Line-shape-based peak locking applied to the optical clock

In high-precision spectroscopy and quantum metrology, the stabilized system parameters at the peak of spectral lines are often required. Here, we propose a line shape-based peak locking (LSBPL) method that stabilizes system parameters at the maximum of arbitrary single-peak spectral lines. Unlike conventional proportional–integral–derivative (PID) control, which relies on a feedback function to convert an error signal into a control output, our approach exploits the intrinsic properties of the spectral line for peak locking. With application to the 171Yb+ ion optical clock using Rabi spectra, we demonstrated a comparable result to that of the PID method when the gain coefficient was set to g=0.3. The proposed LSBPL method also exhibits superior locking performance over the conventional PID method for asymmetric resonance profiles, maintaining stable closed-loop locking even when the PID method becomes ineffective. In addition, machine learning algorithms, especially random forest and support vector regression, are employed to fit single-peak spectral lines when an analytical expression is unavailable. This enables the peak-locking method to be universally applied even without requiring an explicit mathematical model, providing a flexible and generalizable tool for precision spectroscopy and quantum metrology.