arXiv subjects

Han Gao

Publications and source records attributed to Han Gao.

At least 19 recordsLinked to original sources

arXiv2026

Enhancing charge stability of Ge quantum well heterostructures via SiGe layer composition engineering

Composition modulation is a powerful technique for designing materials with tailored properties, fueling the development of advanced semiconductor devices. In this work, we have implemented this technique into Ge quantum well heterostructures, offering a promising avenue to address the critical challenge of charge stability in spin qubit devices. Harnessing the atomic-scale precision of molecular beam epitaxy, we have engineered the band structure of the SiGe top barrier via graded composition modulation, thereby reducing charge accumulation states at the SiGe-dielectric interface and strengthening the effective confinement to the hole gases in the Ge quantum wells. The enhanced charge stability of composition-modulated SiGe/Ge quantum well heterostructures is confirmed in Hall devices, featuring an enlarged stable gate voltage range. We have further fabricated quantum dot devices from the composition-modulated SiGe/Ge quantum well heterostructures and observed remarkably low charge noise with an averaged amplitude of $0.46\,\mathrm{\mu eV}/\mathrm{\sqrt{Hz}}$ at $1\,\mathrm{Hz}$---the lowest reported value for Ge quantum wells grown on silicon. This exceptional charge stability of the quantum dots persists in the few-hole regime, with no observable voltage drift over $\sim$hours. With reduced charge noise and enhanced energy stability, composition-modulated SiGe/Ge heterostructures exhibit significant potential for applications in building high-performance quantum devices, including spin qubits with a long coherence time.

cond-mat.mes-hall↗

arXiv2026

Viscoelasticity reshapes the frequency response of a rotating magnetic particle

A magnetic particle driven by a rotating magnetic field undergoes a transition from synchronous to asynchronous rotation at a critical driving frequency. The asynchronous dynamics is well understood in Newtonian fluids but remains unclear in viscoelastic media. Here, we develop a theoretical description of the asynchronous rotation of a magnetic particle in a Jeffreys-type viscoelastic fluid. The particle's time-averaged angular velocity exhibits a nontrivial frequency dependence that changes from non-monotonic to monotonic as the polymer relaxation time increases. This behavior is explained by the interplay among magnetic driving, viscoelastic relaxation, and frequency-dependent viscous dissipation. We further derive an asymptotic expression that captures the non-monotonic dependence. These results clarify how solvent and polymer contributions jointly control asynchronous rotation and provide a physical basis for guiding relevant applications in complex fluids.

cond-mat.soft↗

arXiv2026

Microwave Response of the Superconducting Diode Effect in Proximitized Bilayer Graphene Interferometers

Microwave irradiation has emerged as a promising means to tune the superconducting diode effect (SDE) in Josephson junction devices. Previous experimental studies have mainly focused on the adiabatic-driving regime, in which the diode efficiency increases monotonically with microwave power and can approach the ideal value of unity. Beyond this regime, however, the microwave response of the SDE remains largely unexplored experimentally. In this work, we investigate the microwave response of the SDE in bilayer-graphene-based superconducting quantum interference devices (SQUIDs) under a broad range of driving frequencies. We show that increasing the driving frequency changes the response characteristics of the diode efficiency to microwave power--the dependence of the diode efficiency evolves from monotonic enhancement with increasing microwave power in the adiabatic regime to non-monotonic behavior beyond this regime, and ultimately to sign-reversal as well oscillatory characteristics at sufficiently high frequencies. We find that these experimentally observed frequency-dependent power response characteristics of the diode efficiency can be qualitatively captured by simulations based on the resistively shunted junction model using the device current-phase relations extracted from the experiments. These results establish SQUIDs made from bilayer graphene as a versatile platform for studying dynamic properties of superconducting junction devices.

cond-mat.mes-hall↗

arXiv2026

Dynamics of a microroller under confinement

Rotating particles can translate when placed near a surface, forming microrollers with a wide range of biomedical and microfluidic applications. In this work, we investigate the dynamics of microrollers in confined microchannels with different geometries by combining experiments, numerical simulations, and scaling analysis. In constricted channels, we find that the translational velocity of a microroller decreases as it approaches the constricted region. In both rectangular and cylindrical channels, velocity reversal occurs as the characteristic channel width decreases. Using the force-free condition for free translation, we develop a systematic scaling framework that can be generalized to different channel geometries. The scaling analysis yields functional dependences of the translational velocity on the degree of confinement, which agree well with both experiments and simulations. Importantly, we demonstrate that the viscous stress generated by the far-field rotlet flow governs the observed velocity reduction and reversal, while the translational resistance resulting from the near-field shear flow suppresses translation under tight confinement. The distinct roles of these flow components revealed by our analysis may provide practical guidance for controlling microroller dynamics in confined fluid environments.

physics.flu-dyn↗

arXiv2026

Layer-Parallel Inference Reduces Encrypted Nonlinear Depth in Transformers

Fully homomorphic encryption (FHE) enables computation on encrypted data, but practical encrypted Transformer inference is bottlenecked by the sequential composition of many nonlinear blocks. We study whether Structured Newton Layer Parallelism (SNLP) can make this inter-layer composition more FHE-friendly: each Transformer block still requires polynomial approximations for operations such as softmax and RMSNorm, but SNLP reduces the layerwise sequential nonlinear depth from L stages to a small number of solver iterations plus linear structured corrections. Using a simulation framework based on Chebyshev polynomial approximations, we measure error accumulation under sequential versus SNLP inference across 8 models and 4 architecture families. On a 0.5B IDN-trained model, SNLP reduces symbolic bootstraps from 53 to 20 (2.65x) with only +1.2% perplexity degradation, while lowering error amplification (1.36x vs. 1.42x). Across all tested models, SNLP has lower amplification than sequential inference. Ablations show that softmax approximation dominates the error budget and CKKS arithmetic noise is negligible in our setting, suggesting that SNLP is complementary to block-level FHE-friendly operator design rather than a replacement for it.

cs.LG↗

arXiv2026

PhysMiner: An Agentic AI Framework for Discovering Turbulence Physics

Uncovering the physical mechanisms of turbulent flows remains a fundamental challenge in fluid mechanics. In particular, conventional velocity-gradient analysis methods suffer from shear contamination, which hinders accurate identification of the dominant physical mechanisms. This study presents PhysMiner, an automated framework integrating the triple decomposition method of the velocity gradient tensor with large language model-driven reasoning for turbulence-physics discovery. The triple decomposition module automatically decomposes flow fields into rigid rotation, pure shearing, and normal straining components, enabling statistical analysis, contour visualization, vortex-line extraction, and threshold-insensitive vortex identification while eliminating shear contamination. These automated capabilities are validated across five benchmarks, ranging from canonical configurations to complex engineering flows. A discover-physics agent combines flow statistics, spatial structures, and literature-derived knowledge to perform pattern recognition and physical inference, while a review Agent iteratively validates physical consistency to ensure reliable conclusions. A continuously evolving Triple Decomposition Library accumulates statistical knowledge from successfully analyzed flows, enabling cross-case comparison and progressive enhancement of inductive capability. The complete PhysMiner pipeline is validated end-to-end on the periodic hill flow, where the framework autonomously generates turbulence modeling recommendations and derives an improved subgrid-scale model with superior Reynolds-stress predictions. PhysMiner is open to the public and establishes a foundation for long-term collaborative advancement in automated turbulence-physics discovery.

physics.flu-dyn↗

arXiv2026

SLIM-RL: Risk-Budgeted Random-Masking RL for Diffusion LLMs Without Trajectory Slicing

Reinforcement learning for diffusion large language models (dLLMs) has largely moved to trajectory-aware methods. The current state of the art, TraceRL, holds that random masking is mismatched with the model's inference trajectory, and it reconstructs that trajectory during training by slicing each rollout into up to K/s trajectory-aligned training samples, a cost that grows with the block size K. We show that this mismatch can be mitigated without reconstructing the trajectory. Our method, SLIM-RL, bounds the commit risk of each rollout step with a tau-budget decoder, reducing aggregate commit risk in the training data. During optimization, SLIM-RL trains on these risk-controlled rollouts with a trace-free random-masking objective that adapts variance-reduction tools, combining sequence-level importance sampling, deterministic quadrature over masking levels under a mean-preserving, monotonically decreasing per-block mask schedule that we introduce. On SDAR-4B, SLIM-RL matches TraceRL's best MATH500 accuracy on only 0.46x its training samples at block size 16, improving over TraceRL by 6.32% on MATH500 and 11.05% on GSM8K under matched dynamic sampling. At block size 4, the 4B SLIM-RL surpasses the larger LLaDA-8B and Dream-7B dLLMs on math, exceeding LLaDA-8B by 10.76% on MATH500 while staying below the autoregressive Qwen2.5-7B. On code, it improves over TraceRL by 4.20% on MBPP and 3.65% on HumanEval. The tau-budget decoder transfers training-free across LLaDA, Dream, and SDAR. The source code is available at https://github.com/laolaorkkkkk/SLIM-RL .

cs.CL↗

arXiv2026

Taming Complexity in Intuitionistic Modal Logic: The Case of FIK and Its Shallow Calculus

Intuitionistic modal logics (IMLs) comprise many systems: from constructive modal logics such as CK and Wijesekera's CCDL to Fischer Servi/Simpson's IK, as well as some recently introduced variants. All of them are characterized by bi-relational semantics and have complete axiomatisations. However, from the perspective of proof theory and complexity, there are strong differences: while for constructive modal logics simple Gentzen calculi suffice, for IK more complex calculi, based on nested or labelled sequents, are needed. As a consequence, the decision problem for constructive modal logics has a PSPACE upper bound, whereas for IK is not known and it is even conjectured to be non-elementary. We study here the proof theory and complexity of FIK, a natural intuitionistic modal logic recently introduced. FIK is strictly in between CCDL and IK, yet it has the same forcing conditions as IK. We define a "shallow" sequent calculus for FIK which is a nested sequent calculus where sequents have at most one level of nesting. We prove its syntactic completeness by showing the admissibility of cut. By means of this calculus we show that decision problem for FIK is in EXPSPACE, whence significantly lower than the complexity conjectured for IK.

cs.LO↗

arXiv2026

Single-Crystalline Al/Ge Heterostructure with an Atomically Sharp Commensurate Interface

A key challenge in developing Al/Ge heterostructures for quantum applications is Al-Ge interdiffusion. This process is facilitated by grain boundaries in polycrystalline films, which degrades interface quality and impairs device performance and reliability. Here, we present epitaxial growth of single-crystalline Al(111) on Ge(111) by molecular beam epitaxy, achieving an atomically flat and sharp interface. At the interface, a commensurate 7-Al-lattice/5-Ge-lattice epitaxial relationship is observed, which dramatically reduces the intrinsic lattice mismatch from 28.4% to about 0.1%. Interestingly, this well-ordered interface does not form below a critical thickness of 0.3 nm. Instead, Al initially nucleates as random clusters, which then transform into two-dimensional (2D) islands and, as Al deposition further increases, eventually develop into a continuous film. By optimizing the growth parameters, we have achieved an ultra-flat Al film with a surface root-mean-square roughness of about 0.16 nm and an ultra-thin continuous film with thickness of only 2 nm. These epitaxially grown Al-Ge heterostructures, with their atomically flat surfaces and sharp interfaces, provide a promising platform for studying topological quantum states.

cond-mat.mtrl-sci↗

arXiv2026

The polarization of thermal dileptons emitted in high-energy heavy-ion collisions

This work presents calculations of thermal dilepton emission and polarization observables. It features a comprehensive framework which comprises virtual photon spectral functions complete at next-to-leading-order in the strong coupling and iEBE-MUSIC hydrodynamic simulations. The polarization of thermal lepton pairs is shown to be sensitive to in-medium properties of the quark-gluon plasma. We consider Pb+Pb collisions performed in conditions specific to the LHC and examine the magnitude and behaviour of the polarization as measured in different frames, the effects of the pre-equilibrium gluon abundance, and we derive a one-to-one mapping between dielectron and dimuon polarization.

nucl-th↗

arXiv2026

A geometry-aligned multi-fidelity framework for uncertainty quantification of wildfire spread

Forward propagation of input uncertainties in physics-based wildfire models is computationally prohibitive, limiting the use of high-fidelity simulators in risk assessment workflows. This work introduces a geometry-aligned bi-fidelity surrogate framework that addresses the convection-dominated nature of wildfire spread by mapping low- and high-fidelity solution snapshots onto a common reference domain prior to basis selection and reconstruction. Unlike conventional bi-fidelity schemes, which combine spatially shifted snapshots and thus suffer from oscillations and excess basis requirements near sharp fronts, the proposed mapping aligns the dominant front geometry through per-variable shift/stretch transforms in 1D and an activity indicator-based affine alignment in 2D, so that reduced bases compare physically corresponding structures rather than displaced ones. Building on the ADfiRe physics-based simulator, we demonstrate the method on 1D and 2D test cases in which low- and high-fidelity models differ in mesh resolution and physical completeness. Across both settings, the geometry-aligned surrogate reproduces full-field temperature and fuel composition with substantially lower error than its unmapped counterpart, eliminates Gibbs-type oscillations near steep gradients, and recovers high-fidelity probability density functions for key quantities of interest (e.g., maximum temperature, evaporated moisture, and burned area). After offline training, online predictions are roughly three orders of magnitude cheaper than direct high-fidelity evaluation, making the framework a practical building block for many-query uncertainty quantification once the offline cost is amortized over enough queries. We discuss the conditions under which the geometric alignment is most effective, its limitations for non-convex or topologically complex fronts, and the path toward validation against real data.

cs.CE↗

arXiv2026

KAT-Coder-V2 Technical Report

We present KAT-Coder-V2, an agentic coding model developed by the KwaiKAT team at Kuaishou. KAT-Coder-V2 adopts a "Specialize-then-Unify" paradigm that decomposes agentic coding into five expert domains - SWE, WebCoding, Terminal, WebSearch, and General - each undergoing independent supervised fine-tuning and reinforcement learning, before being consolidated into a single model via on-policy distillation. We develop KwaiEnv, a modular infrastructure sustaining tens of thousands of concurrent sandbox instances, and scale RL training along task complexity, intent alignment, and scaffold generalization. We further propose MCLA for stabilizing MoE RL training and Tree Training for eliminating redundant computation over tree-structured trajectories with up to 6.2x speedup. KAT-Coder-V2 achieves 79.6% on SWE-bench Verified (vs. Claude Opus 4.6 at 80.8%), 88.7 on PinchBench (surpassing GLM-5 and MiniMax M2.7), ranks first across all three frontend aesthetics scenarios, and maintains strong generalist scores on Terminal-Bench Hard (46.8) and tau^2-Bench (93.9). Our model is publicly available at https://streamlake.com/product/kat-coder.

cs.CL↗

arXiv2026

S2D2: Fast Decoding for Diffusion LLMs via Training-Free Self-Speculation

Block-diffusion language models offer a promising path toward faster-than-autoregressive generation by combining block-wise autoregressive decoding with within-block parallel denoising. However, in the few-step regime needed for practical acceleration, standard confidence-thresholded decoding is often brittle: aggressive thresholds hurt quality, while conservative thresholds require unnecessary denoising steps. Existing approaches that address this issue either require additional training or incur extra test-time compute. We present S2D2, a training-free self-speculative decoding framework for block-diffusion language models. Our key observation is that a block-diffusion model becomes autoregressive when the block size is reduced to one, allowing the same pretrained model to act as both drafter and verifier. S2D2 inserts a speculative verification step into standard block-diffusion decoding and uses lightweight routing policies to decide when verification is worth its cost. This yields a hybrid decoding trajectory in which diffusion proposes tokens in parallel, while the autoregressive mode acts as a local sequence-level critic. Across three mainstream block-diffusion families, S2D2 consistently improves the accuracy-speed tradeoff over strong confidence-thresholding baselines. On SDAR, we observe up to $4.7\times$ speedup over autoregressive decoding, and up to $1.57\times$ over a tuned dynamic decoding baseline while improving accuracy by up to $4.5$ points. On LLaDA2.1-Mini, S2D2 remains complementary to built-in self-correction, including a conservative setting where it is $4.4\times$ faster than the static baseline with slightly higher accuracy.

cs.CL↗

arXiv2026

Beyond VLM-Based Rewards: Diffusion-Native Latent Reward Modeling

Preference optimization for diffusion and flow-matching models relies on reward functions that are both discriminatively robust and computationally efficient. Vision-Language Models (VLMs) have emerged as the primary reward provider, leveraging their rich multimodal priors to guide alignment. However, their computation and memory cost can be substantial, and optimizing a latent diffusion generator through a pixel-space reward introduces a domain mismatch that complicates alignment. In this paper, we propose DiNa-LRM, a diffusion-native latent reward model that formulates preference learning directly on noisy diffusion states. Our method introduces a noise-calibrated Thurstone likelihood with diffusion-noise-dependent uncertainty. DiNa-LRM leverages a pretrained latent diffusion backbone with a timestep-conditioned reward head, and supports inference-time noise ensembling, providing a diffusion-native mechanism for test-time scaling and robust rewarding. Across image alignment benchmarks, DiNa-LRM substantially outperforms existing diffusion-based reward baselines and achieves performance competitive with state-of-the-art VLMs at a fraction of the computational cost. In preference optimization, we demonstrate that DiNa-LRM improves preference optimization dynamics, enabling faster and more resource-efficient model alignment.

cs.CV↗

arXiv2026

Spin splitting, Kondo correlation and singlet-doublet quantum phase transition in a superconductor-coupled InSb nanosheet quantum dot

We realize a superconductor-coupled quantum dot (QD) in an InSb nanosheet, a 2D platform promising for studies of topological superconductivity. The device consists of a superconductor-QD-superconductor junction, where a bottom bilayer gate defines the QD and allows tuning of its coupling to the superconducting leads. The QD exhibits large $g$-factors and strong spin-orbit coupling. Transport measurements reveal Coulomb diamond-shaped differential conductance features with even-odd alternating sizes and pronounced conductance lines associated with the superconducting gap, confirming a few-electron, superconductor-coupled regime. At an odd electron occupation, Kondo signatures emerge, including a zero-bias peak that splits with magnetic field and is logarithmically suppressed at elevated temperatures. We further observe a doublet-singlet quantum phase transition, manifested by a clear change of Andreev bound states from crossing to anticrossing as the coupling strength increases. These results underscore the rich physics of InSb nanosheet QDs and their promise for topological quantum devices.

cond-mat.supr-con↗

arXiv2026

vLLM-Omni: Fully Disaggregated Serving for Any-to-Any Multimodal Models

Any-to-any multimodal models that jointly handle text, images, video, and audio represent a significant advance in multimodal AI. However, their complex architectures (typically combining multiple autoregressive LLMs, diffusion transformers, and other specialized components) pose substantial challenges for efficient model serving. Existing serving systems are mainly tailored to a single paradigm, such as autoregressive LLMs for text generation or diffusion transformers for visual generation. They lack support for any-to-any pipelines that involve multiple interconnected model components. As a result, developers must manually handle cross-stage interactions, leading to huge performance degradation. We present vLLM-Omni, a fully disaggregated serving system for any-to-any models. vLLM-Omni features a novel stage abstraction that enables users to decompose complex any-to-any architectures into interconnected stages represented as a graph, and a disaggregated stage execution backend that optimizes resource utilization and throughput across stages. Each stage is independently served by an LLM or diffusion engine with per-stage request batching, flexible GPU allocation, and unified inter-stage connectors for data routing. Experimental results demonstrate that vLLM-Omni reduces job completion time (JCT) by up to 91.4% compared to baseline methods. The code is public available at https://github.com/vllm-project/vllm-omni.

cs.DC↗

arXiv2026

Supercurrent and multiple Andreev reflections in Ge hut nanowire Josephson Junctions

We report an experimental study of induced superconductivity in Ge hut nanowire Josephson junctions. The Ge hut nanowires are grown on prepatterned SiGe ridges via molecular beam epitaxy (MBE) and Josephson junction devices are fabricated by contacting the nanowires with Al electrodes. Low-temperature current-bias transport measurements of the Josephson junctions are performed and the measurements show that the devices exhibit gate-tunable supercurrent and excess current. The analysis of excess current indicates that the transparency of the Ge hut nanowire Josephson junctions is as high as 85%. Voltage-bias spectroscopy measurements of the devices show multiple Andreev reflections up to the fourth order. With magnetic field and temperature-dependent measurements of the multiple Andreev reflections, the critical field and the critical temperature of the induced superconductivity in the Josephson junctions are extracted to be ~0.12 T and ~1.4 K. The success in introducing superconductivity into Ge hut nanowires will stimulate their applications in building advanced quantum processors.

cond-mat.mes-hall↗

arXiv2026

CoINS: Counterfactual Interactive Navigation via Skill-Aware VLM

Recent Vision-Language Models (VLMs) have demonstrated significant potential in robotic planning. However, they typically function as semantic reasoners, lacking an intrinsic understanding of the specific robot's physical capabilities. This limitation is particularly critical in interactive navigation, where robots must actively modify cluttered environments to create traversable paths. Existing VLM-based navigators are predominantly confined to passive obstacle avoidance, failing to reason about when and how to interact with objects to clear blocked paths. To bridge this gap, we propose Counterfactual Interactive Navigation via Skill-aware VLM (CoINS), a hierarchical framework that integrates skill-aware reasoning and robust low-level execution. Specifically, we fine-tune a VLM, named InterNav-VLM, which incorporates skill affordance and concrete constraint parameters into the input context and grounds them into a metric-scale environmental representation. By internalizing the logic of counterfactual reasoning through fine-tuning on the proposed InterNav dataset, the model learns to implicitly evaluate the causal effects of object removal on navigation connectivity, thereby determining interaction necessity and target selection. To execute the generated high-level plans, we develop a comprehensive skill library through reinforcement learning, specifically introducing traversability-oriented strategies to manipulate diverse objects for path clearance. A systematic benchmark in Isaac Sim is proposed to evaluate both the reasoning and execution aspects of interactive navigation. Extensive simulations and real-world experiments demonstrate that CoINS significantly outperforms representative baselines, achieving a 17\% higher overall success rate and over 80\% improvement in complex long-horizon scenarios compared to the best-performing baseline

cs.RO↗