arXiv subjects

Wei Du

Publications and source records attributed to Wei Du.

At least 37 records · Page 2Linked to original sources

arXiv2026

Time delay measurements with Broken Power Law model

One challenge in strong gravitational lensing cosmography is the measurement of time delays between multiple lensed images, which are essential for constraining the Hubble constant ($H_0$). In this study, we investigate how assumptions about the lens mass profile affect time-delay measurements in lensing systems. Specifically, we implement a Broken Power Law (BPL) mass model within the \textsc{Lenstronomy} framework (Birrer & Amara 2018), which introduces additional flexibility in the radial mass distribution and can phenomenologically capture deviations from a single power-law profile. This model is combined with a numerical approach to compute time delays at the image positions. We validate the BPL implementation using simulated lens systems and compare the results with those from the elliptical power-law (EPL) model. We then apply both model families to the quadruply imaged quasar WGD~2038--4008. Both models provide good fits to the imaging and kinematic data, with a slight preference for the BPL model. When the internal mass-sheet factor is allowed to vary, the inferred Hubble constant in a flat $Λ$CDM cosmology with fixed $Ω_{\rm m}=0.3$ is $H_0 = 75.3^{+23.1}_{-16.3} \ \mathrm{km \ s^{-1} \ Mpc^{-1}}$ for the BPL model and $H_0 = 61.1^{+19.2}_{-13.2} \ \mathrm{km \ s^{-1} \ Mpc^{-1}}$ for the EPL model. For comparison, in the diagnostic case with the internal mass-sheet factor fixed to unity under the same setup, we obtain $H_0 = 74.2^{+20.3}_{-13.8} \ \mathrm{km \ s^{-1} \ Mpc^{-1}}$ for the BPL model and $H_0 = 66.1^{+18.8}_{-12.8} \ \mathrm{km \ s^{-1} \ Mpc^{-1}}$ for the EPL model. This highlights how time-delay cosmography remains sensitive to assumptions about the lens mass profile. With current precision, this difference does not favor one cosmological scenario over another, but underscores the importance of flexible mass modeling.

astro-ph.CO↗

arXiv2026

EmbTracker: Traceable Black-box Watermarking for Federated Language Models

Federated Language Model (FedLM) allows a collaborative learning without sharing raw data, yet it introduces a critical vulnerability, as every untrustworthy client may leak the received functional model instance. Current watermarking schemes for FedLM often require white-box access and client-side cooperation, providing only group-level proof of ownership rather than individual traceability. We propose EmbTracker, a server-side, traceable black-box watermarking framework specifically designed for FedLMs. EmbTracker achieves black-box verifiability by embedding a backdoor-based watermark detectable through simple API queries. Client-level traceability is realized by injecting unique identity-specific watermarks into the model distributed to each client. In this way, a leaked model can be attributed to a specific culprit, ensuring robustness even against non-cooperative participants. Extensive experiments on various language and vision-language models demonstrate that EmbTracker achieves robust traceability with verification rates near 100\%, high resilience against removal attacks (fine-tuning, pruning, quantization), and negligible impact on primary task performance (typically within 1-2\%).

cs.CR↗

arXiv2026

A Robust Geometric Distortion Solution for Main Survey Camera of CSST

The advancement in sensitivity and field of view of next-generation wide-field survey telescopes requires astrometric measurements with high precision, even in the presence of significant geometric distortions. To address this challenge, we develop a Weighted Polynomial Distortion Correction in 2-Phase (WPDC-2P) method. This approach enhances stellar cross-matching, incorporates distance-based weighting into the traditional polynomial fitting, and employs a look-up table to absorb the remaining distortion residuals. Validated on simulated data from the Main Survey Camera of the \emph{Chinese Space Station Survey Telescope} (CSST), incorporating geometric distortions up to approximately $200$ pixels, the method achieves astrometric standard deviation ranging from 0.013 to 0.107 pixels (0.03 pixels for the $g$-1 detector) across all 18 detectors. Under extreme crowding conditions (e.g., globular cluster NGC 2298), the astrometric precision for the $g$-1 detector reaches 0.05-pixel level within the central region ($r_d < 4000$), despite a centroiding precision of $\sim$0.04 pixels. When applied to the Beijing-Arizona Sky Survey data, for which the standard pipeline delivers an astrometric uncertainty of $\sim$20 mas, our method reduces the positional scatter to $ σ_{Δα}=5.494$ mas (0.01 pixels) and $ σ_{Δδ}=9.981$ mas (0.02 pixels) using only a weighted 3rd-order polynomial correction. The method has been integrated into the CSST data processing pipeline and is prepared for further refinement using on-orbit calibration data.

astro-ph.IM↗

arXiv2026

ProtegoFed: Backdoor-Free Federated Instruction Tuning with Interspersed Poisoned Data

Federated Instruction Tuning (FIT) enables collaborative instruction tuning of large language models across multiple organizations (clients) in a cross-silo setting without requiring the sharing of private instructions. Recent findings on natural backdoors and the existing training data collection method suggest that poisoned samples may be pervasive and inadvertently embedded in real-world datasets, potentially distributed across all clients, even if the clients are benign. This work systematically examine this threat in FIT, demonstrating that existing defenses are ineffective when poisoned data is interspersed among all clients. Addressing this challenge entails two major difficulties: identifying the distinctive characteristics of poisoned samples at each client and enabling collaborative defense when some clients are heavily dominated by poisoned samples. To address these difficulties, we identify gradients in the frequency domain as a robust signal to distinguish poisoned data. We further propose a global secondary clustering mechanism that facilitates collaborative identification of poisoned samples across clients. In summary, this paper introduces ProtegoFed, the first backdoor-free FIT framework that accurately detects, removes, and even purifies interspersed poisoned data across clients during the training. Experimental results on four FL datasets show that ProtegoFed identifies $92.00\% \sim 100.00\%$ of poisoned samples, reduces the attack success rate to almost zero, and maintains utility on the main task. Code is available at https://github.com/dongdongzhaoUP/ProtegoFed.

cs.CR↗

arXiv2026

Systematic study of high performance GeSn photodiodes with thick absorber for SWIR and extended SWIR detection

Germanium-tin (GeSn) photodiodes potentiate a viable solution to integrate SWIR and extended SWIR detection technology into CMOS processing line. However, challenges in the growth of thick, high quality GeSn limit the device absorber thickness, making it impossible to ascertain the performance limit of GeSn photodiodes. An in-depth understanding of their device physics and a clear optimization pathway towards commercial-grade devices remain elusive. This work presents a systematic empirical study of GeSn photodiodes with thick absorber (2 to 8% Sn content, up to 2630 nm thick), showing high responsivity up to 0.59 A.W-1 at 1.55 μm and 0.43 A.W-1 at 2 μm wavelengths, low dark current density down to 2 x 10-2 A.cm-2, and high detection cutoff wavelengths up to 2.1 and 2.5 μm at 5% and 8% Sn, respectively. Using specific doping design (P-i-N and N-i-P), an in-depth analysis is presented on the impact of junction position, p-type background carrier concentration, bulk/ surface defects and photocarrier diffusion length - on photodetection performance. Different optimization strategies for GeSn photodiodes, in particular at high Sn content, are proposed.

physics.ins-det↗

arXiv2026

CFHT MegaCam Two Deep Fields Imaging Survey (2DFIS) II: Decoding the Lensing Profile of a "Rotating" Cluster with Deep CFHT Imaging

We present a multi-wavelength analysis of the galaxy cluster RXCJ0110.0+1358 ($z=0.058$), a rotating cluster candidate, combining deep CFHT imaging, SDSS photometry, spectroscopic redshifts, and XMM-Newton X-ray observations. We find a notable discrepancy between the optical and X-ray views: while optical data reveal a pronounced bimodal galaxy distribution with significant kinematic substructure signatures, the X-ray emission exhibits a single, smoothly extended component centered on the BCG. Our weak lensing analysis resolves this discrepancy by revealing that the mass is predominantly concentrated in the southeast ($\log M_{200}/M_\odot = 14.04_{-0.40}^{+0.24}$), while the northwestern substructure has a negligible mass ($\sim 10^{13} M_\odot$). This immense mass disparity rules out the dynamical possibility of a rotating system. We demonstrate that the apparent optical bimodality arises from the projection of a filament, which led optical group-finding algorithms to misclassify these galaxies as cluster members. This contamination creates a spurious substructure that mimics a rotation signal and leads to an overestimation of the luminosity-based halo mass, resolving the observed inconsistencies.

astro-ph.GA↗

arXiv2026

Learning Generative Selection for Best-of-N

Scaling test-time compute via parallel sampling can substantially improve LLM reasoning, but is often limited by Best-of-N selection quality. Generative selection methods, such as GenSelect, address this bottleneck, yet strong selection performance remains largely limited to large models. We show that small reasoning models can acquire strong GenSelect capabilities through targeted reinforcement learning. To this end, we synthesize selection tasks from large-scale math and code instruction datasets by filtering to instances with both correct and incorrect candidate solutions, and train 1.7B-parameter models with DAPO to reward correct selections. Across math (AIME24, AIME25, HMMT25) and code (LiveCodeBench) reasoning benchmarks, our models consistently outperform prompting and majority-voting baselines, often approaching or exceeding much larger models. Moreover, these gains generalize to selecting outputs from stronger models despite training only on outputs from weaker models. Overall, our results establish reinforcement learning as a scalable way to unlock strong generative selection in small models, enabling efficient test-time scaling.

cs.LG↗

arXiv2026

Broadband tunable narrow-linewidth laser based on scattering-enhanced fiber covering E-S-C-L bands

This work demonstrates a broadband tunable narrow-linewidth laser based on scattering-enhanced fiber, covering the E-S-C-L wavelength bands from 1337.47 nm to 1631.39 nm, with a total tuning span of 293.92 nm. The laser employs two semiconductor optical amplifiers (SOAs) centered at 1420 nm and 1550 nm, which are connected into a single ring resonator via polarization multiplexing. Wavelength selection and tunability is realized using an ultra-broadband tunable filter based on a blazed grating. To suppress side longitude modes, an 18-meter-long femtosecond-laser-empowered random scattering fiber is utilized inside the cavity as a feedback medium, yielding an output linewidths between 1.54 kHz and 2.61 kHz. Benefited from the fast response of the galvanometer mirror and short relaxation time of SOAs, wavelength switching time is less than 1 ms under different tuning channels among the wavelength range of near 300 nm. The stable single-longitude-mode operation is maintained across the entire tuning range. The exceptionally broad tuning range and high spectral purity of the laser endow it with significant application potentials across a wide range of fields.

physics.optics↗

arXiv2026

Decoder Gradient Shields: A Family of Provable and High-Fidelity Methods Against Gradient-Based Box-Free Watermark Removal

Box-free model watermarking has gained significant attention in deep neural network (DNN) intellectual property protection due to its model-agnostic nature and its ability to flexibly manage high-entropy image outputs from generative models. Typically operating in a black-box manner, it employs an encoder-decoder framework for watermark embedding and extraction. While existing research has focused primarily on the encoders for the robustness to resist various attacks, the decoders have been largely overlooked, leading to attacks against the watermark. In this paper, we identify one such attack against the decoder, where query responses are utilized to obtain backpropagated gradients to train a watermark remover. To address this issue, we propose Decoder Gradient Shields (DGSs), a family of defense mechanisms, including DGS at the output (DGS-O), at the input (DGS-I), and in the layers (DGS-L) of the decoder, with a closed-form solution for DGS-O and provable performance for all DGS. Leveraging the joint design of reorienting and rescaling of the gradients from watermark channel gradient leaking queries, the proposed DGSs effectively prevent the watermark remover from achieving training convergence to the desired low-loss value, while preserving image quality of the decoder output. We demonstrate the effectiveness of our proposed DGSs in diverse application scenarios. Our experimental results on deraining and image generation tasks with the state-of-the-art box-free watermarking show that our DGSs achieve a defense success rate of 100% under all settings.

cs.CV↗

arXiv2025

Efficient Computation of Integer-constrained Cones for Conformal Parameterizations

We propose an efficient method to compute a small set of integer-constrained cone singularities, which induce a rotationally seamless conformal parameterization with low distortion. Since the problem only involves discrete variables, i.e., vertex-constrained positions, integer-constrained angles, and the number of cones, we alternately optimize these three types of variables to achieve tractable convergence. Central to high efficiency is an explicit construction algorithm that reduces the optimization problem scale to be slightly greater than the number of integer variables for determining the optimal angles with fixed positions and numbers, even for high-genus surfaces. In addition, we derive a new derivative formula that allows us to move the cones, effectively reducing distortion until convergence. Combined with other strategies, including repositioning and adding cones to decrease distortion, adaptively selecting a constrained number of integer variables for efficient optimization, and pairing cones to reduce the number, we quickly achieve a favorable tradeoff between the number of cones and the parameterization distortion. We demonstrate the effectiveness and practicability of our cones by using them to generate rotationally seamless and low-distortion parameterizations on a massive test data set. Our method demonstrates an order-of-magnitude speedup (30$\times$ faster on average) compared to state-of-the-art approaches while maintaining comparable cone numbers and parameterization distortion.

cs.GR↗

arXiv2025

Nemotron-Math: Efficient Long-Context Distillation of Mathematical Reasoning from Multi-Mode Supervision

High-quality mathematical reasoning supervision requires diverse reasoning styles, long-form traces, and effective tool integration, capabilities that existing datasets provide only in limited form. Leveraging the multi-mode generation ability of gpt-oss-120b, we introduce Nemotron-Math, a large-scale mathematical reasoning dataset containing 7.5M solution traces across high, medium, and low reasoning modes, each available both with and without Python tool-integrated reasoning (TIR). The dataset integrates 85K curated AoPS problems with 262K community-sourced StackExchange-Math problems, combining structured competition tasks with diverse real-world mathematical queries. We conduct controlled evaluations to assess the dataset quality. Nemotron-Math consistently outperforms the original OpenMathReasoning on matched AoPS problems. Incorporating StackExchange-Math substantially improves robustness and generalization, especially on HLE-Math, while preserving accuracy on math competition benchmarks. To support efficient long-context training, we develop a sequential bucketed strategy that accelerates 128K context-length fine-tuning by 2--3$\times$ without significant accuracy loss. Overall, Nemotron-Math enables state-of-the-art performance, including 100\% maj@16 accuracy on AIME 2024 and 2025 with Python TIR.

cs.AI↗

arXiv2025

HI-detected Dwarf Galaxies in the FASHI Survey: Insights from Single- and Double-Peaked Emission-Line Samples

We present a sample of low HI mass dwarf galaxies ($M_{\rm HI} < 10^8 M_\odot$) detected by The FAST All Sky HI Survey (FASHI) project. Due to the faint and irregular morphology of these galaxies, the default photometry is often inaccurate. Therefore, we utilized The Dark Energy Camera Legacy Survey (DECaLS) data to perform careful photometric measurements, and find that the low HI mass galaxies have similar stellar mass densities to dwarf elliptical (dE) galaxies. Compared to other dwarf galaxy populations, the HI-selected dwarfs exhibit higher stellar mass densities than ultradiffuse galaxies, and similar densities to HI-selected low-surface-brightness galaxies, albeit with lower stellar masses, suggesting a possible evolutionary connection among these populations. By classifying the galaxies according to their HI spectral-line profiles, we show that the double-peaked sources conform closely to the Tully-Fisher relation, whereas the single-peaked sources follow the Faber-Jackson relation but with large scatter. This indicates that the single-peaked systems are likely dispersion dominated and that the relationship between stellar mass and halo mass in such systems may remain consistent across both low- and high-mass regimes. These findings suggest that HI-selected dwarf galaxies with single-peaked HI profiles may share a similar dynamical state with massive ellipticals, offering new insights into their structural evolution and the diversity of formation pathways for low-mass galaxies.

astro-ph.GA↗

arXiv2025

Scaling Generative Verifiers For Natural Language Mathematical Proof Verification And Selection

Large language models have achieved remarkable success on final-answer mathematical problems, largely due to the ease of applying reinforcement learning with verifiable rewards. However, the reasoning underlying these solutions is often flawed. Advancing to rigorous proof-based mathematics requires reliable proof verification capabilities. We begin by analyzing multiple evaluation setups and show that focusing on a single benchmark can lead to brittle or misleading conclusions. To address this, we evaluate both proof-based and final-answer reasoning to obtain a more reliable measure of model performance. We then scale two major generative verification methods (GenSelect and LLM-as-a-Judge) to millions of tokens and identify their combination as the most effective framework for solution verification and selection. We further show that the choice of prompt for LLM-as-a-Judge significantly affects the model's performance, but reinforcement learning can reduce this sensitivity. However, despite improving proof-level metrics, reinforcement learning does not enhance final-answer precision, indicating that current models often reward stylistic or procedural correctness rather than mathematical validity. Our results establish practical guidelines for designing and evaluating scalable proof-verification and selection systems.

cs.AI↗

arXiv2025

Can Visual Input Be Compressed? A Visual Token Compression Benchmark for Large Multimodal Models

Large multimodal models (LMMs) often suffer from severe inference inefficiency due to the large number of visual tokens introduced by image encoders. While recent token compression methods, such as pruning and merging, have shown promise in reducing redundancy, their evaluation remains fragmented and inconsistent. In this work, we present UniPruneBench, a unified and extensible benchmark for visual token pruning in multimodal LLMs. UniPruneBench provides standardized protocols across six ability dimensions and ten datasets, covering ten representative compression algorithms and three families of LMMs (LLaVA-v1.5, Intern-VL3, and Qwen2.5-VL). Beyond task accuracy, it incorporates system-level metrics such as runtime and prefilling latency to provide a holistic view. Our experiments uncover several key findings: (1) random pruning is a surprisingly strong baseline, (2) no single method consistently outperforms others across scenarios, (3) pruning sensitivity varies significantly across tasks, with OCR being most vulnerable, and (4) pruning ratio is the dominant factor governing performance degradation. We believe UniPruneBench will serve as a reliable foundation for future research on efficient multimodal modeling.

cs.CV↗

arXiv2025

Quantum-elevated Chiral Discrimination for Bio-molecules

Chiral discrimination of enantiomeric biomolecules is vital in chemistry, biology, and medicine. Conventional methods, relying on circularly polarized light, face weak chiroptical signals and potential photodamage. Despite extensive efforts to improve sensitivity under low-photon exposure, classical chiral probes remain fundamentally bounded by the shot-noise limit due to quantum fluctuations. To beat these limitations, we demonstrate quantum-elevated chiral discrimination using continuous-variable polarization-entangled states as moderate-photon-flux, high-sensitivity, quantum-noise-squeezed chiral probes. We achieve a 5 dB improvement beyond the SNL in distinguishing L- and D-amino acids in liquid phase. This non-destructive, biocompatible protocol enables high-sensitivity chiral analysis, with broad implications for drug development, biochemical research, environmental monitoring, and asymmetric synthesis.

quant-ph↗

arXiv2025

DVAGen: Dynamic Vocabulary Augmented Generation

Language models trained with a fixed vocabulary struggle to generalize to novel or out-of-vocabulary words, limiting their flexibility in handling diverse token combinations. Existing dynamic vocabulary approaches attempt to address this limitation but face challenges such as fragmented codebases, lack of support for modern LLMs, and limited inference scalability. To overcome these issues, we introduce DVAGen, a fully open-source, unified framework designed for training, evaluation, and visualization of dynamic vocabulary-augmented language models. Our framework modularizes the pipeline for ease of customization, integrates seamlessly with open-source LLMs, and is the first to provide both CLI and WebUI tools for real-time result inspection. We validate the effectiveness of dynamic vocabulary methods on modern LLMs and demonstrate support for batch inference, significantly improving inference throughput.

cs.CL↗

arXiv2025

NCV: A Node-Wise Consistency Verification Approach for Low-Cost Structured Error Localization in LLM Reasoning

Verifying multi-step reasoning in large language models is difficult due to imprecise error localization and high token costs. Existing methods either assess entire reasoning chains, suffering attention dilution, or rely on expensive multi-sampling. We introduce Node-wise Consistency Verification (NCV), a training-free framework that recasts verification as lightweight binary consistency checks at the node level. By decomposing the chain of thought into interconnected verification nodes, NCV precisely localizes errors and avoids unnecessary long-form generation. Experiments demonstrate that our approach enhances interpretability and efficiency, presenting a scalable solution for reliable LLM reasoning verification. On public datasets, NCV achieves a 10\% to 25\% improvement in F1 scores over baselines while utilizing $6\times$~$58\times$ fewer tokens than traditional methods like CoT-based verifiers.

cs.AI↗

arXiv2025

Llama-Nemotron: Efficient Reasoning Models

We introduce the Llama-Nemotron series of models, an open family of heterogeneous reasoning models that deliver exceptional reasoning capabilities, inference efficiency, and an open license for enterprise use. The family comes in three sizes -- Nano (8B), Super (49B), and Ultra (253B) -- and performs competitively with state-of-the-art reasoning models such as DeepSeek-R1 while offering superior inference throughput and memory efficiency. In this report, we discuss the training procedure for these models, which entails using neural architecture search from Llama 3 models for accelerated inference, knowledge distillation, and continued pretraining, followed by a reasoning-focused post-training stage consisting of two main parts: supervised fine-tuning and large scale reinforcement learning. Llama-Nemotron models are the first open-source models to support a dynamic reasoning toggle, allowing users to switch between standard chat and reasoning modes during inference. To further support open research and facilitate model development, we provide the following resources: 1. We release the Llama-Nemotron reasoning models -- LN-Nano, LN-Super, and LN-Ultra -- under the commercially permissive NVIDIA Open Model License Agreement. 2. We release the complete post-training dataset: Llama-Nemotron-Post-Training-Dataset. 3. We also release our training codebases: NeMo, NeMo-Aligner, and Megatron-LM.

cs.CL↗