arXiv subjects

Wenxin Wu

Publications and source records attributed to Wenxin Wu.

4 recordsLinked to original sources

arXiv2026

Hallucination-R1: Robustness-Oriented Paraphrase Generation for Factual Consistency

Factual hallucination is commonly defined by incorrect factual outputs. We study a paraphrase-induced hallucination setting, where a model answers a factual question correctly in its original form but generates an incorrect answer under a semantically equivalent paraphrase. Such inconsistencies expose latent factual instability under semantic invariance. However, general-purpose paraphrases are often insufficient as robustness-oriented supervision: near-copy paraphrases provide weak signals, while overly diverse paraphrases may break semantic equivalence. In this paper, we propose HALLUCINATION-R1, a robustness-oriented paraphrase generation framework that learns to produce semantically faithful yet robustness-challenging paraphrases for factual consistency. Through two-stage optimization, it first stabilizes meaning-preserving and diverse paraphrasing, then rewards paraphrases that reveal factual consistency degradation in downstream QA models. Experiments on SimpleQuestions, PopQA, and TruthfulQA show that HALLUCINATION-R1 achieves a strong consistency--diversity trade-off and exposes robustness failures across multiple model families and datasets. Further analyses indicate that these failures are not reducible to surface-level artifacts or semantic drift, but reveal non-trivial factual instability under meaning-preserving variation. A lightweight fine-tuning study also shows that HALLUCINATION-R1-generated data improves robust accuracy under paraphrase variations, suggesting its utility for robustness-oriented training. Our code and models are publicly available at https://github.com/yuwenhan07/Hallucination-R1.

cs.CL↗

arXiv2025

Efficient simulation framework for modeling collective emission in ensembles of inhomogeneous solid-state emitters

An efficient simulation framework is proposed to model collective emission in disordered ensembles of quantum emitters. Using a cumulant expansion approach, the computational complexity scales polynomially as opposed to exponentially with the number of emitters, enabling Monte Carlo sampling over a large number of realizations. The framework is applied to model negatively charged silicon-vacancy (SiV$^{-}$) centers inside diamond. Incorporating spatial disorder and inhomogeneous broadening, we obtain statistically averaged responses over hundreds of SiV$^{-}$ clusters. These simulations reveal two signatures of collective behavior. First, dynamics of fully inverted clusters show that superradiant emission occurs only with sufficiently large emitter number and high quantum efficiency. Unlike ideal Dicke superradiance, the burst is substantially suppressed by strong near-field dipole-dipole interaction, consistent with existing theoretical predictions. Second, under continuous-wave excitation we compute photoluminescence-excitation spectra, which exhibit interaction-induced broadening in the distribution of resonance peaks. The corresponding density of states also displays a non-zero skewness. Overall, by incorporating realistic inhomogeneities in emitter clusters, our framework is able to predict statistics for disordered ensembles that can be compared to experiments directly. Our approach generalizes to other types of emitters, including atoms, molecules, and quantum dots, thus providing a practical tool for analyzing collective behavior in realistic quantum systems.

physics.optics↗

arXiv2025

Wafer-Scale All-Dielectric quasi-BIC Metasurfaces: Bridging High-throughput Deep-UV Lithography with Nanophotonic Applications

High quality-factor (Q) dielectric metasurfaces operating in the visible to near-infrared range usually require sub-200 nm features, limiting their fabrication to expensive, low-throughput electron beam lithography. Here, we demonstrate wafer-scale metasurfaces fabricated using deep ultraviolet lithography (DUVL), a workhorse technology in the semiconductor industry. Using a radius and depth perturbation technique in a hole array patterned into a silicon nitride slab, we achieve quasi-bound states in the continuum (qBIC) resonances with measured Q-factors of 150. Critically, we introduce DUV exposure dose as a Q-factor engineering parameter and demonstrate how hole depth control circumvents DUVL resolution limits. Despite stochastic nanoscale variations, the fabricated metasurfaces exhibit spatial uniformity, a consequence of the nonlocal nature of the qBIC resonances. Proof of concept refractive index sensing demonstrates 129 nm/RIU sensitivity while maintaining simple CMOS camera-based resonance shift interrogation. This work bridges scalable semiconductor manufacturing with high-performance nanophotonics, establishing a practical pathway for commercializing metasurface-based sensors, on-chip spectrometers, and integrated photonic systems.

physics.optics↗

arXiv2025

Broadband light extraction from near-surface NV centers using crystalline-silicon antennas

We use crystalline silicon (Si) antennas to efficiently extract broadband single-photon fluorescence from shallow nitrogen-vacancy (NV) centers in diamond into free space. Our design features relatively easy-to-pattern high-index Si resonators on the diamond surface to boost photon extraction by overcoming total internal reflection and Fresnel reflection at the diamond-air interface, and providing modest Purcell enhancement, without etching or otherwise damaging the diamond surface. In simulations, ~17 times more single photons are collected from a single NV center compared to the case without the antenna; in experiments, we observe an enhancement of ~9 times, limited by spatial alignment between the NV and the antenna. Our approach can be readily applied to other color centers in diamond, and more generally to the extraction of light from quantum emitters in wide-bandgap materials.

physics.optics↗