arXiv subjects

Haochen Wang

Publications and source records attributed to Haochen Wang.

At least 55 records · Page 3Linked to original sources

arXiv2026

Exploring selection biases in FRB dispersion-galaxy cross-correlations with magnetohydrodynamical simulations

The dispersion measure (DM) of fast radio bursts (FRBs) in conjunction with their redshifts can be used as powerful probes of the distribution of extragalactic plasma. With a large enough sample, the free-electron--galaxy power spectrum $P_{eg}$ can be measured by cross-correlating FRB DMs with galaxy positions. However, a precise measurement of $P_{eg}$ requires a careful investigation of selection effects: the probability of both observing the FRB DM and obtaining a host galaxy redshift depends on their properties. We ray trace through the magnetohydrodynamic simulation IllustrisTNG to investigate the impact of expected observational selection effects on FRB dispersion--galaxy angular cross-correlations with a sample of 3000 FRBs at $0.3\leq z\leq 0.4$ . Our results show that cross-correlations with such an FRB sample are robust to properties of the FRB host galaxy: this includes DM contributions from the FRB host and optical follow-up selection effects. We also find that such cross-correlations are robust to DM-dependent and scattering selection effects specific to the CHIME/FRB survey. However, a DM-dependent selection effect that cuts off the 10\% most dispersed FRBs at a fixed redshift shell can bias the amplitude of the cross-correlation signal by over 50\% at angular scales of $\sim 0.1^\circ$, corresponding to $\sim$ Mpc physical scales. Our findings highlight the importance of both measuring and accounting for selection effects present in existing FRB surveys, as well as mitigating DM-dependent selection effects in the design of upcoming FRB surveys aiming to probe large-scale structure with FRBs.

astro-ph.CO↗

arXiv2026

The Second CHIME/FRB Catalog of Fast Radio Bursts

We present a catalog of 4539 fast radio bursts (FRBs) observed with the Canadian Hydrogen Intensity Mapping Experiment (CHIME) telescope between 25 July 2018 and 15 September 2023. These bursts originate from 3641 unique sources, including 981 bursts from 83 known repeating sources. For each FRB, the catalog provides a $O(10')$ estimate of sky location along with corresponding measurements of cumulative exposure time and survey sensitivity over the observing period. It includes a total-intensity dynamic spectrum between 400 and 800 MHz at 0.983 ms resolution. From this spectrum, we constrain a model of the burst morphology and measure key parameters such as arrival time, intrinsic temporal width, dispersion measure, scattering time, and flux density. This second catalog includes all FRBs from the first catalog, with every event reprocessed using a uniform and improved analysis framework. We show that previously published inferences remain valid under the updated measurements. We assess consistency of the detection rate across observational parameters, present initial distributions of burst properties, and outline ongoing and future studies that will use this catalog to investigate the nature of FRBs and their utility as astrophysical and cosmological probes.

astro-ph.HE↗

arXiv2026

MMFormalizer: Multimodal Autoformalization in the Wild

Autoformalization, which translates natural language mathematics into formal statements to enable machine reasoning, faces fundamental challenges in the wild due to the multimodal nature of the physical world, where physics requires inferring hidden constraints (e.g., mass or energy) from visual elements. To address this, we propose MMFormalizer, which extends autoformalization beyond text by integrating adaptive grounding with entities from real-world mathematical and physical domains. MMFormalizer recursively constructs formal propositions from perceptually grounded primitives through recursive grounding and axiom composition, with adaptive recursive termination ensuring that every abstraction is supported by visual evidence and anchored in dimensional or axiomatic grounding. We evaluate MMFormalizer on a new benchmark, PhyX-AF, comprising 115 curated samples from MathVerse, PhyX, Synthetic Geometry, and Analytic Geometry, covering diverse multimodal autoformalization tasks. Results show that frontier models such as GPT-5 and Gemini-3-Pro achieve the highest compile and semantic accuracy, with GPT-5 excelling in physical reasoning, while geometry remains the most challenging domain. Overall, MMFormalizer provides a scalable framework for unified multimodal autoformalization, bridging perception and formal reasoning. To the best of our knowledge, this is the first multimodal autoformalization method capable of handling classical mechanics (derived from the Hamiltonian), as well as relativity, quantum mechanics, and thermodynamics. More details are available on our project page: MMFormalizer.github.io

cs.CL↗

arXiv2026

The Squeezed Bispectrum from CHIME HI Emission and Planck CMB Lensing: Current Sensitivity and Forecasts

Line intensity mapping using atomic hydrogen (HI) has the potential to efficiently map large volumes of the universe if the signal can be successfully separated from overwhelmingly bright radio foreground emission. This motivates cross-correlations, to ascertain the cosmological nature of measured HI fluctuations, and to study their connections with galaxies and the underlying matter density field. However, these same foregrounds render the cross-correlation with projected fields such as the lensing of the cosmic microwave background (CMB) difficult. Indeed, the correlated Fourier modes vary slowly along the line of sight, and are thus most contaminated by the smooth-spectrum radio continuum foregrounds. In this paper, we implement a method that avoids this issue by attempting to measure the non-linear gravitational coupling of the small-scale 21cm power from the Canadian Hydrogen Intensity Mapping Experiment (CHIME) with large-scale Planck CMB lensing. This measurement is a position-dependent power spectrum, i.e. a squeezed integrated bispectrum. Using 94 nights of CHIME data between $1.0 < z < 1.3$ and aggressive foreground filtering, we find that the expected signal is five times smaller than the current noise. We forecast that incorporating the additional nights of CHIME data already collected would enable a signal-to-noise ratio of 3, without any further improvements in filtering for foreground cleaning.

astro-ph.CO↗

arXiv2025

DriveVLA-W0: World Models Amplify Data Scaling Law in Autonomous Driving

Scaling Vision-Language-Action (VLA) models on large-scale data offers a promising path to achieving a more generalized driving intelligence. However, VLA models are limited by a ``supervision deficit'': the vast model capacity is supervised by sparse, low-dimensional actions, leaving much of their representational power underutilized. To remedy this, we propose \textbf{DriveVLA-W0}, a training paradigm that employs world modeling to predict future images. This task generates a dense, self-supervised signal that compels the model to learn the underlying dynamics of the driving environment. We showcase the paradigm's versatility by instantiating it for two dominant VLA archetypes: an autoregressive world model for VLAs that use discrete visual tokens, and a diffusion world model for those operating on continuous visual features. Building on the rich representations learned from world modeling, we introduce a lightweight action expert to address the inference latency for real-time deployment. Extensive experiments on the NAVSIM v1/v2 benchmark and a 680x larger in-house dataset demonstrate that DriveVLA-W0 significantly outperforms BEV and VLA baselines. Crucially, it amplifies the data scaling law, showing that performance gains accelerate as the training dataset size increases.

cs.CV↗

arXiv2025

Electrical Stability of Cr2O3/\b{eta}-Ga2O3 and NiOx/\b{eta}-Ga2O3 Heterojunction Diodes

This work reports the electrical characteristics comparison study between Cr2O3 and NiOx based heterojunction diodes (HJD) on halide vapor phase epitaxy (HVPE) grown \b{eta}-Ga2O3 epitaxial layers. Both as-fabricated Cr2O3 and NiOx HJDs exhibited forward current density in a range of 130-150 A/cm^2 at 5 V with rectifying ratios >10^10 and a reverse leakage current density at 10^-8 A/cm^2 at -5 V. The differential specific on-resistance of Cr2O3 and NiOx HJDs was 12.01 mΩ*cm^2 and 12.05 mΩ*cm^2, respectively. Breakdown voltages of Cr2O3 HJDs ranged from 1.4-1.9 kV and 1.5-2.3 kV for NiOx HJDs. Theoretical band alignment between Cr2O3 and \b{eta}-Ga2O3 was calculated from first principles. The ambient exposed NiOx/HVPE \b{eta}-Ga2O3 HJDs forward current density degraded after 10 days while that of Cr2O3/HVPE \b{eta}-Ga2O3 HJDs remained nearly unchanged after the same amount of time. It was later confirmed that the ambient exposed sputtered NiOx sheet resistance (Rsh) degradation gave rise to the reduction of the forward current density of the NiOx based HJDs, and water (H2O) was qualitatively determined to be the agent attributed to the forward conduction degradation by measuring the Rsh of NiOx-on-sapphire reference wafer after exposing it to different environments. The Cr2O3/HVPE \b{eta}-Ga2O3 HJD also exhibited enhanced thermal stability compared to the NiOx/\b{eta}-Ga2O3 heterostructures at elevated temperatures. Interfacial nickel gallate (Ga2NiO4) phase formation expected from phase diagrams can explain the reduced thermal stability of NiOx/\b{eta}-Ga2O3 HJDs. This study indicates that Cr2O3 is a stable p-type oxide for the realization of robust multi-kV \b{eta}-Ga2O3 HJDs.

physics.app-ph↗

arXiv2025

CrossVid: A Comprehensive Benchmark for Evaluating Cross-Video Reasoning in Multimodal Large Language Models

Cross-Video Reasoning (CVR) presents a significant challenge in video understanding, which requires simultaneous understanding of multiple videos to aggregate and compare information across groups of videos. Most existing video understanding benchmarks focus on single-video analysis, failing to assess the ability of multimodal large language models (MLLMs) to simultaneously reason over various videos. Recent benchmarks evaluate MLLMs' capabilities on multi-view videos that capture different perspectives of the same scene. However, their limited tasks hinder a thorough assessment of MLLMs in diverse real-world CVR scenarios. To this end, we introduce CrossVid, the first benchmark designed to comprehensively evaluate MLLMs' spatial-temporal reasoning ability in cross-video contexts. Firstly, CrossVid encompasses a wide spectrum of hierarchical tasks, comprising four high-level dimensions and ten specific tasks, thereby closely reflecting the complex and varied nature of real-world video understanding. Secondly, CrossVid provides 5,331 videos, along with 9,015 challenging question-answering pairs, spanning single-choice, multiple-choice, and open-ended question formats. Through extensive experiments on various open-source and closed-source MLLMs, we observe that Gemini-2.5-Pro performs best on CrossVid, achieving an average accuracy of 50.4%. Notably, our in-depth case study demonstrates that most current MLLMs struggle with CVR tasks, primarily due to their inability to integrate or compare evidence distributed across multiple videos for reasoning. These insights highlight the potential of CrossVid to guide future advancements in enhancing MLLMs' CVR capabilities.

cs.CV↗

arXiv2025

Cr2O3/\b{eta}-Ga2O3 Heterojunction Diodes with Orientation-Dependent Breakdown Electric Field up to 12.9 MV/cm

We report the fabrication of Cr2O3/\b{eta}-Ga2O3 heterojunction diodes using reactive magnetron sputtering of Cr2O3 on highly doped \b{eta}-Ga2O3 bulk substrates along (100), (010), (001), (110), and (011) orientation dependence of high electric field handling capability in \b{eta}-Ga2O3. Additional relative permittivity values in (110) and (011) orientations of \b{eta}-Ga2O3 were computed by using first-principles calculation methods for accurate apparent charge density (ND-NA) extraction and breakdown electric field analysis from capacitance-voltage measurements. The HJDs fabricated on n+ (110) exhibited breakdown electric fields >10 MV/cm up to 12.9 MV/cm, showing the highest experimentally observed parallel-plane junction electric field among \b{eta}-Ga2O3-based junctions. Breakdown electric fields among (100), (010), (001), and (011) orientations showed distinct distribution in the range of 5.13-5.26 MV/cm, 5.10-7.05 MV/cm, 2.70-3.33 MV/cm, and 3.88-4.38 MV/cm, respectively, validating the orientational dependence of parallel-plane junction electric field at breakdown in low-symmetry monoclinic \b{eta}-Ga2O3. The parallel-plane breakdown electric fields (EBr,||) reported in this work were extracted when the device experienced catastrophic breakdown at 100 mA/cm^2 current density compliance, and should not be confused with critical electric field (Ec) as a function of drift layer doping concentration, which accounts for electric-field dependent impact ionization coefficients in Si, SiC and GaN. This study can guide the choice of crystal orientation for high performance gallium oxide-based devices that require high electric field handling capability.

cond-mat.mtrl-sci↗

arXiv2025

MMaDA-Parallel: Multimodal Large Diffusion Language Models for Thinking-Aware Editing and Generation

While thinking-aware generation aims to improve performance on complex tasks, we identify a critical failure mode where existing sequential, autoregressive approaches can paradoxically degrade performance due to error propagation. To systematically analyze this issue, we propose ParaBench, a new benchmark designed to evaluate both text and image output modalities. Our analysis using ParaBench reveals that this performance degradation is strongly correlated with poor alignment between the generated reasoning and the final image. To resolve this, we propose a parallel multimodal diffusion framework, MMaDA-Parallel, that enables continuous, bidirectional interaction between text and images throughout the entire denoising trajectory. MMaDA-Parallel is trained with supervised finetuning and then further optimized by Parallel Reinforcement Learning (ParaRL), a novel strategy that applies semantic rewards along the trajectory to enforce cross-modal consistency. Experiments validate that our model significantly improves cross-modal alignment and semantic consistency, achieving a 6.9\% improvement in Output Alignment on ParaBench compared to the state-of-the-art model, Bagel, establishing a more robust paradigm for thinking-aware image synthesis. Our code is open-sourced at https://github.com/tyfeld/MMaDA-Parallel

cs.CV↗

arXiv2025

MVU-Eval: Towards Multi-Video Understanding Evaluation for Multimodal LLMs

The advent of Multimodal Large Language Models (MLLMs) has expanded AI capabilities to visual modalities, yet existing evaluation benchmarks remain limited to single-video understanding, overlooking the critical need for multi-video understanding in real-world scenarios (e.g., sports analytics and autonomous driving). To address this significant gap, we introduce MVU-Eval, the first comprehensive benchmark for evaluating Multi-Video Understanding for MLLMs. Specifically, our MVU-Eval mainly assesses eight core competencies through 1,824 meticulously curated question-answer pairs spanning 4,959 videos from diverse domains, addressing both fundamental perception tasks and high-order reasoning tasks. These capabilities are rigorously aligned with real-world applications such as multi-sensor synthesis in autonomous systems and cross-angle sports analytics. Through extensive evaluation of state-of-the-art open-source and closed-source models, we reveal significant performance discrepancies and limitations in current MLLMs' ability to perform understanding across multiple videos. The benchmark will be made publicly available to foster future research.

cs.CV↗

arXiv2025

FastMap: Revisiting Structure from Motion through First-Order Optimization

We propose FastMap, a new global structure from motion method focused on speed and simplicity. Previous methods like COLMAP and GLOMAP are able to estimate high-precision camera poses, but suffer from poor scalability when the number of matched keypoint pairs becomes large, mainly due to the time-consuming process of second-order Gauss-Newton optimization. Instead, we design our method solely based on first-order optimizers. To obtain maximal speedup, we identify and eliminate two key performance bottlenecks: computational complexity and the kernel implementation of each optimization step. Through extensive experiments, we show that FastMap is up to 10 times faster than COLMAP and GLOMAP with GPU acceleration and achieves comparable pose accuracy.

cs.CV↗

arXiv2025

Object-centric Video Question Answering with Visual Grounding and Referring

Video Large Language Models (VideoLLMs) have recently demonstrated remarkable progress in general video understanding. However, existing models primarily focus on high-level comprehension and are limited to text-only responses, restricting the flexibility for object-centric, multiround interactions. In this paper, we make three contributions: (i) we address these limitations by introducing a VideoLLM model, capable of performing both object referring for input and grounding for output in video reasoning tasks, i.e., allowing users to interact with videos using both textual and visual prompts; (ii) we propose STOM (Spatial-Temporal Overlay Module), a novel approach that propagates arbitrary visual prompts input at any single timestamp to the remaining frames within a video; (iii) we present VideoInfer, a manually curated object-centric video instruction dataset featuring questionanswering pairs that require reasoning. We conduct comprehensive experiments on VideoInfer and other existing benchmarks across video question answering and referring object segmentation. The results on 12 benchmarks of 6 tasks show that our proposed model consistently outperforms baselines in both video question answering and segmentation, underscoring its robustness in multimodal, object-centric video and image understanding. Project page: https://qirui-chen.github.io/RGA3-release/.

cs.CV↗

arXiv2025

Stellar Mass-Dispersion Measure Correlations Constrain Baryonic Feedback in Fast Radio Burst Host Galaxies

Low redshift fast radio bursts (FRBs) provide robust measurements of the host-galaxy contribution to the dispersion measure (DM), which can constrain the circumgalactic medium (CGM) of the hosts. We curate a sample of 20 nearby FRBs with low scattering timescales and face-on host galaxies with stellar masses ranging from $10^9 < M^* / M_\odot < 10^{11}$. We fit the distribution of the host galaxy DM to a quadratic model as a function of stellar mass with a mass-independent scatter and find that the more massive the host, the lower its host DM. We report that this relation has a negative slope of $m = -97 \pm 44$ pc/cm$^{-3}$ per dex in stellar mass. We compare this measurement to similar fits to three sub-grid models implemented in the CAMELS suite of simulations from Astrid, IllustrisTNG, and SIMBA and find that fine-tuning of the host ISM contribution as a function of stellar mass is required in order to reconcile the observational data with the predictions of the fiducial CAMELS-Astrid model. More generally, models which attribute a positive correlation between stellar mass and host dispersion measure ($m > 0$) to the CGM are in tension with our measurement. We show that this conclusion is robust to a wide range of assumptions, such as the offset distribution of FRBs from their hosts and the statistics of the cosmic contribution to the DM budget along each sightline. Our results indirectly imply a lower limit on the strength of baryonic feedback in the Local Universe $(z < 0.2)$ in isolated $\sim L^*$ halos, complementing results from weak lensing surveys and kSZ observations which target higher halo mass and redshift ranges.

astro-ph.GA↗

arXiv2025

Hita: Holistic Tokenizer for Autoregressive Image Generation

Vanilla autoregressive image generation models generate visual tokens step-by-step, limiting their ability to capture holistic relationships among token sequences. Moreover, because most visual tokenizers map local image patches into latent tokens, global information is limited. To address this, we introduce \textit{Hita}, a novel image tokenizer for autoregressive (AR) image generation. It introduces a holistic-to-local tokenization scheme with learnable holistic queries and local patch tokens. Hita incorporates two key strategies to better align with the AR generation process: 1) {arranging} a sequential structure with holistic tokens at the beginning, followed by patch-level tokens, and using causal attention to maintain awareness of previous tokens; and 2) adopting a lightweight fusion module before feeding the de-quantized tokens into the decoder to control information flow and prioritize holistic tokens. Extensive experiments show that Hita accelerates the training speed of AR generators and outperforms those trained with vanilla tokenizers, achieving \textbf{2.59 FID} and \textbf{281.9 IS} on the ImageNet benchmark. Detailed analysis of the holistic representation highlights its ability to capture global image properties, such as textures, materials, and shapes. Additionally, Hita also demonstrates effectiveness in zero-shot style transfer and image in-painting. The code is available at \href{https://github.com/CVMI-Lab/Hita}{https://github.com/CVMI-Lab/Hita}.

cs.CV↗

arXiv2025

FRB 20250316A: A Brilliant and Nearby One-Off Fast Radio Burst Localized to 13 parsec Precision

Precise localizations of a small number of repeating fast radio bursts (FRBs) using very long baseline interferometry (VLBI) have enabled multiwavelength follow-up observations revealing diverse local environments. However, the 2--3\% of FRB sources that are observed to repeat may not be representative of the full population. Here we use the VLBI capabilities of the full CHIME Outriggers array for the first time to localize a nearby (40 Mpc), bright (kJy), and apparently one-off FRB source, FRB 20250316A, to its environment on 13-pc scales. We use optical and radio observations to place deep constraints on associated transient emission and the properties of its local environment. We place a $5σ$ upper limit of $L_{\mathrm{9.9~\mathrm{GHz}}} < 2.1\times10^{25}~\mathrm{erg~s^{-1}~Hz^{-1}}$ on spatially coincident radio emission, a factor of 100 lower than any known compact persistent radio source associated with an FRB. Our KCWI observations allow us to characterize the gas density, metallicity, nature of gas ionization, dust extinction and star-formation rate through emission line fluxes. We leverage the exceptional brightness and proximity of this source to place deep constraints on the repetition of FRB 20250316A, and find it is inconsistent with all well-studied repeaters given the non-detection of bursts at lower spectral energies. We explore the implications of a measured offset of 190$\pm20$ pc from the center of the nearest star-formation region, in the context of progenitor channels. FRB 20250316A marks the beginning of an era of routine localizations for one-off FRBs on tens of mas-scales, enabling large-scale studies of their local environments.

astro-ph.HE↗

arXiv2025

Measurement of the Dispersion$\unicode{x2013}$Galaxy Cross-Power Spectrum with the Second CHIME/FRB Catalog

The dispersion of extragalactic fast radio bursts (FRBs) can serve as a powerful probe of the diffuse plasma between and surrounding galaxies, which contains most of the Universe's baryons. By cross-correlating the dispersion of background FRBs with the locations of foreground galaxies, we can study the relative spatial distributions of plasma and galaxies on scales of 0.1 to 50 Mpc, which are strongly affected by feedback processes in galaxy formation. Here we present the measurement of the dispersion$\unicode{x2013}$galaxy angular cross-power spectrum between 2873 FRBs from the Second CHIME/FRB Catalog and nearly 6 million galaxies from the Dark Energy Spectroscopic Instrument (DESI) Legacy Imaging Survey. Over five photometric galaxy redshift bins spanning $0.05 < z <0.5$ and at 5.1$σ$ significance, we make the first definitive detection of spatial correlations in FRB dispersion measure due to cosmic structure. While parameter inferences should be interpreted with caution because of incomplete modelling of both the signal and systematic errors, our data indicate that the plasma$\unicode{x2013}$galaxy cross-power spectrum cuts off relative to the matter power spectrum at a scale $k_\textrm{cut}^{-1}=0.9^{+0.4}_{-0.4}\,\textrm{Mpc}$. This scale is consistent with those X-ray stacking analyses that suggest dark-matter halos with group-scale masses are largely evacuated of their baryons by feedback processes. Our study demonstrates that FRBs are promising tools to discern the physics of baryonic structure formation and will only become more powerful as FRB surveys expand.

astro-ph.CO↗

arXiv2025

Mitigating antenna gain errors with HyFoReS in CHIME simulations

Hybrid Foreground Residual Subtraction (HyFoReS) is a new family of algorithms designed to remove systematics-induced foreground contamination for 21-cm intensity mapping data. Previously, the algorithm was shown to be effective in mitigating beam perturbations in sky maps from the Canadian Hydrogen Intensity Mapping Experiment (CHIME). In this study, we apply HyFoReS to CHIME simulations and test the algorithm's ability to mitigate antenna gain-type systematics in polarized visibilities. Simulating a two-cylinder telescope similar to the CHIME pathfinder, we find that HyFoReS reduces foreground bias caused by bandpass perturbations to a level below the thermal noise, provided that the RMS value of the perturbations is on the order of $10^{-4}$ or lower. When tested with complex antenna-dependent gain errors, HyFoReS can reduce residual foreground bias in the power spectrum by up to three orders of magnitude. While noise bias and second-order perturbations are currently the limiting factors for the algorithm, we have demonstrated that HyFoReS can suppress gain-induced foreground leakage in polarized data from 21-cm telescopes, aiding in the detection of the 21-cm auto-power spectrum for hydrogen intensity mapping experiments.

astro-ph.IM↗

arXiv2025

Integrating Learning-Based Manipulation and Physics-Based Locomotion for Whole-Body Badminton Robot Control

Learning-based methods, such as imitation learning (IL) and reinforcement learning (RL), can produce excel control policies over challenging agile robot tasks, such as sports robot. However, no existing work has harmonized learning-based policy with model-based methods to reduce training complexity and ensure the safety and stability for agile badminton robot control. In this paper, we introduce Hamlet, a novel hybrid control system for agile badminton robots. Specifically, we propose a model-based strategy for chassis locomotion which provides a base for arm policy. We introduce a physics-informed "IL+RL" training framework for learning-based arm policy. In this train framework, a model-based strategy with privileged information is used to guide arm policy training during both IL and RL phases. In addition, we train the critic model during IL phase to alleviate the performance drop issue when transitioning from IL to RL. We present results on our self-engineered badminton robot, achieving 94.5% success rate against the serving machine and 90.7% success rate against human players. Our system can be easily generalized to other agile mobile manipulation tasks such as agile catching and table tennis. Our project website: https://dreamstarring.github.io/HAMLET/.

cs.RO↗