NICER, and EP observations, we obtain a coherent quadratic timing solution, yielding an orbital period of 374.15013 (2) s and an orbital decay rate of Ṗ = −4.7 (1) × 10−11 s s−1. This orbital decay exceeds that measured in the prototypical DD binaries HM Cnc and V407 Vul. Assuming that the observed orbital evolution is primarily driven by gravitational-wave (GW) angular momentum loss, the inferred chirp mass is ∼ 0.43 M⊙, placing the source among the most massive known systems of this class. The phase-averaged spectra of NICER and EP-Follow-up X-ray Telescope (FXT) are described by a soft thermal component with temperatures of ∼126 and ∼144 eV, respectively, confirming the supersoft nature of the source. Phase-resolved spectroscopy reveals a clear decrease in temperature across the bright phase in both instruments, indicating a structured emission region with significant temperature gradients. These results establish eRASSU J060839.5–704014 as one of the most rapidly evolving ultracompact DD binaries presently known, belonging to the rare class of direct-impact ultracompact binaries, and a promising verification source for future low-frequency GW studies.

Improvements to AI Systems:

1. Orbital Decay Prediction Model – Train an AI system on the phase-connected timing solution (Ṗ = −4.7 × 10−11 s s−1) to predict future orbital periods and eclipse times for ultracompact double-degenerate binaries, enabling automated scheduling of follow-up observations.

2. Chirp Mass Inference Engine – Develop a Bayesian neural network that takes observed orbital decay rates and automatically infers chirp mass (e.g., 0.43 M⊙) and system inclination, reducing manual astrophysical modeling for new GW-driven binaries.

3. Phase-Resolved Spectral Classifier – Build a deep learning classifier that uses phase-resolved X-ray spectra (temperature gradients like 126→144 eV) to automatically identify structured emission regions and classify direct-impact vs. stream-impact binaries without human intervention.

4. Multi-Mission Data Fusion Pipeline – Create an AI pipeline that phase-connects heterogeneous time-series data from XMM-Newton, NICER, and Einstein Probe, handling gaps, instrument offsets, and noise to produce coherent timing solutions autonomously.

5. GW Source Prioritization Tool – Implement a reinforcement learning agent that ranks newly discovered ultracompact binaries by their orbital decay rate and chirp mass, flagging the most promising verification sources for low-frequency GW observatories (e.g., LISA) in real time.

What the Improved AI System Can Do:

• Automatically detect and characterize the fastest-evolving double-degenerate binaries from raw X-ray light curves.

• Provide real-time alerts for orbital decay anomalies that might indicate additional mass transfer or magnetic braking effects.

• Generate synthetic phase-resolved spectra for training other models, improving robustness in low-signal observations.

• Predict the detectability of GW signals from such binaries, optimizing observation strategies for future space-based detectors.