arXiv subjects

Cheng Li

Publications and source records attributed to Cheng Li.

At least 73 records · Page 4Linked to original sources

arXiv2026

SE3D: Testing the recovery of stellar population, dust and structural properties on mock-observed toy model and simulated galaxies

The translation from direct observables to physical properties of galaxies is a key step in reconstructing their evolutionary histories. Star-dust geometry and inhomogeneous stellar populations can induce spatial variations in the mass-to-light ratio, complicating this process. In this paper, we present tests of SE3D, a novel modelling framework built around a radiative transfer emulator, aimed at tackling this problem. We test the ability to recover known intrinsic properties of toy model and TNG50 simulated galaxies by jointly fitting mock observations of their multi-wavelength photometric and structural properties. We find an encouraging performance ($\lesssim$ 0.1 dex) for several key characteristics, including the bulk stellar mass, dust mass and SFR, as well as their respective radial extents. We point out limitations, and investigate the impact of various sources of model mismatch. Among them, mismatch in the shapes of star formation histories contributes most, with radial and azimuthal structure and stellar metallicity distributions playing a progressively more minor role. We also analyse the evolution from z=2 to z=0 of resolved stellar and dust properties of TNG galaxies, as measured intrinsically and expressed in their distribution across UVJ and IRX-$β$ diagnostic diagrams. We test different methods to assign dust to the simulation, and find a persistent lack of Mdust/Mstar evolution and a more limited dynamic range across the diagnostic diagrams compared to observations.

astro-ph.GA↗

arXiv2026

Generating quantum entanglement from sunlight

Energy consumption is becoming a serious bottleneck for integrating quantum technologies within the existing global information infrastructure. In photonic architectures, considerable energy overheads stem from using lasers, whose high coherence was long considered indispensable for quantum state preparation. Here, we demonstrate that natural, incoherent sunlight can successfully produce quantum-entangled states via spontaneous parametric down-conversion. We detect polarization-entangled photon pairs with a concurrence of $0.905\pm0.053$ and a Bell state fidelity of $0.939\pm0.027$. Importantly, the system violates Bell's inequality with $S=2.5408\pm0.2171$, exceeding the classical threshold of 2, while maintaining generation rates comparable to laser-based setups. These findings pave the way for sustainable quantum applications in resource-limited environments like interplanetary missions.

quant-ph↗

arXiv2026

Filtering Interlopers with Photometry and Diagnostic Features: A Machine Learning Framework Validated with CSST Slitless Spectroscopy

The slitless spectroscopic method employed by missions such as Euclid and the Chinese Space-station Survey Telescope (CSST) faces a fundamental challenge: spectroscopic redshifts derived from their data are susceptible to emission-line misidentification due to the limited spectral resolution and signal-to-noise ratio. This effect systematically introduces interloper galaxies into the sample. Conventional strict selection not only struggles to secure high redshift purity but also drastically reduces completeness by discarding valuable data. To overcome this limitation, we develop an XGBoost classifier that leverages photometric properties and spectroscopic diagnostics to construct a high-purity redshift catalog while maximizing completeness. We validate this method on a simulated sample with spectra generated by the CSST emulator for slitless spectroscopy. Of the $\sim$62 million galaxies that obtain valid redshifts (parent sample), approximately 43% achieve accurate measurements, defined as $|Δz| \leqslant 0.002(1+z)$. From this parent sample, the XGBoost classifier selects galaxies with a selection efficiency of 42.3% on the test set and 42.2% when deployed on the entire parent sample. Crucially, among the retained galaxies, 96.6% (parent sample: 96.5%) achieve accurate measurements, while the outlier fraction ($|Δz|>0.01(1+z)$) is constrained to 0.13% (0.11%). We verified that simplified configurations that exclude either spectroscopic diagnostics (except the measured redshift) or photometric data yield significantly higher outlier fractions, increasing by factors of approximately 3.5 and 6.3, respectively, with the latter case also introducing notable catastrophic interloper contamination. This framework effectively resolves the purity-completeness trade-off, enabling robust large-scale cosmological studies with CSST and similar surveys.

astro-ph.CO↗

arXiv2026

Low Light Image Enhancement Challenge at NTIRE 2026

This paper presents a comprehensive review of the NTIRE 2026 Low Light Image Enhancement Challenge, highlighting the proposed solutions and final results. The objective of this challenge is to identify effective networks capable of producing clearer and visually compelling images in diverse and challenging conditions by learning representative visual cues with the purpose of restoring information loss due to low-contrast and noisy images. A total of 195 participants registered for the first track and 153 for the second track of the competition, and 22 teams ultimately submitted valid entries. This paper thoroughly evaluates the state-of-the-art advances in (joint denoising and) low-light image enhancement, showcasing the significant progress in the field, while leveraging samples of our novel dataset.

cs.CV↗

arXiv2026

Juno Microwave Radiometer Observations Reveal A Warmer Polar Atmosphere on Jupiter

The intriguing circumpolar cyclone pattern at Jupiter's poles raises fundamental questions about how these systems are organized vertically and, further, how the planet's internal heat shapes and sustains them in the absence of solar insolation. We report recent close-in observations of Jupiter's north pole acquired by NASA's Juno Microwave Radiometer (MWR), which achieved comprehensive microwave mapping of the region at an unprecedentedly high resolution. Using six-channel measurements from eleven perijove passes (PJ51-PJ61) poleward of 75N, we derive polar-mean nadir brightness temperatures and limb-darkening spectra that together point to two equally plausible atmospheric scenarios: (1) a dry-adiabatic profile with slightly depleted ammonia gas at a few bars, or (2) a moist-adiabatic profile with uniform ammonia. Markov chain Monte Carlo retrievals yield a deep ammonia abundance of 354.8+12.0/-11.0 ppmv (3+/-0.1 x solar) and a water abundance of 1.8+1.5/-1.1 x 1000 ppmv (2.1+1.8/-1.3 x solar), resembling previous estimates at lower latitudes. Remarkably, the north pole is found to be 6-7 K warmer than the equator at the 1-bar level, although the inferred difference is close to the 1-sigma uncertainty level. If confirmed, this result would suggest an enhanced internal heat flux toward the poles, which is consistent with the more intense lightning activity observed at high latitudes.

astro-ph.EP↗

arXiv2026

Interpreting the strong clustering of ultra-diffuse galaxies by halo spin bias

We use the IllustrisTNG300-ODM simulation to investigate the spin bias of low-mass halos and its connection to the strong clustering of ultra-diffuse galaxies (UDGs) reported by Zhang et al. (2025). By comparing two halo spin definitions-one using only bound particles ($λ_{\rm b}$) and another including unbound particles ($λ_{\rm a}$)-we demonstrate that the spin bias of low-mass halos critically depends on the definition. While $λ_{\rm a}$ yields stronger clustering for higher-spin halos at all masses, $λ_{\rm b}$ produces an inverted trend below $M_{\rm h}\sim 10^{11} \rm M_{\odot}/h$. This discrepancy is driven by a subset of halos in high-density environments that have large $λ_{\rm a}$ but small $λ_{\rm b}$. Using an empirical model implemented in SDSS-like mocks, we link the stellar surface-mass-density ($Σ_\ast$) of a galaxy to $λ_{\rm a}$ of its host halo and find an anti-correlation that more diffuse dwarfs tend to reside in higher-spin halos. The model naturally reproduces the observed strong clustering of UDGs within the standard $Λ$CDM framework without invoking exotic assumptions such as self-interacting dark matter. The high fraction of unbound particles in UDG hosts likely originates from tidal fields in dense regions, an effect particularly significant for low-mass halos. We discuss how the angular momentum of a halo represented by $λ_{\rm a}$ may be transferred to the gas to affect size and surface density of the galaxy that forms in the halo.

astro-ph.GA↗

arXiv2026

LC4-DViT: Land-cover Creation for Land-cover Classification with Deformable Vision Transformer

Land-cover underpins ecosystem services, hydrologic regulation, disaster-risk reduction, and evidence-based land planning; timely, accurate land-cover maps are therefore critical for environmental stewardship. Remote sensing-based land-cover classification offers a scalable route to such maps but is hindered by scarce and imbalanced annotations and by geometric distortions in high-resolution scenes. We propose LC4-DViT (Land-cover Creation for Land-cover Classification with Deformable Vision Transformer), a framework that combines generative data creation with a deformation-aware Vision Transformer. A text-guided diffusion pipeline uses GPT-4o-generated scene descriptions and super-resolved exemplars to synthesize class-balanced, high-fidelity training images, while DViT couples a DCNv4 deformable convolutional backbone with a Vision Transformer encoder to jointly capture fine-scale geometry and global context. On eight classes from the Aerial Image Dataset (AID)-Beach, Bridge, Desert, Forest, Mountain, Pond, Port, and River-DViT achieves 0.9572 overall accuracy, 0.9576 macro F1-score, and 0.9510 Cohen' s Kappa, improving over a vanilla ViT baseline (0.9274 OA, 0.9300 macro F1, 0.9169 Kappa) and outperforming ResNet50, MobileNetV2, and FlashInternImage. Cross-dataset experiments on a three-class SIRI-WHU subset (Harbor, Pond, River) yield 0.9333 overall accuracy, 0.9316 macro F1, and 0.8989 Kappa, indicating good transferability. An LLM-based judge using GPT-4o to score Grad-CAM heatmaps further shows that DViT' s attention aligns best with hydrologically meaningful structures. These results suggest that description-driven generative augmentation combined with deformation-aware transformers is a promising approach for high-resolution land-cover mapping.

cs.CV↗

arXiv2026

Accelerating Long-Tail Generation in Synchronous RLHF Training via Adaptive Tensor Parallelism

Reinforcement Learning from Human Feedback (RLHF) has become a key post-training paradigm for improving model quality. However, the synchronous three-stage RLHF pipeline is often bottlenecked by the generation stage, where response-length skew causes the effective batch size to shrink rapidly during decoding, leaving GPUs underutilized while a few long responses remain unfinished. Mainstream frameworks employ a static tensor parallelism (TP) configuration that cannot adapt to changing batch characteristics, leaving substantial performance headroom unexplored. We propose PAT, an adaptive TP method that dynamically reconfigures TP during the generation stage of each RLHF iteration. PAT introduces two key techniques. First, a predictor-guided online reconfiguration method decides both the reconfiguration point and the target TP configuration based on offline profiling, triggering reconfiguration only when the predicted latency benefit outweighs the reconfiguration overhead. Second, a lightweight online reconfiguration mechanism updates only the states and layouts affected by TP changes: it adapts unfinished decoding states through a cost-model-based choice between KV-cache migration and recomputation, performs in-place weight resharding, and reuses cached communication groups. We implement PAT on top of SGLang and integrate it with the VeRL framework. Evaluations on LLaMA3.1-8B and Qwen3-14B using DeepScaleR show that PAT reduces generation latency by up to 34.6% and end-to-end RLHF training iteration latency by up to 27.2% compared to the original VeRL setup.

cs.AI↗

arXiv2026

The neutral scalars of type-II 2HDM+S under the LHC

The 2HDM+S is a singlet extension of the Two-Higgs-Doublet Model (2HDM), which offers rich collider phenomenology. In this paper, we parametrize the 2HDM+S with the Higgs masses and mixing angles, which provide a model-independent framework to study the collider signature. Under five benchmark scenarios, we obtain the 95\% C.L. exclusion regions in the Type-II 2HDM+S parameter space by incorporating the SM-like 125~GeV Higgs precision measurements, beyond the Standard Model Higgs direct searches, $Z$-pole precision measurements and $B$-physics observables. We present the results in the Higgs boson masses vs $\tanβ$, Higgs boson masses vs mixing angles, $\tanβ$ vs mixing angles and doublet Higgs boson masses vs singlet Higgs boson mass parameter space. We explore the complementarity between direct and indirect Higgs searches, as well as conventional Higgs search channels and exotic Higgs search channels. Compared to the 2HDM scenarios, we find that exotic channels such as $A/H \rightarrow Z h_S/ZA_S$ can probe large part of the parameter spaces, especially for moderate $1<\tanβ<7$ region where the conventional channels in the 2HDM cannot contribute much.

hep-ph↗

arXiv2026

Radial Profiles of Binary Fraction in Elliptical Galaxies

The radial profile of binary fraction may vary with environment and is of significant importance for studying the formation mechanisms of binary stars and their dynamical evolution within globular clusters (GCs) and galaxies. However, existing studies remain limited to the Milky Way and its neighboring galaxies. Leveraging the method proposed by Zhang et al. for estimating the variation of binary fraction from integrated spectral features, we analyze a sample of 513 elliptical galaxies drawn from the Mapping Nearby Galaxies at Apache Point Observatory (MaNGA) survey to measure their radial binary fraction profiles. Our results show that after accounting for the effect induced by radial variations in the stellar population (SP), the median SP-subtracted binary fraction, $r_{\rm b,sub}^{\rm med}$, becomes approximately flat. For nearly all elliptical galaxies in our sample, the variation in binary fraction relative to the galaxy center at $1R_e$ is less than 5%. No clear correlation is found between the binary fraction gradient and the gradients of SP properties. Moreover, we also compare differences between ultraviolet (UV) upturn and non-UV upturn galaxies. The overall binary fraction profiles and SP properties of the non-UV upturn galaxies in our sample are comparable to those of the UV upturn galaxies. This similarity may arise from the presence of residual star formation (RSF) in the non-UV upturn systems.

astro-ph.GA↗

arXiv2026

Mars Thermospheric Polar Warming at Aphelion: Dynamical Processes Studied Using M-GITM

Aphelion Thermospheric Polar Warming (TPW), first identified in 2024 by Mars Atmosphere and Volatile EvolutioN (MAVEN) observations, is a dynamical heating phenomenon in the Martian atmosphere that exists in the winter hemisphere near the aphelion solstice. Studying the formation mechanism of aphelion TPW will help us better understand the energy budget of the Martian thermosphere. In this study, we investigate aphelion TPW using the Mars Global Ionosphere Thermosphere global circulation model (M-GITM). The simulation results show that the local dust storms have little impact on the formation of aphelion TPW. The simulated thermospheric temperature difference between the polar region and the low-latitude region is considerably lower than the value observed, which suggests that some important atmospheric processes are not captured by M-GITM. To investigate potential causes, we conduct sensitivity tests on solar insolation, gravity waves, and model horizontal resolution. The sensitivity test on solar insolation shows that the magnitude of aphelion TPW increases with increasing solar insolation. We also find that gravity waves play a critical role in modulating dynamical heating, as their suppression increases the latitudinal temperature difference. Model resolution has minimal impact on polar warming but affects thermospheric structure at low latitudes. These findings highlight the importance of refining the representation of dynamical processes, especially the parameterization of subgrid-scale internal gravity waves in the Martian general circulation model to better capture thermospheric dynamics.

astro-ph.EP↗

arXiv2026

RetroInfer: A Vector Storage Engine for Scalable Long-Context LLM Inference

Recent large language models (LLMs) are rapidly extending their context windows, yet inference throughput lags due to increasing GPU memory and bandwidth demands. This is because the key-value (KV) cache, an intermediate structure storing token representations, grows linearly with context length and requires an iterative linear scan for attention computation. A promising direction to accelerate long-context inference is to exploit attention's inherent sparsity by offloading the KV cache to CPU memory and retrieving only a small subset of tokens important to the current generation step. However, prior sparse attention approaches struggle to balance accuracy and retrieval cost due to varying sparsity patterns and inefficient GPU-CPU memory management. We present RetroInfer, a vector storage engine that realizes a sparsity-based KV cache for long-context inference. RetroInfer introduces an Attention-aWare VEctor index (wave index), which fundamentally improves the tradeoff between attention accuracy and retrieval cost through tripartite attention approximation, accuracy-bound attention estimation, and segmented clustering. We also design the wave buffer, a GPU-CPU buffer manager that assigns computation and manages data across heterogeneous hardware. We evaluate RetroInfer across a range of models and workloads, demonstrating up to 4.4X decoding throughput over full attention at 120K context and up to 12.2X over sparse attention baselines at 1 million tokens -- all while preserving full-attention-level accuracy.

cs.LG↗

arXiv2026

Amorphous Nanoconfinement Enables Self-sustaining Sabatier Reaction at Ambient Conditions

The Sabatier reaction, the catalytic hydrogenation of CO2 into CH4, offers a cornerstone for carbon capture and utilization, and in-situ resource utilization during space exploration; however, it faces a fundamental thermodynamic-kinetic paradox: although highly exothermic, conventional catalysts still require continuous external heating to activate CO2 and maintain stable operation. Here we report an amorphous silica-embedded ruthenium catalyst that enables a long-term self-sustaining autothermal Sabatier reaction dispensing with external energy supply. Operating under ambient conditions, this system achieves a record-high CH4 yield of 0.50 mol gcat-1 h-1 with 100% selectivity, stable operation for over 2,000 hours, and a record-low catalyst bed temperature down to 100 oC. This exceptional self-sustaining behavior stems from the synergistic effect of the catalyst's ultralow effective thermal conductivity (0.27 W m-1 K-1), induced by amorphous nanoconfinement, and its superior intrinsic activity. This synergy generates localized hot spots at Ru sites while suppressing macroscopic heat loss. In situ measurements further reveal CH4 formation even at 54 oC and identify a *CO-mediated pathway for CO2 methanation. The reaction ignites readily with a lighter or focused sunlight and persists even under forced convection from an electric fan, demonstrating strong environmental tolerance. By removing the need for constant energy input, this "ignite-and-forget" system paves the way for decentralized Power-to-Gas systems and autonomous fuel production in resource-constrained environments like Mars.

cond-mat.mtrl-sci↗

arXiv2026

Fast and Provably Accurate Sequential Designs using Hilbert Space Gaussian Processes

Gaussian processes are widely used for accurate emulation of unknown surfaces in sequential design of expensive simulation experiments. Integrated mean squared error (IMSE) is an effective acquisition function for sequential designs based on Gaussian processes. However, existing approaches struggle with its implementation because the required integrals often lack closed-form expressions for most kernel functions. We propose a novel and computationally efficient Hilbert space Gaussian process approximation for the IMSE acquisition function, where a truncated eigenbasis representation of the integral enables closed-form evaluation. We establish sharp global non-asymptotic bounds for both the approximation error of isotropic kernels and the resulting error in the acquisition function. In a series of numerical experiments with $γ$-stabilizing, the proposed method achieves substantially lower prediction error and reduced computation time compared to existing benchmarks. These results demonstrate that the proposed Hilbert space Gaussian process framework provides an accurate and computationally efficient approach for Gaussian process based sequential design.

math.ST↗

arXiv2026

Search for Dark Matter in 2HDMS at LHC and future Lepton Colliders

We investigate the phenomenological prospects of the Two Higgs Doublet and Complex Singlet Scalar Extension (2HDMS) in the context of dark matter (DM) and Higgs phenomenology. The 2HDMS provides an enlarged Higgs sector along with a DM candidate. In this work, we perform an exhaustive scan to find representative benchmarks which are consistent with all theoretical and experimental constraints. We choose benchmarks with light, intermediate and massive DM masses and in some cases, also accommodate the 95 GeV excess in $b\bar{b}$ and $γγ$ channels observed at the Large Electron-Positron Collider (LEP) and Large Hadron Collider (LHC). We focus on the relevant signatures at the LHC and at proposed future lepton colliders including electron-positron and muon colliders. Using a cut and count analysis, we show that while the High Luminosity LHC (HL-LHC) may give a hint of new physics, future lepton colliders prove to be efficient discovery probes for the 2HDMS.

hep-ph↗

arXiv2026

AdaCluster: Adaptive Query-Key Clustering for Sparse Attention in Video Generation

Video diffusion transformers (DiTs) suffer from prohibitive inference latency due to quadratic attention complexity. Existing sparse attention methods either overlook semantic similarity or fail to adapt to heterogeneous token distributions across layers, leading to model performance degradation. We propose AdaCluster, a training-free adaptive clustering framework that accelerates the generation of DiTs while preserving accuracy. AdaCluster applies an angle-similarity-preserving clustering method to query vectors for higher compression, and designs a euclidean-similarity-preserving clustering method for keys, covering cluster number assignment, threshold-wise adaptive clustering, and efficient critical cluster selection. Experiments on CogVideoX-2B, HunyuanVideo, and Wan-2.1 on one A40 GPU demonstrate up to 1.67-4.31x speedup with negligible quality degradation.

cs.CV↗

arXiv2026

Bias-constrained multimodal intelligence for equitable and reliable clinical AI

The integration of medical imaging and clinical text has enabled the emergence of generalist artificial intelligence (AI) systems for healthcare. However, pervasive biases, such as imbalanced disease prevalence, skewed anatomical region distributions, heterogeneous imaging protocols, and demographic disparities, pose significant challenges to the fairness and reliability of vision-language systems in real-world clinical settings. Here we present BiasCareVL, a bias-aware multimodal learning framework that introduces bias control directly into model design, rather than treating it as a post hoc correction. BiasCareVL incorporates adaptive uncertainty modeling with optional human-in-the-loop refinement to regulate the influence of dominant data patterns and to promote equitable reasoning under distributional imbalance. Trained on 3.44 million samples spanning over 15 imaging modalities, the framework supports diverse clinical tasks, including visual question answering, disease classification, segmentation, and report generation within a unified representation space. Across eight public benchmarks covering dermatology, oncology, radiology, and pathology, BiasCareVL consistently outperforms 20 state-of-the-art methods, with pronounced gains in clinically challenging scenarios, including over 10% accuracy improvement in multi-class skin lesion diagnosis and more than 20% Dice improvement in small tumor segmentation. Furthermore, BiasCareVL achieves diagnostic performance exceeding human accuracy with substantially reduced time requirements when evaluated with board-certified radiologists. By open-sourcing BiasCareVL, we aim to promote a transparent, reproducible, and equitable future for AI in healthcare, paving the way for general-purpose, trustworthy, and clinically reliable AI systems.

cs.CV↗

arXiv2026

A comparison of the spin-phonon behaviour of Fe$_2$P-based magnetocaloric materials

Magnetic refrigeration can provide an environmentally friendly technology to reduce significantly the energy consumption of cooling devices. To retain the sustainability of the device, all parts must be made from abundant materials, excluding e.g. rare earth elements. As such, materials based on Fe$_2$P have shown great potential for magnetocaloric devices. In this study, Fe$_2$P and FeMnP$_{0.55}$Si$_{0.45}$, have been studied using magnetometry, neutron scattering and theoretical modelling with the aim to understand the ferromagnetic transition, related to the magnetocaloric effect. Analysis of the diffraction data of Fe$_2$P showed that it is the Fe$_{3g}$-site that drives the magnetic transition as the Fe$_{3f}$ does not have any magnetic contribution at the magnetic transition temperature. For FeMnP$_{0.55}$Si$_{0.45}$, the magnetic transition is more gradual, on both sites, with coexistence of the para- and ferromagnetic phases close to the magnetic transition. The temperature dependent magnetic structure behaviour are well in agreement with our first principles calculations. Both Fe$_2$P and FeMnP$_{0.55}$Si$_{0.45}$ showed two distinct regions, at different length scales, in their S(\textbf{Q},$ω$) spectra. The two length scales can be modelled using a different set of magnetic spin states (S), using S$\rm _{Fe}$~=~2 and S$\rm _{Mn}$~=~2.5, consistent with the ground state of the magnetic atoms. QENS at low Q (Q~\textless{}~0.5~Å) shows similar magnetic processes in both compounds with uncorrelated magnetism below the magnetic transition temperature. The uncorrelated state highlights that the magnetic anisotropy does not play a major role in the formation of the magnetic state. Furthermore, this emphasises the existence of a two part system in FeMn(P,Si)-based compounds, that drives the magnetic transition and in turn the magnetocaloric effect.

cond-mat.mtrl-sci↗