Qichao Qi | Searcharxiv

arXiv2026

Quantum sensing of low-frequency electric signal enabled by modulated auxiliary field in Rydberg atoms

Rydberg atoms have emerged as a versatile and efficient platform for high-sensitivity quantum sensing of free-space electric fields, with remarkable progress in detecting low-frequency signals. To date, low-frequency Rydberg receivers have relied on a constant bias field, typically realized via intra-cell electrodes or Rydberg plasmas generated by photoelectric effects or inter-atomic interactions. While these approaches improve sensitivity, they suffer from inherent challenges in calibration, long-term stability, and robustness, hindering practical deployment. Here, we propose, design, and experimentally demonstrate a quantum sensing scheme for low-frequency electric signals using modulated auxiliary fields in Rydberg atoms. Unlike conventional methods that employ external DC electric fields that are often fully shielded by adsorbed atom layers on the cell walls, we introduce an AC-field modulation strategy. The incoming low-frequency signal mixes with the auxiliary field, and together they induce Stark shifts of the Rydberg level. These shifts are mapped onto the probe laser via electromagnetically induced transparency (EIT), in a manner analogous to heterodyne detection. We demonstrate a sensitivity of $7.5 \pm 2.6~\mathrm{μV/(cm\cdot Hz^{1/2})}$ at 5 kHz and a minimal detectable field of $0.26 \pm 0.04~\mathrm{μV/cm}$ with an integration time of 1000 s. Furthermore, we extend this approach to systematically analyze the performance of generalized auxiliary fields containing multiple frequency components. By virtue of modulated auxiliary field and quantum frequency mixing, our results establish a robust and systematic framework for quantum sensing of low-frequency electric fields with Rydberg atoms, offering improved sensitivity, stability, and immunity to environmental drifts.

quant-ph↗

arXiv2026

Hierarchical Dual-Strategy Unlearning for Biomedical and Healthcare Intelligence Using Imperfect and Privacy-Sensitive Medical Data

Large language models (LLMs) exhibit exceptional performance but pose substantial privacy risks due to training data memorization, particularly within healthcare contexts involving imperfect or privacy-sensitive patient information. We present a hierarchical dual-strategy framework for selective knowledge unlearning that precisely removes specialized knowledge while preserving fundamental medical competencies. Our approach synergistically integrates geometric-constrained gradient updates to selectively modulate target parameters with concept-aware token-level interventions that distinguish between preservation-critical and unlearning-targeted tokens via a unified four-level medical concept hierarchy. Comprehensive evaluations on the MedMCQA (surgical) and MHQA (anxiety, depression, trauma) datasets demonstrate superior performance, achieving an 82.7% forgetting rate and 88.5% knowledge preservation. Notably, our framework maintains robust privacy guarantees while requiring modification of only 0.1% of parameters, addressing critical needs for regulatory compliance, auditability, and ethical standards in clinical research.

cs.LG↗