arXiv subjects

Hui Pan

Publications and source records attributed to Hui Pan.

At least 19 recordsLinked to original sources

arXiv2026

Analytical Floquet Quantum Statistics from Nonequilibrium Green's Functions

We derive an analytical expression for the steady-state quantum statistics of periodically driven quantum systems coupled to a bath using the nonequilibrium Green's function (NEGF) formalism. By embedding Floquet theory into NEGF, we obtain closed expressions for the retarded, advanced, and lesser Green's functions in the Floquet representation, yielding the Floquet Fermi distribution in which the steady-state occupation is expressed as a weighted sum of Fermi functions shifted by integer multiples of the driving frequency. The weights are determined solely by the Fourier components of the micromotion operator, providing a transparent interpretation of Floquet sideband occupations. Our analysis extends beyond the diagonal commuting Hamiltonians treated in earlier work, and further shows that the robust Floquet distribution remains valid for a broad class of weakly coupled bath spectral functions beyond the ideal featureless-bath approximation. Finally, we establish a Floquet version of the Landauer formula for the DC part of the current, in which the equilibrium Fermi functions are replaced by their Floquet-modified counterparts. Together, these results provide a coherent description of Floquet quantum statistics and transport in periodically driven open quantum systems.

cond-mat.stat-mech↗

arXiv2026

RUBRIC: Realism--Utility Balanced Ranking for Imbalanced Classification

Class imbalance poses a fundamental challenge in risk-sensitive applications such as fraud detection and medical diagnosis, where minority-class samples are scarce yet critical for accurate classification. Existing oversampling methods generate synthetic samples to rebalance class distributions; however, they often produce large numbers of low-quality candidates that distort decision boundaries or introduce artifacts, leading to overfitting and degraded generalization. In this work, we introduce RUBRIC, a generator-agnostic filtering framework that formulates synthetic sample selection as a quality-over-quantity optimization problem. RUBRIC ranks candidates using a realism-utility trade-off: realism is quantified by a learned discriminator that distinguishes real samples from synthetic samples, while utility captures proximity to the decision boundary through a concave margin-based scoring function. We show that, under mild regularity conditions, the proposed filtering strategy monotonically tightens the generalization bound for margin-based classifiers by jointly reducing distribution shift and suppressing near-negative tail contributions. Through extensive experiments on credit-card fraud detection and other imbalanced benchmarks, we demonstrate that RUBRIC improves F1-macro and recall while maintaining comparable ROC-AUC across several generators. We also provide explicit lambda-sensitivity analysis to show how users can recover AUPRC when ranking quality is prioritized.

cs.LG↗

arXiv2026

ParaTutor: Coordinating Parent and Child Math Tutoring through Role Separated LLM Scaffolding

Parent and child tutoring is a collaborative learning setting with asymmetric roles. Parents guide children s problem solving, while children are expected to remain actively engaged in understanding and reasoning. However, most LLM based learning systems are designed for single users or relatively symmetric collaboration, leaving parent and child tutoring with distinct instructional roles underexplored. Through a formative study, we found that parent and child math tutoring was often disrupted by cognitive misalignment, emotional escalation, and method mismatch. To address these challenges, we present ParaTutor, a multiple agents LLM based scaffolding system for home math word problem tutoring. ParaTutor distributes support across user roles by providing parents with strategy, language, repair, and phase scaffolds, while providing children with visual grounding for problem interpretation. We evaluated ParaTutor with 23 parent and child dyads (children aged 10 to 12) across four tutoring conditions that varied how LLM assistance was delivered. Results show that generic LLM assistance often provided useful explanations but did not consistently support parent led tutoring or children s active reasoning. In contrast, ParaTutor helped redistribute tutoring work across parents and children, increased children s engagement with word problems, supported shared understanding through visual grounding, and helped parents translate LLM generated methods into child facing tutoring moves. These findings suggest that in family learning, the value of LLM support depends not only on model capability, but also on how support is coordinated across users with different roles. Our work contributes design implications for LLM systems that support role sensitive scaffolding in parent and child learning.

cs.HC↗

arXiv2026

Amortized Reasoning Tree Search: Decoupling Proposal and Decision in Large Language Models

Reinforcement Learning with Verifiable Rewards (RLVR) has established itself as the dominant paradigm for instilling rigorous reasoning capabilities in Large Language Models. While effective at amplifying dominant behaviors, we identify a critical pathology in this alignment process: the systematic suppression of valid but rare (low-likelihood under the base model distribution) reasoning paths. We theoretically characterize this phenomenon as a "Normalization Squeeze," where the interplay between mode-seeking policy gradients and finite sampling acts as a high-pass likelihood filter, driving the probability of rare correct traces to statistical extinction. To counteract this collapse without discarding the base model's latent diversity, we propose Amortized Reasoning Tree Search (ARTS). Unlike standard approaches that force internalization via parameter updates, ARTS prioritizes deliberation by decoupling generation from verification. We introduce a Flow Matching objective that repurposes the verifier to estimate the conservation of probability flow, enabling robust navigation through sparse, high-entropy search spaces where traditional discriminative objectives fail. Extensive experiments on the MATH-500 benchmark demonstrate that ARTS achieves a performance of 74.6% (BoN@16), effectively matching fully fine-tuned policies (74.7%) without modifying the generative backbone. Crucially, on the long-tail subset where coupled RL optimization collapses to 0% pass@k, ARTS uniquely recovers significant performance, suggesting that disentangling verification from generation offers a more robust pathway for solving complex reasoning tasks.

cs.LG↗

arXiv2025

Clarification of Floquet--Enhanced Thermal Emission Through the Nonequilibrium Green's Function Formalism

Floquet engineering offers a powerful route to enhance emission in time-modulated media. Here, we investigate the influence of time-modulated permittivity in silicon carbide on its intensity spectrum. We consider both the nonequilibrium Green's function approach and the macroscopic quantum electrodynamics approach, and establish their formal compatibility by deriving the Lippmann-Schwinger equation in both cases. To analyze spectral features, we propose several methods for decomposing the electric field into positive- and negative-frequency components, along with the criteria required for physical consistency. Our analytical and numerical results show that, when defined appropriately, the intensity spectrum avoids divergence, though the resulting enhancement remains modest. These findings provide a unified theoretical foundation for modeling time-dependent media, and reinforce the utility of Floquet engineering as a versatile platform for tailoring emission dynamics.

cond-mat.mes-hall↗

arXiv2025

Enhancing far-field thermal radiation by Floquet engineering

Time modulation introduces a dynamic degree of freedom for tailoring thermal radiation beyond the limits of static materials. Here we investigate far-field thermal radiation from a periodically time-modulated SiC film under the Floquet nonequilibrium Green's function framework. We show that time modulation enables radiative energy transfer into the far field that surpasses the limit imposed by the equilibrium thermal fluctuations. This enhancement originates from the modulation-induced coupling between evanescent surface phonon polaritons and propagating modes, effectively bridging the energy and momentum mismatch through frequency conversion. Notably, even at zero temperature, the film emits a finite radiative heat flux due to nonequilibrium photon occupation generated by the modulation. The radiative output grows with increasing modulation strength, highlighting the role of external work in driving far-field emission. These results establish time modulation as an effective mechanism for bridging near-field and far-field regimes, opening new pathways for active thermal radiation control.

cond-mat.mes-hall↗

arXiv2025

Anomalous Scaling Laws of Dispersion Interactions in Anisotropic Nanostructures

The van der Waals (vdW) dispersion interaction between two finite neutral objects typically follows the standard nonretarded $d^{-6}$ law. Here, we reveal an anomalous $d^{-10}$ scaling law between nanostructures with strong geometric or electric anisotropy, driven intrinsically by symmetry-restricted plasmon interactions. At finite anisotropy ratios, a scaling crossover from $d^{-10}$ to $d^{-6}$ occurs due to plasmon mode competition, marked by a finite critical separation. Furthermore, we demonstrate tunability of interlayer vdW forces in two-dimensional materials with strong in-plane electronic anisotropy. By pushing the conventional lower bound of vdW scaling laws, these findings open new opportunities for tailoring nanoscale forces, with potential applications in low-stiction nanomechanical devices, vdW superstructure assembly, metamaterials, and molecular simulations.

cond-mat.mes-hall↗

arXiv2025

Comparative Analysis of Pre-trained Deep Learning Models and DINOv2 for Cushing's Syndrome Diagnosis in Facial Analysis

Cushing's syndrome is a condition caused by excessive glucocorticoid secretion from the adrenal cortex, often manifesting with moon facies and plethora, making facial data crucial for diagnosis. Previous studies have used pre-trained convolutional neural networks (CNNs) for diagnosing Cushing's syndrome using frontal facial images. However, CNNs are better at capturing local features, while Cushing's syndrome often presents with global facial features. Transformer-based models like ViT and SWIN, which utilize self-attention mechanisms, can better capture long-range dependencies and global features. Recently, DINOv2, a foundation model based on visual Transformers, has gained interest. This study compares the performance of various pre-trained models, including CNNs, Transformer-based models, and DINOv2, in diagnosing Cushing's syndrome. We also analyze gender bias and the impact of freezing mechanisms on DINOv2. Our results show that Transformer-based models and DINOv2 outperformed CNNs, with ViT achieving the highest F1 score of 85.74%. Both the pre-trained model and DINOv2 had higher accuracy for female samples. DINOv2 also showed improved performance when freezing parameters. In conclusion, Transformer-based models and DINOv2 are effective for Cushing's syndrome classification.

cs.LG↗

arXiv2024

MobileNetV2: A lightweight classification model for home-based sleep apnea screening

This study proposes a novel lightweight neural network model leveraging features extracted from electrocardiogram (ECG) and respiratory signals for early OSA screening. ECG signals are used to generate feature spectrograms to predict sleep stages, while respiratory signals are employed to detect sleep-related breathing abnormalities. By integrating these predictions, the method calculates the apnea-hypopnea index (AHI) with enhanced accuracy, facilitating precise OSA diagnosis. The method was validated on three publicly available sleep apnea databases: the Apnea-ECG database, the UCDDB dataset, and the MIT-BIH Polysomnographic database. Results showed an overall OSA detection accuracy of 0.978, highlighting the model's robustness. Respiratory event classification achieved an accuracy of 0.969 and an area under the receiver operating characteristic curve (ROC-AUC) of 0.98. For sleep stage classification, in UCDDB dataset, the ROC-AUC exceeded 0.85 across all stages, with recall for Sleep reaching 0.906 and specificity for REM and Wake states at 0.956 and 0.937, respectively. This study underscores the potential of integrating lightweight neural networks with multi-signal analysis for accurate, portable, and cost-effective OSA screening, paving the way for broader adoption in home-based and wearable health monitoring systems.

cs.LG↗

arXiv2024

Asymmetry-induced radiative heat transfer in Floquet systems

Time modulation opens new avenues for light, heat control, and energy harvesting, yet the impact of nonequilibrium dynamics of microscopic particles remains largely unexplored. We develop a microscopic theory to describe radiative heat transfer in such Floquet systems. Significant heat transfer occurs due to differences in electronic properties between parallel metal plates, despite identical driving protocols and temperatures. This arises from a unique exponential-staircase distribution of radiative photons, induced by nonequilibrium electronic fluctuations, and can be tuned via both microscopic properties and driving parameters. Our work highlights the importance of nonequilibrium microscopic details, unlocking new opportunities for active cooling, thermophotovoltaics, thermal imaging and manipulation, and carrier dynamics probing.

cond-mat.mes-hall↗

arXiv2024

Large Bulk Photovoltaic Effect of Nitride Perovskite LaWN3 as Photocatalyst for Hydrogen Evolution Reaction: First Principles Calculation

Bulk photovoltaic effect in noncentrosymmetric materials is a fundamental and significant property that holds potential for high-efficiency energy harvesting, such as photoelectric application and photocatalysis. Here, based on first principles calculation, we explore the electronic structure, dielectric property, shift current, and photocatalytic performance of novel nitride perovskite LaWN3. Our calculations show that LaWN3 possesses large dielectric constants and shift current. The shift current can be enhanced by considering spin-orbit coupling and is switchable by ferroelectric polarization, which suggests LaWN3 is a promising candidate for logic and neuromorphic photovoltaic devices driven by ferroelectric polarization. Additionally, LaWN3 shows advanced photocatalytic hydrogen evolution reaction as a photocatalyst. Especially, the (110) surface represents low surface energy and Gibbes free energy, implying that the (110) surface may be exposed to the active surface. Our finding highlights potential applications of novel polar nitride perovskite LaWN3 in various fields not only photoelectric devices but also photocatalysis.

cond-mat.mtrl-sci↗

arXiv2024

Semantic-Rearrangement-Based Multi-Level Alignment for Domain Generalized Segmentation

Domain generalized semantic segmentation is an essential computer vision task, for which models only leverage source data to learn the capability of generalized semantic segmentation towards the unseen target domains. Previous works typically address this challenge by global style randomization or feature regularization. In this paper, we argue that given the observation that different local semantic regions perform different visual characteristics from the source domain to the target domain, methods focusing on global operations are hard to capture such regional discrepancies, thus failing to construct domain-invariant representations with the consistency from local to global level. Therefore, we propose the Semantic-Rearrangement-based Multi-Level Alignment (SRMA) to overcome this problem. SRMA first incorporates a Semantic Rearrangement Module (SRM), which conducts semantic region randomization to enhance the diversity of the source domain sufficiently. A Multi-Level Alignment module (MLA) is subsequently proposed with the help of such diversity to establish the global-regional-local consistent domain-invariant representations. By aligning features across randomized samples with domain-neutral knowledge at multiple levels, SRMA provides a more robust way to handle the source-target domain gap. Extensive experiments demonstrate the superiority of SRMA over the current state-of-the-art works on various benchmarks.

cs.CV↗

arXiv2020

Carrier-potential interaction for high-Tc superconductivity

The origin of high-temperature superconductivity has been widely debated since its discovery. Here, we propose a model to reveal the mechanism based on the interaction between carrier and local potential. In this model, the potential that is analogous to the lattice point is composed of localized charges and its vibration mediates the coupling of mobile carriers. A Hamiltonian that describes the vibration, coupling, and various interactions among the ordered potentials and carriers is established. By analyzing the Hamiltonian, we find that the vibration of local potential and the interactions, which are determined by the carrier density, control the transition temperature. We show that the transition temperature is high if the local potential is composed of electrons and the mobile carrier is hole because of the strong coupling between them. By replacing the local potential with lattice point, our model is equivalent to the BCS theory. Therefore, our model may provide a general theoretical description on the superconductivity.

cond-mat.supr-con↗

arXiv2017

The process of 3D-printed skull models for the anatomy education

Objective The 3D printed medical models can come from virtual digital resources, like CT scanning. Nevertheless, the accuracy of CT scanning technology is limited, which is 1mm. In this situation, the collected data is not exactly the same as the real structure and there might be some errors causing the print to fail. This study presents a common and practical way to process the skull data to make the structures correctly. And then we make a skull model through 3D printing technology, which is useful for medical students to understand the complex structure of skull. Materials and Methods The skull data is collected by the CT scan. To get a corrected medical model, the computer-assisted image processing goes with the combination of five 3D manipulation tools: Mimics, 3ds Max, Geomagic, Mudbox and Meshmixer, to reconstruct the digital model and repair it. Subsequently, we utilize a low-cost desktop 3D printer, Ultimaker2, with polylactide filament (PLA) material to print the model and paint it based on the atlas. Result After the restoration and repairing, we eliminate the errors and repair the model by adding the missing parts of the uploaded data within 6 hours. Then we print it and compare the model with the cadaveric skull from frontal, left, right and anterior views respectively. The printed model can show the same structures and also the details of the skull clearly and is a good alternative of the cadaveric skull.

cs.OH↗

arXiv2017

Trigonal warping induced terraced spin texture and nearly perfect spin polarization in graphene with Rashba effect

Electrical tunability of spin polarization has been a focus in spintronics. Here, we report that the trigonal warping (TW) effect, together with spin-orbit coupling (SOC), can lead to two distinct magnetoelectric effects in low-dimensional systems. Taking graphene with Rashba SOC as example, we study the electronic properties and spin-resolved scattering of system. It is found that the TW effect gives rise to a terraced spin texture in low-energy bands and can render significant spin polarization in the scattering, both resulting in an efficient electric control of spin polarization. Our work unveils not only SOC but also the TW effect is important for low-dimensional spintronics.

cond-mat.mtrl-sci↗

arXiv2016

Fullerene antiferromagnetic reconstructed spinterface subsurface layer dominates multi-orbitals spin-splitting and large magnetic moment in C60

The interfaces between organic molecules and metal surfaces with layered antiferromagnetic order have gained increasing interests in the field of antiferromagnetic spintronics. The C60 layered AFM spinterfaces have been studied for C60 bonded only to the outermost ferromagnetic layer. Using density functional theory calculations, here we demonstrate that C60 adsorption can reconstruct the layered AFM Cr(001) surface so that C60 bonds to the top two Cr layers with opposite spin direction. Surface reconstruction drastically changes C60 s spintronic properties 1 the spin-split p-d hybridization involve multi-orbitals of C60 and metal double layers, 2 the subsurface layer dominates the C60 spin properties, and 3) reconstruction induces a large magnetic moment in C60 of 0.58 B, which is a synergetic effect of the top two layers as a result of a magnetic direct-exchange interaction. Understanding these complex spinterfaces phenomena is a crucial step for their device applications. The surface reconstruction can be realized by annealing at above room temperature in experiments.

physics.comp-ph↗

arXiv2015

Quantum anomalous Hall effect in atomic crystal layers from in-plane magnetization

We theoretically report that, with \textit{in-plane} magnetization, the quantum anomalous Hall effect (QAHE) can be realized in two-dimensional atomic crystal layers with preserved inversion symmetry but broken out-of-plane mirror reflection symmetry. We take the honeycomb lattice as an example, where we find that the low-buckled structure, which makes the system satisfy the symmetric criteria, is crucial to induce QAHE. The topologically nontrivial bulk gap carrying a Chern number of $\mathcal{C}=\pm1$ opens in the vicinity of the saddle points $M$, where the band dispersion exhibits strong anisotropy. We further show that the QAHE with electrically tunable Chern number can be achieved in Bernal-stacked multilayer systems, and the applied interlayer potential differences can dramatically decrease the critical magnetization to make the QAHE experimentally feasible.

cond-mat.mes-hall↗

arXiv2015

Exploration of CPT violation via time-dependent geometric quantities embedded in neutrino oscillation through fluctuating matter

We propose a new approach to explore CPT violation of neutrino oscillation through a fluctuating matter based on time-dependent geometric quantities. By mapping the neutrino oscillation onto a Poincare sphere structure, we obtain an analytic solution of master equation and further define the geometric quantities, i.e., radius of Poincare sphere and geometric phase. We find that the mixing process between electron and muon neutrinos can be described by the radius of Poincare sphere that depends on the intrinsic CP-violating angle. Such a radius reveals a dynamic mechanism of CPT-violation, i.e., both spontaneous symmetry breaking and Majorana-Dirac neutrino confusion. We show that the time-dependent geometric phase can be used to find the neutrino nature and observe the CPT-violation because it is strongly enhanced under the neutrino propagation. We further show that the time-dependent geometric phase can be easily detected by simulating the neutrino oscillation based on fluctuating magnetic fields in nuclear magnetic resonance, which makes the experimental observation of CPT-violation possible in the neutrino mixing and oscillation.

quant-ph↗