arXiv subjects

Yu Chen

Publications and source records attributed to Yu Chen.

At least 127 records · Page 7Linked to original sources

arXiv2026

OmniVideoBench: Towards Audio-Visual Understanding Evaluation for Omni MLLMs

Recent advances in multimodal large language models (MLLMs) have demonstrated substantial potential in video understanding. However, existing benchmarks fail to comprehensively evaluate synergistic reasoning capabilities across audio and visual modalities, often neglecting either one of the modalities or integrating them in a logically inconsistent manner. To bridge this gap, we introduce OmniVideoBench, a large-scale and rigorously designed benchmark dedicated to assessing synergistic audio-visual understanding, with a strong emphasis on modality complementarity and logical consistency. Specifically, OmniVideoBench comprises 1000 high-quality question-answer(QA) pairs, each annotated with step-by-step reasoning traces, derived from 628 diverse videos ranging from several seconds to 30 minutes, and manually verified to guarantee complete correctness and uniqueness. Moreover, OmniVideoBench encompasses 13 carefully designed question types, covering temporal reasoning, spatial localization, counting, causal inference, summarization, and beyond, thereby capturing the essential challenges of video understanding. Evaluation of multiple MLLMs on OmniVideoBench reveals a pronounced gap between model performance and human reasoning, with open-source models lagging significantly behind their closed-source counterparts, underscoring the inherent difficulty of genuine audio-visual reasoning. We will release OmniVideoBench to foster the development of MLLMs with stronger and more generalizable reasoning capabilities.

cs.AI↗

arXiv2026

Hitting time for Hamilton cycles in pseudorandom graphs

Consider the random subgraph process on a base graph $G$ with $n$ vertices: we generate a sequence $\{G_t\}_{t=0}^{|E(G)|}$ by taking a uniformly random ordering of the edges of $G$ and then adding these edges one by one to the empty graph $G_0$ on the same vertex set. We prove that there is a constant $C > 0$ such that if $G$ is an $(n,d,λ)$-graph with $d/λ\ge C$, then with high probability, the hitting time for the appearance of a Hamilton cycle coincides with the hitting time for reaching minimum degree $2$. This resolves questions posed by Alon--Krivelevich in 2019 and by Frieze--Krivelevich in 2002. As a consequence, we determine the sharp threshold for Hamilton cycles in $(n,d,λ)$-graphs with $d/λ\ge C$ for all $d$ sufficiently large. Lastly, we extend our result to the minimum degree $2k$ versus $k$ edge-disjoint Hamilton cycles setting for $k \leq c\cdot \min\{d,\log n\}$ where $c$ is a constant depending on $C$. This advances on a question asked by Frieze.

math.CO↗

arXiv2026

Damage Prediction of Sintered α-SiC Using Thermo-mechanical Coupled Fracture Model

A three-way coupled thermo-mechanical fracture model is presented to predict the damage of brittle ceramics, in particular α-SiC, over a wide range of temperatures (20-1400 C). Predicting damage over such a range of temperatures is crucial for thermal protection systems for many systems such as spacecraft. The model, which has been implemented in MOOSE, is divided into three modules: elasticity, damage phase field, and heat conduction. Analytical approaches for determining crack length scales are presented for both simple tension and simple shear. Validation tests are conducted for both flexural strength and fracture toughness over the specified range of temperatures. Flexural strength simulation results fall within the uncertainty region of the experimental data, and mode I fracture toughness simulation results are also in agreement with the experimental data. Mode II and mixed mode fracture toughness simulations results are presented with the modified G-criterion. Finally, the parallel computing capabilities of the model is considered in various scalability tests.

cond-mat.mtrl-sci↗

arXiv2026

The effect of chemical vapor infiltration process parameters on flexural strength of porous α-SiC: A numerical model

The flexural strength variability of α-SiC based ceramics at elevated temperatures creates the need for an Integrated Computational Materials Engineering (ICME) framework that relates the strength of a specimen directly to its manufacturing process. To create this ICME framework a model must first be developed which establishes a relationship between the chemical vapor infiltration (CVI) process and parameters, the resulting mesoscale pores, and the overall macroscale flexural strength. Here a nonlinear single pore model of CVI is developed used in conjunction with a four-way coupled themo-mechanical damage model. The individual components of the model are tested and a sample system under a four-point bending test is explored. Results indicate that specimens with an initial porosity greater than 30% require temperatures below 1273 K to maintain structural integrity, while those with initial porosities less than 30% are temperature-independent, allowing for optimization of the CVI processing time without compromising strength.

cond-mat.mtrl-sci↗

arXiv2026

GFRRN: Explore the Gaps in Single Image Reflection Removal

Prior dual-stream methods with the feature interaction mechanism have achieved remarkable performance in single image reflection removal (SIRR). However, they often struggle with (1) semantic understanding gap between the features of pre-trained models and those of reflection removal models, and (2) reflection label inconsistencies between synthetic and real-world training data. In this work, we first adopt the parameter efficient fine-tuning (PEFT) strategy by integrating several learnable Mona layers into the pre-trained model to align the training directions. Then, a label generator is designed to unify the reflection labels for both synthetic and real-world data. In addition, a Gaussian-based Adaptive Frequency Learning Block (G-AFLB) is proposed to adaptively learn and fuse the frequency priors, and a Dynamic Agent Attention (DAA) is employed as an alternative to window-based attention by dynamically modeling the significance levels across windows (inter-) and within an individual window (intra-). These components constitute our proposed Gap-Free Reflection Removal Network (GFRRN). Extensive experiments demonstrate the effectiveness of our GFRRN, achieving superior performance against state-of-the-art SIRR methods.

cs.CV↗

arXiv2026

World Simulation with Video Foundation Models for Physical AI

We introduce [Cosmos-Predict2.5], the latest generation of the Cosmos World Foundation Models for Physical AI. Built on a flow-based architecture, [Cosmos-Predict2.5] unifies Text2World, Image2World, and Video2World generation in a single model and leverages [Cosmos-Reason1], a Physical AI vision-language model, to provide richer text grounding and finer control of world simulation. Trained on 200M curated video clips and refined with reinforcement learning-based post-training, [Cosmos-Predict2.5] achieves substantial improvements over [Cosmos-Predict1] in video quality and instruction alignment, with models released at 2B and 14B scales. These capabilities enable more reliable synthetic data generation, policy evaluation, and closed-loop simulation for robotics and autonomous systems. We further extend the family with [Cosmos-Transfer2.5], a control-net style framework for Sim2Real and Real2Real world translation. Despite being 3.5$\times$ smaller than [Cosmos-Transfer1], it delivers higher fidelity and robust long-horizon video generation. Together, these advances establish [Cosmos-Predict2.5] and [Cosmos-Transfer2.5] as versatile tools for scaling embodied intelligence. To accelerate research and deployment in Physical AI, we release source code, pretrained checkpoints, and curated benchmarks under the NVIDIA Open Model License at https://github.com/nvidia-cosmos/cosmos-predict2.5 and https://github.com/nvidia-cosmos/cosmos-transfer2.5. We hope these open resources lower the barrier to adoption and foster innovation in building the next generation of embodied intelligence.

cs.CV↗

arXiv2026

Universal observable as a signal of chiral anomaly in lattice Weyl fermions

The Adler-Bell-Jackiw chiral anomaly is shown to retain its Lorentz-invariant form, $\partial_μJ^μ_5 \propto \mathbf{E} \cdot \mathbf{B}$, in lattice Weyl systems beyond moderate magnetic fields, where neither Lorentz nor rotational symmetry is present. We show that the longitudinal and Hall magnetoconductivities factorize into a product of a universal part, governed by the chiral anomaly, and a non-universal part that depends on the density of states at the Fermi level. A rotationally invariant observable $\varkappa = σ(c_V/T)^2$ is introduced as a robust signature of the anomaly, where $σ$ denotes the Euclidean norm of the longitudinal and Hall conductivities and $c_V$ is the specific heat density. This quantity follows a universal $B^2$ dependence and scales as $|\cosΘ|$, with $Θ$ being the angle between $\mathbf{E}$ and $\mathbf{B}$. Through analytical derivation and full numerical simulation, we establish that $\varkappa$ remains universal independent of system parameters and of the orientation of the magnetic or electric field for fixed $Θ$. The emergent SO(3) symmetry in $\varkappa$ persists despite the absence of isotropy in both the microscopic model and the low-energy effective theory.

cond-mat.mes-hall↗

arXiv2026

Dissipative Spectroscopy

We introduce dissipative spectroscopy as a framework for extracting spectral information from quantum systems via controlled dissipation. By establishing a general dissipative response theory applicable to both Markovian and non-Markovian environments, we develop a protocol to access the dissipative spectrum (DS) through driven oscillation-dissipation resonance. We show that the DS can identify two-particle soft modes near quantum critical points and, on the normal-phase side, predict the emergence of macroscopic order exhibiting power-law growth following a dissipation quench. These distinctive signatures appear in quasiparticle-dominant regimes, previously considered trivial. Furthermore, we introduce extended dissipative susceptibilities that capture leading memory effects and demonstrate their utility in a dissipative fermionic model. Our results indicate that the DS is readily accessible and offers a versatile tool for probing equilibrium properties as well as predicting nonequilibrium dissipative dynamics.

quant-ph↗

arXiv2026

Single Image Reflection Separation via Dual Prior Interaction Transformer

Single image reflection separation aims to separate the transmission and reflection layers from a mixed image. Existing methods typically combine general priors from pre-trained models with task-specific priors such as text prompts and reflection detection. However, the transmission prior, as the most direct task-specific prior for the target transmission layer, has not been effectively modeled or fully utilized, limiting performance in complex scenarios. To address this issue, we propose a dual-prior interaction framework based on lightweight transmission prior generation and effective prior fusion. First, we design a Local Linear Correction Network (LLCN) that finetunes pre-trained models based on the physical constraint T=SI+B, where S and B represent pixel-wise and channel-wise scaling and bias transformations. LLCN efficiently generates high-quality transmission priors with minimal parameters. Second, we construct a Dual-Prior Interaction Transformer (DPIT) that employs a dual-stream channel reorganization attention mechanism. By reorganizing features from general and transmission priors for attention computation, DPIT achieves deep fusion of both priors, fully exploiting their complementary information. Experimental results on multiple benchmark datasets demonstrate that the proposed method achieves state-of-the-art performance.

cs.CV↗

arXiv2026

Measuring Spin-Charge Separation by an Off-diagonal Dissipative Response

Fractionalization of symmetry - exemplified by spin-charge separation in the 1D Hubbard model and fractional charges in the fractional quantum Hall effect - is a typical strongly correlated phenomena in quantum many-body systems. Despite the success in measuring velocity differences, however, it is still quite challenging in probing emergent excitations' anomalous dimensions experimentally. We propose a off-diagonal dissipative response protocol, leveraging dissipative response theory (DRT), to directly detect spin-charge separation. By selectively dissipating spin-$\downarrow$ particles and measuring the spin-$\uparrow$ response, we uncover a universal temporal signature: the off-diagonal response exhibits a crossover from cubic-in-time ($t^3$) growth at short times to linear-in-time ($t$) decay at long times. Crucially, the coefficients $\varkappa^s$ (short-time) and $\varkappa^l$ (long-time) encode the distinct anomalous dimensions and velocities of spinons and holons, providing unambiguous evidence of fractionalization. This signal vanishes trivially without spin-charge separation. Our predictions, verified numerically via tDMRG, with microscopic parameters linking with Luttinger parameters by Bethe ansatz, establish off-diagonal dissipative response as a probe of quantum fractionalization in synthetic quantum matter.

cond-mat.quant-gas↗

arXiv2026

MLLM-CTBench: A Benchmark for Continual Instruction Tuning with Reasoning Process Diagnosis

Continual instruction tuning(CIT) during the post-training phase is crucial for adapting multimodal large language models (MLLMs) to evolving real-world demands. However, the progress is hampered by the lack of benchmarks with rigorous, protocol-consistent evaluation. To bridge this gap, we introduce MLLM-CTBench, a comprehensive benchmark for CIT of MLLMs, covering seven challenging tasks across six diverse domains. MLLM-CTBench makes three key contributions. First, we establish a multidimensional evaluation framework that jointly assesses final-answer accuracy and process-level reasoning quality, where Chain-of-Thought (CoT) traces serve as an observable signal to diagnose catastrophic forgetting beyond answer-only evaluation. Second, we conduct a large-scale evaluation of continual learning methods by systematically assessing eight representative algorithms from four major families under a unified protocol across task orders, providing actionable insights for algorithm design. Third, we expand the scope from Supervised Fine-Tuning (SFT) to Reinforcement Fine-Tuning (RFT) in CIT. By investigating GRPO, an on-policy RL algorithm that stabilizes updates through explicit KL-divergence control to a prior policy, we aim to analyze how this mechanism affects cross-task knowledge retention. Our experiments yield several findings:(1) Process-level reasoning quality is often more resilient to catastrophic forgetting than final-answer accuracy, and forgetting is primarily driven by degradation in domain knowledge. (2) Model capability is critical factor influencing continual learning outcomes, with stronger baseline models exhibiting greater resistance to catastrophic forgetting. (3) On-policy RFT (GRPO), with its inherent KL control, achieves more stable cross-task retention than SFT. While removing KL control can amplify forgetting despite potential gains on new ones.

cs.CL↗

arXiv2026

Lower Bounds on Flow Sparsifiers with Steiner Nodes

Given a large graph $G$ with a set of its $k$ vertices called terminals, a \emph{quality-$q$ flow sparsifier} is a small graph $G'$ that contains the terminals and preserves all multicommodity flows between them up to some multiplicative factor $q\ge 1$, called the \emph{quality}. Constructing flow sparsifiers with good quality and small size ($|V(G')|$) has been a central problem in graph compression. The most common approach of constructing flow sparsifiers is contraction: first compute a partition of the vertices in $V(G)$, and then contract each part into a supernode to obtain $G'$. When $G'$ is only allowed to contain all terminals, the best quality is shown to be $O(\log k/\log\log k)$ and $Ω(\sqrt{\log k/\log\log k})$. In this paper, we show that allowing a few Steiner nodes does not help much in improving the quality. Specifically, there exist $k$-terminal graphs such that, even if we allow $k\cdot 2^{(\log k)^{Ω(1)}}$ Steiner nodes in its contraction-based flow sparsifier, the quality is still $Ω\big((\log k)^{0.3}\big)$.

cs.DS↗

arXiv2026

SA-CAISR: Stage-Adaptive and Conflict-Aware Incremental Sequential Recommendation

Sequential recommendation (SR) aims to predict a user's next action by learning from their historical interaction sequences. In real-world applications, these models require periodic updates to adapt to new interactions and evolving user preferences. While incremental learning methods facilitate these updates, they face significant challenges. Replay-based approaches incur high memory and computational costs, and regularization-based methods often struggle to discard outdated or conflicting knowledge. To overcome these challenges, we propose SA-CAISR, a Stage-Adaptive and Conflict-Aware Incremental Sequential Recommendation framework. As a buffer-free framework, SA-CAISR operates using only the old model and new data, directly addressing the high costs of replay-based techniques. SA-CAISR introduces a novel Fisher-weighted knowledge-screening mechanism that dynamically identifies outdated knowledge by estimating parameter-level conflicts between the old model and new data, selectively removing obsolete knowledge while preserving compatible historical patterns. This dynamic balance between stability and adaptability allows our method to achieve state-of-the-art performance in incremental SR. Specifically, SA-CAISR improves Recall@20 by 2.0% on average across datasets, while reducing memory usage by 97.5% and training time by 46.9% compared to the best baseline. This efficiency allows real-world systems to rapidly update user profiles with minimal computational overhead, ensuring more timely and accurate recommendations.

cs.IR↗

arXiv2026

ASSESS: A Semantic and Structural Evaluation Framework for Statement Similarity

Despite significant strides in statement autoformalization, a critical gap remains in the development of automated evaluation metrics capable of assessing formal translation quality. Existing metrics often fail to balance semantic and structural information: string-based methods neglect semantics, whereas proof-based approaches offer no graded similarity when proofs fail. To address these issues, we introduce ASSESS (A Semantic and Structural Evaluation Framework for Statement Similarity), which captures syntactic structure by transforming formal statements into operator trees and computes a real-valued similarity score using our novel TransTED (Transformation Tree Edit Distance) Similarity metric by incorporating semantic transformations. For rigorous validation, we present EPLA (Evaluating Provability and Likeness for Autoformalization), a benchmark comprising 1,247 expert-annotated formal statement pairs derived from miniF2F and ProofNet, distinctively labeled for both semantic provability and structural likeness. Experiments on the EPLA benchmark demonstrate that TransTED Similarity surpasses existing methods, achieving state-of-the-art accuracy and Kappa score. The benchmark dataset, code, and detailed experimental results are available at https://github.com/XiaoyangLiu-sjtu/ASSESS.

cs.LG↗

arXiv2026

Punishment in bipartite societies

From ant-acacia mutualism to performative conflict resolution among Inuit, dedicated punishments between distinct subsets of a population are widespread and can reshape the evolutionary trajectory of cooperation. Existing studies have focused on punishments within a homogeneous population, paying little attention to cooperative dynamics in a situation where belonging to a subset is equally important to the actual strategy represented by an actor. To fill this gap, we here study a bipartite population where cooperator agents in a public goods game penalize exclusively those defectors who belong to the alternative subset. We find that cooperation can emerge and remain stable under symmetric intergroup punishment. In particular, at low punishment intensity and at a small value of the enhancement factor of the dilemma game, intergroup punishment promotes cooperation more effectively than a uniformly applied punishment. Moreover, intergroup punishment in bipartite populations tends to be more favorable for overall social welfare. When this incentive is balanced, cooperators can collectively restrain defectors of the alternative set via aggregate interactions in a randomly formed working group, offering a more effective incentive. Conversely, breaking the symmetry of intergroup punishment inhibits cooperation, as the imbalance creates an Achilles' heel in the enforcement structure. Our work, thus, reveals symmetry in intergroup punishment as a unifying principle behind cooperation across human and biological systems.

q-bio.PE↗

arXiv2026

Broadband silicon polarization beam splitter based on Floquet engineering

A broadband silicon polarization beam splitter (PBS) is proposed and experimentally demonstrated based on Floquet-engineered directional couplers. The total length of the coupling structure is 20 um . By periodically modulating the waveguide width of the directional couplers, the power exchange between the two waveguides for the transverse-electric (TE) mode is suppressed, whereas the power coupling for the transverse-magnetic (TM) mode is enhanced. The fabricated PBS exhibits polarization extinction ratios (PERs) > 20 dB for both polarizations over a broad wavelength range of 1483 nm-1620 nm. Additionally, the measured insertion losses (ILs) are 0.15 dB and 1.2 dB at 1550 nm for TE and TM polarizations, respectively.

physics.optics↗

arXiv2026

Blockchain Technology for Public Services: A Polycentric Governance Synthesis

National governments are increasingly adopting blockchain to enhance transparency, trust, and efficiency in public service delivery. However, evidence on how these technologies are governed across national contexts remains fragmented and overly focused on technical features. Using Polycentric Governance Theory, this study conducts a systematic review of peer-reviewed research published between 2021 and 2025 to examine blockchain-enabled public services and the institutional, organizational, and information-management factors shaping their adoption. Following PRISMA guidelines, we synthesize findings from major digital government and information systems databases to identify key application domains, including digital identity, electronic voting, procurement, and social services, and analyze the governance arrangements underpinning these initiatives. Our analysis reveals that blockchain adoption is embedded within polycentric environments characterized by distributed authority, inter-organizational coordination, and layered accountability. Rather than adopting full decentralization, governments typically utilize hybrid and permissioned designs that allow for selective decentralization alongside centralized oversight, a pattern we conceptualize as "controlled polycentricity." By reframing blockchain as a governance infrastructure that encodes rules for coordination and information-sharing, this study advances digital government theory beyond simple adoption metrics. The findings offer theoretically grounded insights for researchers and practical guidance for policymakers seeking to design and scale sustainable blockchain-enabled public services.

cs.CY↗

arXiv2026

Post-Disaster Resource Redistribution and Cooperation Evolution Based on Two-Layer Network Evolutionary Games

In the aftermath of large-scale disasters, the scarcity of resources and the paralysis of infrastructure raise severe challenges to effective post-disaster recovery. Efficient coordination between shelters and victims plays a crucial role in building community resilience, yet the evolution of two-layer behavioral feedback between these two groups through network coupling remains insufficiently understood. Here, this study develops a two-layer network to capture the cross-layer coupling between shelters and victims. The upper layer uses a post-disaster emergency resource redistribution model within the framework of the public goods game, while the lower layer adopts a cooperative evolutionary game to describe internal victim interactions. Monte Carlo simulations on scale-free networks reveal threshold effects of incentives: moderate public goods enhancement and subsidies promote cooperation, whereas excessive incentives induce free-riding. In contrast, credible and well-executed punishment effectively suppresses defection. Targeted punishment of highly connected shelters significantly enhances cooperation under resource constraints. A comparative analysis using a network generated from the actual coordinates of Beijing shelters confirms the model's generality and practical applicability. The findings highlight the importance of calibrated incentives, enforceable sanctions, and structural targeting in fostering robust cooperation across organizational and individual levels in post-disaster environments.

physics.soc-ph↗