Searcharxiv⌕ Search

arXiv subjects

Zheng Ma

Publications and source records attributed to Zheng Ma.

At least 37 records · Page 2Linked to original sources
arXiv2026

Early metal-enriched baryon cycling before the midpoint of cosmic reionization

Models predict that chemical enrichment and gas redistribution should begin rapidly once star formation starts, but direct constraints at the earliest epochs have been scarce. Here we show that metal-enriched gas in multiple ionic phases was already present around galaxies before the midpoint of cosmic reionization. Using JWST/NIRSpec rest-frame ultraviolet spectroscopy from SPURS, we detect blueshifted metal absorption in three galaxies at $7.2<z<9.3$. The detected transitions span neutral, low-ionization, and high-ionization species, including O I, Si II, C II, Si IV, and C IV, with velocity offsets of $|Δv|\sim 50$--$250\,\mathrm{km\,s^{-1}}$ relative to nebular systemic redshifts. The ionic coexistence, overlapping velocity structure, and equivalent-width ratios are consistent with outflowing or otherwise kinematically disturbed galaxy-associated gas, implying rapid metal enrichment. These results show that key conditions for baryon cycling were established in at least a subset of luminous galaxies within the first several hundred million years of cosmic time, well before the completion of reionization.

astro-ph.GA↗
arXiv2026

Decoupled Latent Optimization of Diffusion Models for Full Waveform Inversion

Full waveform inversion (FWI) recovers subsurface velocity from seismic recordings by solving a severely ill-posed, nonconvex PDE-constrained optimization. Classical regularizers stabilize the inversion but fail to reproduce realistic geological structures; recent diffusion-prior methods improve realism at the cost of a fragile trade-off between data fidelity and prior consistency. We propose Decoupled Latent Optimization (DLO), which relaxes the standard latent-optimization formulation into a quadratic-penalty objective over an auxiliary physical variable and a latent variable. The data-fidelity gradient acts in physical space, the diffusion sampler contributes only through a decoded prior sample, and the standard smoothed-velocity initialization of classical FWI is preserved. On the OpenFWI benchmark, DLO outperforms classical regularizers and existing diffusion-based methods under clean, noisy, and missing-trace acquisitions. The prior, trained on 70*70 OpenFWI models, transfers directly to the Marmousi and Overthrust benchmarks, where DLO recovers intricate fault structures and remains robust to initialization smoothing and measurement noise.

cs.LG↗
arXiv2026

GeoSym127K: Scalable Symbolically-verifiable Synthesis for Multimodal Geometric Reasoning

Large Multimodal Models (LMMs) often struggle with geometric reasoning due to visual hallucinations and a lack of mathematically precise Chain-of-Thought (CoT) data. To address this, we propose the GeoSym Engine, an automated and scalable neuro-symbolic framework. By leveraging a type-conditional grammar and an analytic SymGT Solver, it derives exact symbolic ground truths and seamlessly integrates with a robust rendering pipeline to produce high-precision geometric diagrams. Using this engine, we construct GeoSym127K, a difficulty-stratified dataset featuring 51K high-resolution images, 127K questions with symbolic ground truths, and 55K answer-verified CoT QA pairs. We also introduce GeoSym-Bench, an expert-curated suite of 511 complex samples for rigorous evaluation. Through extensive supervised fine-tuning (SFT), we demonstrate that GeoSym drives concentrated improvements specifically on diagram-dependent and multi-step geometry tasks. Our Qwen3-VL-8B model gains an absolute +22.21% on the MathVerse Vision-Only subset and reaches 61.52% (+6.19% improvement) on WeMath, mitigating long-horizon logic fragmentation and outperforming advanced closed-source models like Doubao-1.8. Furthermore, applying Reinforcement Learning with Verifiable Rewards (RLVR) via GRPO reveals that initializing from structural SFT checkpoints substantially elevates the performance ceiling over zero-shot RL. Driven by deterministic exact-match signals, this showcases the robust scaling potential of our verifiable reasoning synthesis. Datasets and code are available at https://huggingface.co/datasets/Tomie0506/GeoSym127K and https://github.com/Tomie56/GeoSym127K.

cs.CV↗
arXiv2026

GraphPI: Efficient Protein Inference with Graph Neural Networks

The integration of deep learning approaches in biomedical research has been transformative, enabling breakthroughs in various applications. Despite these strides, its application in protein inference is impeded by the scarcity of extensively labeled datasets, a challenge compounded by the high costs and complexities of accurate protein annotation. In this study, we introduce GraphPI, a novel framework that treats protein inference as a node classification problem. We treat proteins as interconnected nodes within a protein-peptide-PSM graph, utilizing a Graph Neural Network-based architecture to elucidate their interrelations. To address label scarcity, we train the model on a set of unlabeled public protein datasets with pseudo-labels derived from an existing protein inference algorithm, enhanced by self-training to iteratively refine labels based on confidence scores. Contrary to prevalent methodologies necessitating dataset-specific training, our research illustrates that GraphPI, due to the well normalized nature of Percolator features, exhibits universal applicability without dataset-specific fine-tuning, a feature that not only mitigates the risk of overfitting but also enhances computational efficiency. Our empirical experiments reveal notable performance on various test datasets and deliver significantly reduced computation times compared to common protein inference algorithms.

cs.LG↗
arXiv2026

OpenMobile: Building Open Mobile Agents with Task and Trajectory Synthesis

Mobile agents powered by vision-language models have demonstrated impressive capabilities in automating mobile tasks, with recent leading models achieving a marked performance leap, e.g., nearly 70% success on AndroidWorld. However, these systems keep their training data closed and remain opaque about their task and trajectory synthesis recipes. We present OpenMobile, an open-source framework that synthesizes high-quality task instructions and agent trajectories, with two key components: (1) The first is a scalable task synthesis pipeline that constructs a global environment memory from exploration, then leverages it to generate diverse and grounded instructions. and (2) a policy-switching strategy for trajectory rollout. By alternating between learner and expert models, it captures essential error-recovery data often missing in standard imitation learning. Agents trained on our data achieve competitive results across three dynamic mobile agent benchmarks: notably, our fine-tuned Qwen2.5-VL and Qwen3-VL reach 51.7% and 64.7% on AndroidWorld, far surpassing existing open-data approaches. Furthermore, we conduct transparent analyses on the overlap between our synthetic instructions and benchmark test sets, and verify that performance gains stem from broad functionality coverage rather than benchmark overfitting. We release data and code at https://njucckevin.github.io/openmobile/ to bridge the data gap and facilitate broader mobile agent research.

cs.AI↗
arXiv2026

The Way We Tally Becomes the Tale: the Impact of Selection Strategies on the Inferred Evolution of Little Red Dots Across Cosmic Time

Little Red Dots (LRDs) have emerged as a key population linked to early black hole growth, yet photometric selections have predominantly targeted only the most extreme red systems, thereby shaping our current understanding of this new population of objects. In this work, we deliberately explore a broad range of optical redness while enforcing stringent compactness and visual inspection to ensure robustness and minimize contamination. Leveraging the depth and multiwavelength coverage of the JWST Advanced Deep Extragalactic Survey (JADES) data in the GOODS-North and GOODS-South fields, we construct the largest photometric census of LRDs to date in these fields, comprising 412 sources over $z\approx2\text{--}11$ across $\approx349.6$ arcmin$^2$. We show that classic extreme color cuts isolate only a minor fraction of this population ($\lesssim25\%$), while the majority of LRDs span a broader, largely unexplored parameter space. We quantify how selection strategies impact UV and optical luminosity functions and number density evolution, finding that current demographic trends of LRDs are strongly driven by selection biases and further limited by incomplete identification at both high and low redshift. Spectroscopically confirmed LRDs reveal a continuous range of spectral shapes, consistent with varying Active Galactic Nucleus (AGN) and host contributions in agreement with recent findings. Our results demonstrate that commonly adopted, purity-driven selections bias current demographic constraints toward the most extreme systems, potentially misrepresenting the diversity and evolution of the LRD population. Accounting for these selection effects is essential for interpreting LRDs and their role in early black hole growth.

astro-ph.GA↗
arXiv2026

Light-modulated exchange bias in multiferroic heterostructures

Magnetic straintronics, the strain-mediated control of magnetic anisotropy, has emerged as a key direction for next-generation energy-efficient technologies. In multiferroic heterostructures, magnetoelectric coupling is typically achieved by applying an electric field on a ferroelectric phase, inducing strain through the converse piezoelectric effect, which is then transferred to the adjacent ferromagnetic phase. As an alternative, strain can be remotely modulated through the photostrictive effect induced by light. While light-driven control of magnetic anisotropy has been explored, optical modulation of more complex phenomena such as exchange bias remains largely unaddressed. Here, we demonstrate significant light-induced modulation of exchange bias and magnetization switching at room temperature in a Pb(Mg1/3Nb2/3)O3-Pb(Zr,Ti)O3 (PMN-PZT)/Fe80Ga20(FeGa)/Ir20Mn80(IrMn) multiferroic heterostructure, driven by visible-light-photostriction. The magnetization state correlates with the light intensity, enabling multi-level states with light power densities as low as 0.1 W cm-2. These findings suggest a promising route toward low-power, multistate, and wireless opto-magnetic memory applications.

cond-mat.mtrl-sci↗
arXiv2026

SSI-DM: Singularity Skipping Inversion of Diffusion Models

Inverting real images into the noise space is essential for editing tasks using diffusion models, yet existing methods produce non-Gaussian noise with poor editability due to the inaccuracy in early noising steps. We identify the root cause: a mathematical singularity that renders inversion fundamentally ill-posed. We propose Singularity Skipping Inversion of Diffusion Models (SSI-DM), which bypasses this singular region by adding small noise before standard inversion. This simple approach produces inverted noise with natural Gaussian properties while maintaining reconstruction fidelity. As a plug-and-play technique compatible with general diffusion models, our method achieves superior performance on public image datasets for reconstruction and interpolation tasks, providing a principled and efficient solution to diffusion model inversion.

cs.CV↗
arXiv2026

Clumps in High-Redshift Galaxies: Mass Scaling and Radial Trends from JADES

Massive star-forming clumps are a prominent feature of high-redshift galaxies and are thought to trace gravitational fragmentation, feedback, and bulge growth in gas-rich disks. We present a statistical analysis of clumps in $\sim$3600 galaxies spanning $2 \lesssim z \lesssim 8$ from deep JWST/NIRCam imaging in the JADES GOODS--South field. Clumps are identified as residual features after subtracting smooth Sérsic profiles, enabling a uniform, rest-frame optical census of sub-galactic structure. We characterize their physical properties, size--mass relations, and spatial distributions to constrain models of sub-galactic structure formation and evolution. We find that clumps in our sample are typically low-mass ($10^{\sim7-8}M_\odot$), actively star-forming, and show diverse gas-phase metallicity, dust attenuation, and stellar population properties. Their sizes and average pairwise separations increase with cosmic time (toward lower redshift), consistent with inside-out disk growth. The clump mass function follows a power law with slope $α= -1.50_{-0.17}^{+0.19}$, consistent with fragmentation in turbulent disks. We find a deficit of relatively young clumps near galaxy centers and a radial transition in the size--mass relation: outer clumps exhibit steeper, near-virial slopes ($R_{\rm e}\propto M_*^{\sim 0.3}$), while inner clumps follow flatter trends ($R_{\rm e}\propto M_*^{\sim 0.2}$), consistent with structural evolution via migration or disruption. These results provide new constraints on the formation, survival, and dynamical evolution of clumps, highlighting their role in shaping galaxy morphology during the peak of cosmic star formation.

astro-ph.GA↗
arXiv2026

FailureMem: A Failure-Aware Multimodal Framework for Autonomous Software Repair

Multimodal Automated Program Repair (MAPR) extends traditional program repair by requiring models to jointly reason over source code, textual issue descriptions, and visual artifacts such as GUI screenshots. While recent LLM-based repair systems have shown promising results, existing approaches face several limitations: rigid workflow pipelines restrict exploration during debugging, visual reasoning is often performed over full-page screenshots without localized grounding, and failed repair attempts are rarely transformed into reusable knowledge. To address these challenges, we propose FailureMem, a multimodal repair framework that integrates three key mechanisms: a hybrid workflow-agent architecture that balances structured localization with flexible reasoning, active perception tools that enable region-level visual grounding, and a Failure Memory Bank that converts past repair attempts into reusable guidance. Experiments on SWE-bench Multimodal demonstrate FailureMem improves the resolved rate over GUIRepair by 3.7%.

cs.SE↗
arXiv2026

Robust Physics-Guided Diffusion for Full-Waveform Inversion

We develop a robust physics-guided diffusion framework for full-waveform inversion that combines a score-based generative prior with likelihood guidance computed through wave-equation simulations. We adopt a transport-based data-consistency potential (Wasserstein-2), incorporating wavefield enhancement via bounded weighting and observation-dependent normalization, thereby improving robustness to amplitude imbalance and time/phase misalignment. On the inference side, we introduce a preconditioned guided reverse-diffusion scheme that adapts the guidance strength and spatial scaling throughout the reverse-time dynamics, yielding a more stable and effective data-consistency guidance step than standard diffusion posterior sampling (DPS). Numerical experiments on OpenFWI datasets demonstrate improved reconstruction quality over deterministic optimization baselines and standard DPS under comparable computational budgets.

math.NA↗
arXiv2026

Asymptotic-Preserving Neural Networks based on Even-odd Decomposition for Multiscale Gray Radiative Transfer Equations

We present a novel Asymptotic-Preserving Neural Network (APNN) approach utilizing even-odd decomposition to tackle the nonlinear gray radiative transfer equations (GRTEs). Our AP loss demonstrates consistent stability concerning the small Knudsen number, ensuring the neural network solution uniformly converges to the diffusion limit solution. This APNN method alleviates the rigorous conservation requirements while simultaneously incorporating an auxiliary deep neural network, distinguishing it from the APNN method based on micro-macro decomposition for GRTE. Several numerical problems are examined to demonstrate the effectiveness of our proposed APNN technique.

math.NA↗
arXiv2026

Abundant Population of Broad H$α$ Emitters in the GOODS-N Field Revealed by CONGRESS, FRESCO, and JADES

We present a spectroscopic search for broad H$α$ emitters at z$\approx$3.7-6.5 in the GOODS-N field, utilizing JWST/NIRCam slitless spectroscopy from FRESCO and CONGRESS, complemented by JADES imaging. We identify 19 broad H$α$ emitters with FWHM$>$1000 km/s at z$\approx$4-5.5, including 9 new sources. The broad H$α$ luminosity function (LF) derived from our sample is consistent with those of other JWST-selected broad-line AGN reported in the literature. The black hole masses and AGN bolometric luminosities, inferred from the broad H$α$ components, indicate that most sources are accreting at ~10% of the Eddington limit. We derive their host stellar masses via SED fitting and find higher $M_{BH}/M_{*}$ ratios relative to the local $M_{BH}-M_{*}$ relations, consistent with previous studies. We find that 42% of the sample do not satisfy the widely-used color selection criteria for Little Red Dots (LRDs), with the majority of these sources lacking the characteristic steep rest-optical red slope, indicating that the LRD selection is highly incomplete when selecting AGN galaxies. A comparison of the average SEDs between our sample and LRDs selected in the same field reveals that the steep red slopes observed in some LRDs are likely due to line-boosting effects as previously suggested. Furthermore, we find that 68% of color-selected LRDs with H$α$ detections in the NIRCam/Grism spectra do not exhibit broad-line features. While the limited sensitivity of the grism spectra may hinder the detection of broad-line components in faint sources, our findings still highlight the enigmatic nature of the LRD population.

astro-ph.GA↗
arXiv2026

Empowering Small Language Models with Factual Hallucination-Aware Reasoning for Financial Classification

Small language models (SLMs) are increasingly used for financial classification due to their fast inference and local deployability. However, compared with large language models, SLMs are more prone to factual hallucinations in reasoning and exhibit weaker classification performance. This raises a natural question: Can mitigating factual hallucinations improve SLMs' financial classification? To address this, we propose a three-step pipeline named AAAI (Association Identification, Automated Detection, and Adaptive Inference). Experiments on three representative SLMs reveal that: (1) factual hallucinations are positively correlated with misclassifications; (2) encoder-based verifiers effectively detect factual hallucinations; and (3) incorporating feedback on factual errors enables SLMs' adaptive inference that enhances classification performance. We hope this pipeline contributes to trustworthy and effective applications of SLMs in finance.

cs.AI↗
arXiv2025

Quantifying the Impact of Structured Output Format on Large Language Models through Causal Inference

Structured output from large language models (LLMs) has enhanced efficiency in processing generated information and is increasingly adopted in industrial applications. Prior studies have investigated the impact of structured output on LLMs' generation quality, often presenting one-way findings. Some suggest that structured format enhances completeness and factual accuracy, while others argue that it restricts the reasoning capacity of LLMs and leads to reductions in standard evaluation metrics. Potential limitations of these assessments include restricted testing scenarios, weakly controlled comparative settings, and reliance on coarse metrics. In this work, we present a refined analysis using causal inference. Based on one assumed and two guaranteed constraints, we derive five potential causal structures characterizing the influence of structured output on LLMs' generation: (1) collider without m-bias, (2) collider with m-bias, (3) single cause from instruction, (4) single cause from output format, and (5) independence. Across seven public and one developed reasoning tasks, we find that coarse metrics report positive, negative, or neutral effects of structured output on GPT-4o's generation. However, causal inference reveals no causal impact in 43 out of 48 scenarios. In the remaining 5, 3 involve multifaceted causal structures influenced by concrete instructions. Further experiments show that OpenAI-o3 are more resilient to output formats than general-purpose GPT-4o and GPT-4.1, highlighting an unaware advantage of reasoning models.

cs.CL↗
arXiv2025

Quasar Radiative Feedback May Suppress Galaxy Growth on Intergalactic Scales at $z = 6.3$

We present observational evidence that intense ionizing radiation from a luminous quasar suppresses nebular emission in nearby galaxies on intergalactic scales at $z=6.3$. Using JWST/NIRCam grism spectroscopy from the SAPPHIRES and EIGER programs, we identify a moderate but statistically significant decline in [O\,\textsc{iii}]\,$\lambda5008$ luminosity relative to the UV continuum ($L_{5008}/L_{1500}$) among galaxies within $\sim$ 7 comoving Mpc (cMpc) of the quasar J0100$+$2802, the most UV-luminous quasar known at this epoch ($M_{1450}=-29.26$). While $L_{1500}$ remains roughly constant with transverse distance, $L_{5008}$ increases significantly, suggesting suppression of very recent star formation toward the quasar. The effect persists after controlling for completeness, local density, and UV luminosity, and correlates with the projected photoionization-rate profile $Γ_{\mathrm{qso}}$. A weaker but directionally consistent suppression in $L_{5008}/L_{1500}$ is also observed along the line of sight. The transverse suppression radius ($\sim$ 7 cMpc) implies a recent radiative episode with a cumulative duration $\sim$ 3.1 Myr, shorter than required for thermal photoheating to dominate and thus more naturally explained by rapid H$_2$ photodissociation and related radiative processes. Environmental effects alone appear insufficient to explain the signal. Our results provide direct, geometry-based constraints on large-scale quasar radiative feedback and recent quasar lifetimes.

astro-ph.GA↗
arXiv2025

Towards Multiple Missing Values-resistant Unsupervised Graph Anomaly Detection

Unsupervised graph anomaly detection (GAD) has received increasing attention in recent years, which aims to identify data anomalous patterns utilizing only unlabeled node information from graph-structured data. However, prevailing unsupervised GAD methods typically presuppose complete node attributes and structure information, a condition hardly satisfied in real-world scenarios owing to privacy, collection errors or dynamic node arrivals. Existing standard imputation schemes risk "repairing" rare anomalous nodes so that they appear normal, thereby introducing imputation bias into the detection process. In addition, when both node attributes and edges are missing simultaneously, estimation errors in one view can contaminate the other, causing cross-view interference that further undermines the detection performance. To overcome these challenges, we propose M$^2$V-UGAD, a multiple missing values-resistant unsupervised GAD framework on incomplete graphs. Specifically, a dual-pathway encoder is first proposed to independently reconstruct missing node attributes and graph structure, thereby preventing errors in one view from propagating to the other. The two pathways are then fused and regularized in a joint latent space so that normals occupy a compact inner manifold while anomalies reside on an outer shell. Lastly, to mitigate imputation bias, we sample latent codes just outside the normal region and decode them into realistic node features and subgraphs, providing hard negative examples that sharpen the decision boundary. Experiments on seven public benchmarks demonstrate that M$^2$V-UGAD consistently outperforms existing unsupervised GAD methods across varying missing rates.

cs.LG↗
arXiv2025

DeepRTE: Pre-trained Attention-based Neural Network for Radiative Transfer

In this paper, we propose a novel neural network approach, termed DeepRTE, to address the steady-state Radiative Transfer Equation (RTE). The RTE is a differential-integral equation that governs the propagation of radiation through a participating medium, with applications spanning diverse domains such as neutron transport, atmospheric radiative transfer, heat transfer, and optical imaging. Our DeepRTE framework demonstrates superior computational efficiency for solving the steady-state RTE, surpassing traditional methods and existing neural network approaches. This efficiency is achieved by embedding physical information through derivation of the RTE and mathematically-informed network architecture. Concurrently, DeepRTE achieves high accuracy with significantly fewer parameters, largely due to its incorporation of mechanisms such as multi-head attention. Furthermore, DeepRTE is a mesh-free neural operator framework with inherent zero-shot capability. This is achieved by incorporating Green's function theory and pre-training with delta-function inflow boundary conditions into both its architecture design and training data construction. The efficacy of the proposed approach is substantiated through comprehensive numerical experiments.

cs.LG↗