arXiv subjects

Xue Zhang

Publications and source records attributed to Xue Zhang.

At least 37 records · Page 2Linked to original sources

arXiv2026

Exploring Memory Effects: Sparse Identification in Vector-Borne Diseases

Predicting the human burden of vector-borne diseases from limited surveillance data remains a major challenge, particularly in the presence of nonlinear transmission dynamics and delayed effects arising from vector ecology and human behavior. We develop a data-driven framework based on an extension of Sparse Identification of Nonlinear Dynamics (SINDy) to systems with distributed memory, enabling discovery of transmission mechanisms directly from time series data. Using severe fever with thrombocytopenia syndrome (SFTS) as a case study, we show that this approach can uncover key features of tick-borne disease dynamics using only human incidence and local temperature data, without imposing predefined assumptions on human case reporting. We further demonstrate that predictive performance is substantially enhanced when the data-driven model is coupled with mechanistic representations of tick-host transmission pathways informed by empirical studies. The framework supports systematic sensitivity analysis of memory kernels and behavioral parameters, identifying those most influential for prediction accuracy. Although the approach prioritizes predictive accuracy over mechanistic transparency, it yields sparse, interpretable integral representations suitable for epidemiological forecasting. This hybrid methodology provides a scalable strategy for forecasting vector-borne disease risk and informing public health decision-making under data limitations.

math.DS↗

arXiv2026

Exploring the central engines of gamma-ray bursts from prompt light curves

Hyperaccreting stellar-mass black hole systems are leading candidates for the central engines of gamma-ray bursts (GRBs). Their jets are thought to be powered by either the Blandford-Znajek (BZ) process or neutrino-dominated accretion flows (NDAFs), but discriminating between these mechanisms remains challenging. To address this, we propose using the luminosity decay slope (parameter d) of GRB light curves to distinguish between the BZ and NDAF mechanisms, thereby linking the light-curve morphology to the central engine physics. By analysing 85 single-peaked GRBs with fast-rise, exponential-decay (FRED) profiles observed by Swift/BAT using 64 ms background-subtracted light curves, we fit the decay slope (parameter d) with the empirical Kocevski-Ryde-Liang (KRL) function and compare the results with theoretical predictions for the BZ (d approximately 1.67) and the NDAF (d approximately 3.7 to 7.8) mechanisms. We find that the decay slope (parameter d) can differentiate central engine mechanisms, with 15 GRBs consistent with the BZ mechanism and 22 supporting the NDAF mechanism. However, most events exhibit slopes within the range between 2 and 4, suggesting a hybrid of mechanisms, with NDAF being dominant.

astro-ph.HE↗

arXiv2026

Squared Bessel processes under nonlinear expectation

In this paper, we define the squared G-Bessel process as the square of the modulus of a class of G-Brownian motions and establish that it is the unique solution to a stochastic differential equation. We then derive several path properties of the squared G-Bessel process, which are more profound in the capacity sense. Furthermore, we provide upper and lower bounds for the Laplace transform of the squared G-Bessel process. Finally, we prove that the time-space transformed squared G-Bessel process is a G'-CIR process.

math.PR↗

arXiv2026

Think Natively: Unlocking Multilingual Reasoning with Consistency-Enhanced Reinforcement Learning

Large Reasoning Models (LRMs) have achieved remarkable performance on complex reasoning tasks by adopting the ``think-then-answer'' paradigm, which enhances both accuracy and interpretability. However, current LRMs exhibit two critical limitations when processing non-English languages: (1) They often struggle to maintain input-output language consistency; (2) They generally perform poorly with wrong reasoning paths and lower answer accuracy compared to English. These limitations significantly compromise the interpretability of reasoning processes and degrade the user experience for non-English speakers, hindering the global deployment of LRMs. To address these limitations, we propose M-Thinker, which is trained by the GRPO algorithm that involves a Language Consistency (LC) reward and a novel Cross-lingual Thinking Alignment (CTA) reward. Specifically, the LC reward defines a strict constraint on the language consistency between the input, thought, and answer. Besides, the CTA reward compares the model's non-English reasoning paths with its English reasoning path to transfer its own reasoning capability from English to non-English languages. Through an iterative RL procedure, our M-Thinker-1.5B/4B/7B models not only achieve nearly 100% language consistency and superior performance on two multilingual benchmarks (MMATH and PolyMath), but also exhibit excellent generalization on out-of-domain languages.

cs.CL↗

arXiv2025

Design of 3D Beamforming and Deployment Strategies for ISAC-based HAPS Systems

This paper explores high-altitude platform station (HAPS) systems enabled by integrated sensing and communication (ISAC), in which a HAPS simultaneously transmits communication signals and synthetic aperture radar (SAR) imaging signals to support multi-user communication while performing ground target sensing. Taking into account the operational characteristics of SAR imaging, we consider two HAPS deployment strategies: (i) a quasi-stationary HAPS that remains fixed at an optimized location during SAR operation, following the stop-and-go scanning model; and (ii) a dynamic HAPS that continuously adjusts its flight trajectory along a circular path. For each strategy, we aim at maximizing the weighted sum-rate throughput for communication users while ensuring that SAR imaging requirements, such as beampattern gain and signal-to-noise ratio (SNR), are satisfied. This is achieved by jointly optimizing the HAPS deployment strategy, i.e., its placement or trajectory, along with three-dimensional (3D) transmit beamforming, under practical constraints including transmit power limits, energy consumption, and flight dynamics. Nevertheless, the formulated optimization problems corresponding to the two deployment strategies are inherently non-convex. To address the issue, we propose efficient algorithms that leverage both convex and non-convex optimization techniques to obtain high-quality suboptimal solutions. Numerical results demonstrate the effectiveness and advantages of the proposed approaches over benchmark schemes.

eess.SP↗

arXiv2025

Reconstruction of dark energy using DESI DR2

Using a model-independent Gaussian process (GP) method to reconstruct the dimensionless luminosity distance $D$ and its derivatives, we derive the evolution of the dimensionless Hubble parameter $E$, the deceleration parameter $q$, and the state parameter $w$ of dark energy. We utilize the PantheonPlus, SH0ES, and Gamma Ray Burst (GRB) data to derive the dimensionless luminosity distance $D$. Additionally, we employ observational $H(z)$ data (OHD) and baryon acoustic oscillations (BAO) from Dark Energy Spectroscopic Instrument (DESI) Data Release 2 (DR2) to obtain the first derivative of the dimensionless luminosity distance $D^{'}$. To obtain the reconstructed $D$ and $D^{'}$, we utilize the fiducial value from each dataset, with particular emphasis on the varying $H_0$. According to the reconstruction results obtained from PantheonPlus+SH0ES+GRB+OHD and PantheonPlus+SH0ES+GRB+OHD+DESI data, we find that $E$ are consistent with the predictions of the $Λ$CDM model at a $2σ$ confidence level within the redshift range of $z<2$. However, the reconstruction results for $q$ exhibit deviations from the $Λ$CDM model in the range of $z<0.3$. Furthermore, we observe that the mean value of $w$ exhibits evolving behavior, transiting from $w < -1$ to $w > -1$ around $z_{\rm wt}=0.464^{+0.235}_{-0.120}$. Combining data from DESI DR2 can slightly enhance the accuracy of our constraints.

astro-ph.CO↗

arXiv2025

Controlling the Exchange Field of Surface Spin Impurities via DC Voltages

Recent advances in scanning tunneling microscopy have enabled quantum-coherent control of single surface spins via all-electric electron spin resonance (ESR). Such control requires magnetoelectric coupling, since spin resonance is a magnetic effect. We show that a magnetic tip induces a bias-dependent exchange field on a localized Anderson impurity via virtual particle exchange with the magnetic lead. This field differs from Heisenberg exchange and can be tuned, reversed, or suppressed by the bias voltage. Our model reproduces bias-controlled resonance shifts for S = 1/2 titanium atoms and Fe(II) phthalocyanine, enabling spin control via the exchange field and revealing the magnetoelectric mechanism behind all-electric ESR for spin-based quantum technologies.

cond-mat.mes-hall↗

arXiv2025

The BINGO/ABDUS Project: Forecast for cosmological parameters from a mock Fast Radio Bursts survey

There are various surveys that will provide excellent data to search for and localize Fast Radio Bursts (FRBs). The BINGO project will be one such survey, and this collaboration has already estimated a FRB detection rate that the project will yield. We present a forecast of the future constraints on our current cosmological model that the BINGO FRB detections and localizations will have when added to other current cosmological datasets. We quantify the dispersion measure (DM) as a function of redshift ($z$) for the BINGO FRB mock sample. Furthermore, we use current datasets (Supernovae, Baryonic Acoustic Oscillations, and Cosmic Microwave Background data) prior to assessing the efficacy of constraining dark energy models using Monte Carlo methods. Our results show that spatially localized BINGO FRB dataset will provide promising constraints on the population of host galaxies intrinsic DM and be able to measure the nuisance parameters present within a FRB cosmological analysis. They will also provide alternative estimates on other parameters such as the Hubble constant and the dark energy equation of state. In particular, we should see that BINGO FRB data can put constraints on the degenerate $w-H_0$ plane, which the CMB is incapable of measuring, allowing FRBs to be a viable alternative to BAO to constrain the dark energy equation of state. We conclude that FRBs remain a promising future probe for cosmology and that the FRBs localized by the BINGO project will contribute significantly to our knowledge of the current cosmological model.

astro-ph.CO↗

arXiv2025

CM-Align: Consistency-based Multilingual Alignment for Large Language Models

Current large language models (LLMs) generally show a significant performance gap in alignment between English and other languages. To bridge this gap, existing research typically leverages the model's responses in English as a reference to select the best/worst responses in other languages, which are then used for Direct Preference Optimization (DPO) training. However, we argue that there are two limitations in the current methods that result in noisy multilingual preference data and further limited alignment performance: 1) Not all English responses are of high quality, and using a response with low quality may mislead the alignment for other languages. 2) Current methods usually use biased or heuristic approaches to construct multilingual preference pairs. To address these limitations, we design a consistency-based data selection method to construct high-quality multilingual preference data for improving multilingual alignment (CM-Align). Specifically, our method includes two parts: consistency-guided English reference selection and cross-lingual consistency-based multilingual preference data construction. Experimental results on three LLMs and three common tasks demonstrate the effectiveness and superiority of our method, which further indicates the necessity of constructing high-quality preference data.

cs.CL↗

arXiv2025

A Disease-Centric Vision-Language Foundation Model for Precision Oncology in Kidney Cancer

The non-invasive assessment of increasingly incidentally discovered renal masses is a critical challenge in urologic oncology, where diagnostic uncertainty frequently leads to the overtreatment of benign or indolent tumors. In this study, we developed and validated RenalCLIP using a dataset of 27,866 CT scans from 8,809 patients across nine Chinese medical centers and the public TCIA cohort, a visual-language foundation model for characterization, diagnosis and prognosis of renal mass. The model was developed via a two-stage pre-training strategy that first enhances the image and text encoders with domain-specific knowledge before aligning them through a contrastive learning objective, to create robust representations for superior generalization and diagnostic precision. RenalCLIP achieved better performance and superior generalizability across 10 core tasks spanning the full clinical workflow of kidney cancer, including anatomical assessment, diagnostic classification, and survival prediction, compared with other state-of-the-art general-purpose CT foundation models. Especially, for complicated task like recurrence-free survival prediction in the TCIA cohort, RenalCLIP achieved a C-index of 0.726, representing a substantial improvement of approximately 20% over the leading baselines. Furthermore, RenalCLIP's pre-training imparted remarkable data efficiency; in the diagnostic classification task, it only needs 20% training data to achieve the peak performance of all baseline models even after they were fully fine-tuned on 100% of the data. Additionally, it achieved superior performance in report generation, image-text retrieval and zero-shot diagnosis tasks. Our findings establish that RenalCLIP provides a robust tool with the potential to enhance diagnostic accuracy, refine prognostic stratification, and personalize the management of patients with kidney cancer.

eess.IV↗

arXiv2025

AlignDistil: Token-Level Language Model Alignment as Adaptive Policy Distillation

In modern large language models (LLMs), LLM alignment is of crucial importance and is typically achieved through methods such as reinforcement learning from human feedback (RLHF) and direct preference optimization (DPO). However, in most existing methods for LLM alignment, all tokens in the response are optimized using a sparse, response-level reward or preference annotation. The ignorance of token-level rewards may erroneously punish high-quality tokens or encourage low-quality tokens, resulting in suboptimal performance and slow convergence speed. To address this issue, we propose AlignDistil, an RLHF-equivalent distillation method for token-level reward optimization. Specifically, we introduce the reward learned by DPO into the RLHF objective and theoretically prove the equivalence between this objective and a token-level distillation process, where the teacher distribution linearly combines the logits from the DPO model and a reference model. On this basis, we further bridge the accuracy gap between the reward from the DPO model and the pure reward model, by building a contrastive DPO reward with a normal and a reverse DPO model. Moreover, to avoid under- and over-optimization on different tokens, we design a token adaptive logit extrapolation mechanism to construct an appropriate teacher distribution for each token. Experimental results demonstrate the superiority of our AlignDistil over existing methods and showcase fast convergence due to its token-level distributional reward optimization.

cs.CL↗

arXiv2025

Fundamental limits via CRB of semi-blind channel estimation in Massive MIMO systems

This paper investigates the asymptotic behavior of the deterministic and stochastic Cramér-Rao Bounds (CRB) for semi-blind channel estimation in massive multiple-input multiple-output (MIMO) systems. We derive and analyze mathematically tractable expressions for both metrics under various asymptotic regimes, which govern the growth rates of the number of antennas, the number of users, the training sequence length, and the transmission block length. Unlike the existing work, our results show that the CRB can be made arbitrarily small as the transmission block length increases, but only when the training sequence length grows at the same rate and the number of users remains fixed. However, if the number of training sequences remains proportional to the number of users, the channel estimation error is always lower-bounded by a non-vanishing constant. Numerical results are presented to support our findings and demonstrate the advantages of semi-blind channel estimation in reducing the required number of training sequences.

eess.SP↗

arXiv2025

Less, but Better: Efficient Multilingual Expansion for LLMs via Layer-wise Mixture-of-Experts

Continually expanding new languages for existing large language models (LLMs) is a promising yet challenging approach to building powerful multilingual LLMs. The biggest challenge is to make the model continuously learn new languages while preserving the proficient ability of old languages. To achieve this, recent work utilizes the Mixture-of-Experts (MoE) architecture to expand new languages by adding new experts and avoid catastrophic forgetting of old languages by routing corresponding tokens to the original model backbone (old experts). Although intuitive, this kind of method is parameter-costly when expanding new languages and still inevitably impacts the performance of old languages. To address these limitations, we analyze the language characteristics of different layers in LLMs and propose a layer-wise expert allocation algorithm (LayerMoE) to determine the appropriate number of new experts for each layer. Specifically, we find different layers in LLMs exhibit different representation similarities between languages and then utilize the similarity as the indicator to allocate experts for each layer, i.e., the higher similarity, the fewer experts. Additionally, to further mitigate the forgetting of old languages, we add a classifier in front of the router network on the layers with higher similarity to guide the routing of old language tokens. Experimental results show that our method outperforms the previous state-of-the-art baseline with 60% fewer experts in the single-expansion setting and with 33.3% fewer experts in the lifelong-expansion setting, demonstrating the effectiveness of our method.

cs.CL↗

arXiv2025

A universal scaling law for active diffusion in complex media

Using granular experiments and computer simulations, we investigate the long-time diffusion of active tracers in a broad class of complex media composed of frozen obstacles of diverse structures. By introducing a dimensionless persistence length $Q = v_d τ_r / d_t$, we propose a modified scaling relation that independently collapses experimental and simulation results across active and passive particles, diverse media, and distinct propulsion mechanisms. Our results reveal a universal active diffusion-structure relation that holds across both equilibrium and nonequilibrium regimes, providing a simple predictive framework for active diffusion in complex environments.

cond-mat.soft↗

arXiv2025

Quantum-enhanced weak absorption estimation with correlated photons

Conventional absorption spectroscopy relies on coherent laser sources, and in turn suffers from the inherent limitation of shot noise, especially in estimating weak absorption. Here we propose a measurement strategy with correlated photons to determine the weak absorption by distinguishing the output with and without photons, dubbed as the on-off measurement. We demonstrate that absorption spectroscopy that incorporates quantum correlations is capable of estimating weak absorption down to a single-photon level, even in noisy environments, achieving a precision comparable to that obtained through 1000 photons in conventional absorption spectroscopy. Our strategy provides a new method to probe fragile systems with weak absorption, avoiding the occurrence of light-induced damage.

quant-ph↗

arXiv2025

A Dual-Space Framework for General Knowledge Distillation of Large Language Models

Knowledge distillation (KD) is a promising solution to compress large language models (LLMs) by transferring their knowledge to smaller models. During this process, white-box KD methods usually minimize the distance between the output distributions of the teacher model and the student model to transfer more information. However, we reveal that the current white-box KD framework exhibits two limitations: a) bridging probability distributions from different output spaces will limit the similarity between the teacher model and the student model; b) this framework cannot be applied to LLMs with different vocabularies. One of the root causes for these limitations is that the distributions from the teacher and the student for KD are output by different prediction heads, which yield distributions in different output spaces and dimensions. Therefore, in this paper, we propose a dual-space knowledge distillation (DSKD) framework that unifies the prediction heads of the teacher and the student models for KD. Specifically, we first introduce two projectors with ideal initialization to project the teacher/student hidden states into the student/teacher representation spaces. After this, the hidden states from different models can share the same head and unify the output spaces of the distributions. Furthermore, we develop an exact token alignment (ETA) algorithm to align the same tokens in two differently-tokenized sequences. Based on the above, our DSKD framework is a general KD framework that supports both off-policy and on-policy KD, and KD between any two LLMs regardless of their vocabularies. Extensive experiments on instruction-following, mathematical reasoning, and code generation benchmarks show that DSKD significantly outperforms existing methods based on the current white-box KD framework and surpasses other cross-tokenizer KD methods for LLMs with different vocabularies.

cs.CL↗

arXiv2025

The Impact of the Hubble Tension on the Evidence for Dynamical Dark Energy

Recent findings from the Dark Energy Spectroscopic Instrument (DESI) Data Release 2 (DR2) favor a dynamical dark energy characterized by a phantom crossing feature. This result also implies a lower value of the Hubble constant, thereby intensifying the so-called Hubble tension. To alleviate the Hubble tension, we consider the early dark energy and explore its impact on the evidence for dynamical dark energy including the Hubble constant calibrated by the SH0ES collaboration. We find that incorporating SH0ES prior with CMB, DESI DR2 BAO and Pantheon Plus/Union3/DESY5 data reduces the preference to dynamical dark energy to $1.5σ/1.4σ/2.4σ$ level, respectively. Our results suggest a potential tension between the Hubble constant $H_0$ of the SH0ES measurement and the phantom-to-quintessence transition in dark energy favored by DESI DR2 BAO data.

astro-ph.CO↗

arXiv2025

Constitution or Collapse? Exploring Constitutional AI with Llama 3-8B

As language models continue to grow larger, the cost of acquiring high-quality training data has increased significantly. Collecting human feedback is both expensive and time-consuming, and manual labels can be noisy, leading to an imbalance between helpfulness and harmfulness. Constitutional AI, introduced by Anthropic in December 2022, uses AI to provide feedback to another AI, greatly reducing the need for human labeling. However, the original implementation was designed for a model with around 52 billion parameters, and there is limited information on how well Constitutional AI performs with smaller models, such as LLaMA 3-8B. In this paper, we replicated the Constitutional AI workflow using the smaller LLaMA 3-8B model. Our results show that Constitutional AI can effectively increase the harmlessness of the model, reducing the Attack Success Rate in MT-Bench by 40.8%. However, similar to the original study, increasing harmlessness comes at the cost of helpfulness. The helpfulness metrics, which are an average of the Turn 1 and Turn 2 scores, dropped by 9.8% compared to the baseline. Additionally, we observed clear signs of model collapse in the final DPO-CAI model, indicating that smaller models may struggle with self-improvement due to insufficient output quality, making effective fine-tuning more challenging. Our study suggests that, like reasoning and math ability, self-improvement is an emergent property.

cs.AI↗