arXiv subjects

Kai Ma

Publications and source records attributed to Kai Ma.

At least 19 recordsLinked to original sources

arXiv2026

Entwine: Coordinating Tiled Computation and Fine-Grained Communication across GPUs

Modern high-performance GPU computations partition tensors into tiles to exploit data reuse and parallelism. Individual tile computations complete earlier than the full tensor computation, creating opportunities to overlap computation and communication. However, a mismatch between computation and communication progress can limit these opportunities. Communication stalls when no data is ready, and may lag when data arrives in bursts. Communication can also slow computation by consuming shared resources, offsetting the benefits of overlap. We present Entwine, which coordinates tile computation order, fine-grained communication, and SM resource allocation to minimize overall completion time. Entwine reorders tile computation to produce data for communication at a more regular pace. Entwine couples this schedule with fine-grained SM-based communication to process tile results with low latency and low overhead. Since the communication kernel also consumes SM resources, Entwine coordinates their allocation to balance communication progress against computation slowdown. Across representative tensor-parallel LLM workloads, Entwine achieves a geomean speedup of 1.232x (up to 1.433x) over cuBLAS+NCCL, and outperforms state-of-the-art overlap baselines by 3.1-9.8% in geomean. We will open-source our implementation upon publication.

cs.DC↗

arXiv2026

Exploring millicharged particles in laboratory and astrophysical strong-field regimes

The probe of light dark particles beyond the Standard Model (SM) under a strong-field environment has drawn significant attention. In this work, we investigate the potential to search for and constrain light millicharged particles (MCPs) via strong electromagnetic fields in both laboratory laser experiments and astrophysical environments such as magnetars. We propose the MCP pair production from nonlinear Compton scattering through the interaction of a relativistic electron beam with a high-intensity laser pulse. The Furry picture and Volkov solution of Dirac equation in a background electromagnetic field are used to describe the electrons and MCPs under an external classical laser field. We calculate the cross sections of nonlinear Compton scattering to MCP pairs and take into account the irreducible SM background with missing neutrinos. We also revisit the MCP pair production via the Schwinger mechanism from magnetars with ultra-strong magnetic field and parallel electric field in polar gap. The energy loss due to the Schwinger pair production of MCPs and electric field acceleration is evaluated based on Ruderman-Sutherland model for confirmed magnetars. We find that the constraints from highly magnetized magnetars and the search potential in laboratory laser experiments are complementary.

hep-ph↗

arXiv2026

TORL-VLA: Tactile Guided Online Reinforcement Learning for Contact-Rich Manipulation

Vision-Language-Action (VLA) models have become a powerful framework for robotic manipulation, and recent studies have introduced tactile or force feedback into VLAs to address contact-rich tasks. However, these models are typically deployed as offline policies. When contact conditions shift from the training distribution, the policy cannot perform online adaptation, leading to problems such as inappropriate contact forces and inefficient retries. Therefore, we propose TORL-VLA, a tactile-guided online reinforcement learning framework that couples tactile feedback with policy refinement for contact-rich manipulation. Our method introduces a tactile-derived wrench-aware VLA to predict reference actions and future wrench sequences, while a lightweight online RL module is used to refine the reference actions. To stabilize learning from mixed exploratory policy-generated and human-intervention data, we introduce an intervention-censored critic that prevents post-intervention success from being wrongly credited to policy-generated actions preceding intervention. Real-robot experiments on long-horizon contact-rich tasks, including latch manipulation, coffee-cup placement, and egg handling, show that TORL-VLA improves success rates at both subtask and full-task levels, as well as time-bounded execution efficiency over strong baselines. Project page: https://torl-vla.github.io/

cs.RO↗

arXiv2026

RDT2: Exploring the Scaling Limit of UMI Data Towards Zero-Shot Cross-Embodiment Generalization

Vision-Language-Action (VLA) models hold promise for generalist robotics but currently struggle with data scarcity, architectural inefficiencies, and the inability to generalize across different hardware platforms. We introduce RDT2, a robotic foundation model built upon a 7B parameter VLM designed to enable zero-shot deployment on novel embodiments for open-vocabulary tasks. To achieve this, we collected one of the largest open-source robotic datasets--over 10,000 hours of demonstrations in diverse families--using an enhanced, embodiment-agnostic Universal Manipulation Interface (UMI). Our approach employs a novel three-stage training recipe that aligns discrete linguistic knowledge with continuous control via Residual Vector Quantization (RVQ), flow-matching, and distillation for real-time inference. Consequently, RDT2 becomes one of the first models that simultaneously zero-shot generalizes to unseen objects, scenes, instructions, and even robotic platforms. Besides, it outperforms state-of-the-art baselines in dexterous, long-horizon, and dynamic downstream tasks like playing table tennis. See https://rdt-robotics.github.io/rdt2/ for more information.

cs.RO↗

arXiv2025

Fragmentation of neutron-rich carbon isotopes on light targets at 27.5 MeV/nucleon

Experimental and theoretical investigation of the fragmentation reaction in Fermi-energy domain is currently of particular importance for not only the nuclear physics but also some interdisciplinary fields. In the present work, neutron-rich $^{14}$C and $^{16}$C ion beams at 27.5 MeV/nucleon were used to bombard carbon and polyethylene (CD$_{2}$)$_{n}$ targets. Energy and angular distributions of the produced fragments were measured. Background events originating from the carbon content in (CD$_{2}$)$_{n}$ target were efficiently excluded using an extended $E-P$ plot method. Experimental results are systematically analyzed by using HIPSE-SIMON dynamic model. The comparison reveals that, for the carbon target, the HIPSE-SIMON calculation overestimates the yields of the beam-velocity component for fragments near the projectile and also the energy phase space for fragments far away from the projectile, suggesting fine tuning of the overall interaction profile adopted in the model. In contrast, for reactions with the deuteron target, the model calculation can reasonably reproduce the experimental data. The implication of the fragmentation mechanism to the validity of the invariant mass method, as frequently used to reconstruct the clustering resonant structures in light nuclei, is also discussed.

nucl-ex↗

arXiv2025

Universal Dexterous Functional Grasping via Demonstration-Editing Reinforcement Learning

Reinforcement learning (RL) has achieved great success in dexterous grasping, significantly improving grasp performance and generalization from simulation to the real world. However, fine-grained functional grasping, which is essential for downstream manipulation tasks, remains underexplored and faces several challenges: the complexity of specifying goals and reward functions for functional grasps across diverse objects, the difficulty of multi-task RL exploration, and the challenge of sim-to-real transfer. In this work, we propose DemoFunGrasp for universal dexterous functional grasping. We factorize functional grasping conditions into two complementary components - grasping style and affordance - and integrate them into an RL framework that can learn to grasp any object with any functional grasping condition. To address the multi-task optimization challenge, we leverage a single grasping demonstration and reformulate the RL problem as one-step demonstration editing, substantially enhancing sample efficiency and performance. Experimental results in both simulation and the real world show that DemoFunGrasp generalizes to unseen combinations of objects, affordances, and grasping styles, outperforming baselines in both success rate and functional grasping accuracy. In addition to strong sim-to-real capability, by incorporating a vision-language model (VLM) for planning, our system achieves autonomous instruction-following grasp execution.

cs.RO↗

arXiv2025

SteadyDancer: Harmonized and Coherent Human Image Animation with First-Frame Preservation

Preserving first-frame identity while ensuring precise motion control is a fundamental challenge in human image animation. The Image-to-Motion Binding process of the dominant Reference-to-Video (R2V) paradigm overlooks critical spatio-temporal misalignments common in real-world applications, leading to failures such as identity drift and visual artifacts. We introduce SteadyDancer, an Image-to-Video (I2V) paradigm-based framework that achieves harmonized and coherent animation and is the first to ensure first-frame preservation robustly. Firstly, we propose a Condition-Reconciliation Mechanism to harmonize the two conflicting conditions, enabling precise control without sacrificing fidelity. Secondly, we design Synergistic Pose Modulation Modules to generate an adaptive and coherent pose representation that is highly compatible with the reference image. Finally, we employ a Staged Decoupled-Objective Training Pipeline that hierarchically optimizes the model for motion fidelity, visual quality, and temporal coherence. Experiments demonstrate that SteadyDancer achieves state-of-the-art performance in both appearance fidelity and motion control, while requiring significantly fewer training resources than comparable methods. The model has been publicly released at \url{https://mcg-nju.github.io/steadydancer-web}.

cs.CV↗

arXiv2025

LMM-IR: Large-Scale Netlist-Aware Multimodal Framework for Static IR-Drop Prediction

Static IR drop analysis is a fundamental and critical task in the field of chip design. Nevertheless, this process can be quite time-consuming, potentially requiring several hours. Moreover, addressing IR drop violations frequently demands iterative analysis, thereby causing the computational burden. Therefore, fast and accurate IR drop prediction is vital for reducing the overall time invested in chip design. In this paper, we firstly propose a novel multimodal approach that efficiently processes SPICE files through large-scale netlist transformer (LNT). Our key innovation is representing and processing netlist topology as 3D point cloud representations, enabling efficient handling of netlist with up to hundreds of thousands to millions nodes. All types of data, including netlist files and image data, are encoded into latent space as features and fed into the model for static voltage drop prediction. This enables the integration of data from multiple modalities for complementary predictions. Experimental results demonstrate that our proposed algorithm can achieve the best F1 score and the lowest MAE among the winning teams of the ICCAD 2023 contest and the state-of-the-art algorithms.

cs.LG↗

arXiv2025

Production of Light Dark Particles from Nonlinear Compton Scattering Between Intense Laser and Muon or Proton Beam

The laser of an intense electromagnetic field promotes the studies of strong-field particle physics in high-intensity frontier. Particle accelerator facilities in the world produce high-quality muon and proton beams. In this work, we propose the nonlinear Compton scattering to light dark particles through the collision between intense laser pulse and muon or proton beam. We take light dark photon and axion-like particle as illustrative dark particles. The cross sections of relevant nonlinear Compton scattering to dark photon or axion-like particle are calculated. We also analyze the background processes with missing neutrinos. The prospective sensitivity shows that the laser-induced process provides a complementary and competitive search of new invisible particles lighter than about 1 MeV.

hep-ph↗

arXiv2025

SAM 2++: Tracking Anything at Any Granularity

Due to the varying granularity of target states across different tasks, most existing trackers are tailored to a single task, which specificity limits their generalization, preventing them from effectively utilizing multi-task training data and leading to redundancy in both model design and parameters. Although recent unified vision models share partial architectures across tasks, they usually retain task-specific interfaces and overlook the common tracking principle behind different granularities, leaving a gap for truly unified video tracking. To unify video tracking tasks, we present SAM 2++, a unified framework that can handle target states at different granularities, including masks, boxes, and points, through an integrated design of prompt encoding, output decoding, and memory representation. First, to handle different target granularities, we design task-specific prompts that map diverse task inputs into general prompt embeddings, together with a Unified Decoder that produces task results in a common output form without redesigning the overall pipeline. Next, to satisfy memory matching, the core operation of tracking, we introduce a task-adaptive memory mechanism that unifies memory across different granularities while preserving their distinct state semantics, preventing full parameter sharing from causing interference across granularities. Finally, we introduce Tracking-Any-Granularity, the first large and diverse video tracking dataset with rich annotations at three granularities. It is constructed through a customized data engine with phased manual annotation and model-assisted completion, providing a comprehensive resource for training, benchmarking, and analyzing unified tracking models. Comprehensive experiments confirm that SAM 2++ sets a new state of the art across diverse tracking tasks at different granularities, establishing a unified and robust tracking framework.

cs.CV↗

arXiv2025

Prospects for toponium formation at the LHC in the single-lepton mode

We investigate the formation of toponium in the single-leptonic final state at the LHC. Our study builds on our recently proposed framework that incorporates the associated non-perturbative effects into Monte Carlo simulations through the Green's function of the non-relativistic QCD Hamiltonian and the re-weighting of hard-scattering matrix elements. This allows us to perform a phenomenological analysis that demonstrates that a statistically significant excess from toponium formation could already be accessible in Run~2 data. Moreover, our results highlight observables that provide handles for signal characterisation and establish the single-leptonic channel as a competitive and complementary avenue for the ongoing exploration of toponium signatures at colliders.

hep-ph↗

arXiv2025

Laser-assisted Light-by-Light Scattering in Born-Infeld and Axion-like Particle Theories

The precision measurements of well-known light-by-light reactions lead to important insights of nonlinear quantum electrodynamics (QED) vacuum polarization. The laser of an intense electromagnetic field strength provides an essential tool for exploring nonlinear QED and new physics beyond Standard Model (SM) in the high-precision frontier. In this work, we propose to search for low-energy light-by-light scattering in the collision of a photon beam and a laser pulse of classical background field. We aim to investigate the impact of Born-Infeld (BI) and axion-like particle (ALP) theories on laser-assisted light-by-light scattering. We calculate the QED light-by-light scattering cross section using complete QED helicity amplitudes, and then combine them with the amplitudes in BI or ALP theory to evaluate the total cross section. The laser-assisted SM light-by-light scattering should be observable in future experiments with very moderate integrated luminosities. The sensitivity of laser-assisted light-by-light scattering to BI and ALP parameters is presented.

hep-ph↗

arXiv2025

Agentic Reinforced Policy Optimization

Large-scale reinforcement learning with verifiable rewards (RLVR) has demonstrated its effectiveness in harnessing the potential of large language models (LLMs) for single-turn reasoning tasks. In realistic reasoning scenarios, LLMs can often utilize external tools to assist in task-solving processes. However, current RL algorithms inadequately balance the models' intrinsic long-horizon reasoning capabilities and their proficiency in multi-turn tool interactions. To bridge this gap, we propose Agentic Reinforced Policy Optimization (ARPO), a novel agentic RL algorithm tailored for training multi-turn LLM-based agents. Through preliminary experiments, we observe that LLMs tend to exhibit highly uncertain behavior, characterized by an increase in the entropy distribution of generated tokens, immediately following interactions with external tools. Motivated by this observation, ARPO incorporates an entropy-based adaptive rollout mechanism, dynamically balancing global trajectory sampling and step-level sampling, thereby promoting exploration at steps with high uncertainty after tool usage. By integrating an advantage attribution estimation, ARPO enables LLMs to internalize advantage differences in stepwise tool-use interactions. Our experiments across 13 challenging benchmarks in computational reasoning, knowledge reasoning, and deep search domains demonstrate ARPO's superiority over trajectory-level RL algorithms. Remarkably, ARPO achieves improved performance using only half of the tool-use budget required by existing methods, offering a scalable solution for aligning LLM-based agents with real-time dynamic environments. Our code and datasets are released at https://github.com/dongguanting/ARPO

cs.LG↗

arXiv2025

PositionIC: Unified Position and Identity Consistency for Image Customization

Recent subject-driven image customization excels in fidelity, yet fine-grained instance-level spatial control remains an elusive challenge, hindering real-world applications. This limitation stems from two factors: a scarcity of scalable, position-annotated datasets, and the entanglement of identity and layout by global attention mechanisms. To this end, we introduce PositionIC, a unified framework for high-fidelity, spatially controllable multi-subject customization. First, we present BMPDS, the first automatic data-synthesis pipeline for position-annotated multi-subject datasets, effectively providing crucial spatial supervision. Second, we design a lightweight, layout-aware diffusion framework that integrates a novel visibility-aware attention mechanism. This mechanism explicitly models spatial relationships via an NeRF-inspired volumetric weight regulation to effectively decouple instance-level spatial embeddings from semantic identity features, enabling precise, occlusion-aware placement of multiple subjects. Extensive experiments demonstrate PositionIC achieves state-of-the-art performance on public benchmarks, setting new records for spatial precision and identity consistency. Our work represents a significant step towards truly controllable, high-fidelity image customization in multi-entity scenarios. Code and data: https://github.com/MeiGen-AI/PositionIC.

cs.CV↗

arXiv2025

Deep reinforcement learning-based joint real-time energy scheduling for green buildings with heterogeneous battery energy storage devices

Green buildings (GBs) with renewable energy and building energy management systems (BEMS) enable efficient energy use and support sustainable development. Electric vehicles (EVs), as flexible storage resources, enhance system flexibility when integrated with stationary energy storage systems (ESS) for real-time scheduling. However, differing degradation and operational characteristics of ESS and EVs complicate scheduling strategies. This paper proposes a model-free deep reinforcement learning (DRL) method for joint real-time scheduling based on a combined battery system (CBS) integrating ESS and EVs. We develop accurate degradation models and cost estimates, prioritize EV travel demands, and enable collaborative ESS-EV operation under varying conditions. A prediction model optimizes energy interaction between CBS and BEMS. To address heterogeneous states, action coupling, and learning efficiency, the DRL algorithm incorporates double networks, a dueling mechanism, and prioritized experience replay. Experiments show a 37.94 percent to 40.01 percent reduction in operating costs compared to a mixed-integer linear programming (MILP) approach.

eess.SY↗

arXiv2025

Cybersquatting in Web3: The Case of NFT

Cybersquatting refers to the practice where attackers register a domain name similar to a legitimate one to confuse users for illegal gains. With the growth of the Non-Fungible Token (NFT) ecosystem, there are indications that cybersquatting tactics have evolved from targeting domain names to NFTs. This paper presents the first in-depth measurement study of NFT cybersquatting. By analyzing over 220K NFT collections with over 150M NFT tokens, we have identified 8,019 cybersquatting NFT collections targeting 654 popular NFT projects. Through systematic analysis, we discover and characterize seven distinct squatting tactics employed by scammers. We further conduct a comprehensive measurement study of these cybersquatting NFT collections, examining their metadata, associated digital asset content, and social media status. Our analysis reveals that these NFT cybersquatting activities have resulted in a significant financial impact, with over 670K victims affected by these scams, leading to a total financial exploitation of $59.26 million. Our findings demonstrate the urgency to identify and prevent NFT squatting abuses.

cs.CR↗

arXiv2025

Mono-lepton Signature of a Neutrino-philic Dark Fermion at Hadron Colliders

Searching for dark matter at high-energy colliders and direct detection experiments can effectively cover nearly the entire mass range from the MeV to the TeV scale. In this paper, we focus on four-fermion contact interactions formulated within the framework of Effective Field Theory. Specifically, we present a detailed analysis of mono-lepton production at the LHC. Our results demonstrate that tensor operators exhibit superior sensitivity in the mono-lepton channel, constraining energy scales up to 3\,TeV for a nearly massless dark fermion using current LHC data. Moreover, these operators mediate both spin-independent and spin-dependent absorption processes in nuclear targets. A systematic comparison of constraints between direct detection experiments and collider measurements reveals the LHC's distinct advantage in exploring sub-GeV dark matter candidates while maintaining competitive sensitivity at the TeV scale. Notably, direct detection experiments such as Super-Kamiokande and Borexino achieve complementary constraints in the 10-100\,TeV mass range through their unique capabilities: utilization of light nuclei targets, large exposure volumes, and distinctive features of the recoil energy spectra.

hep-ph↗

arXiv2025

VRoPE: Rotary Position Embedding for Video Large Language Models

Rotary Position Embedding (RoPE) has shown strong performance in text-based Large Language Models (LLMs), but extending it to video remains a challenge due to the intricate spatiotemporal structure of video frames. Existing adaptations, such as RoPE-3D, attempt to encode spatial and temporal dimensions separately but suffer from two major limitations: positional bias in attention distribution and disruptions in video-text transitions. To overcome these issues, we propose Video Rotary Position Embedding (VRoPE), a novel positional encoding method tailored for Video-LLMs. Specifically, we introduce a more balanced encoding strategy that mitigates attention biases, ensuring a more uniform distribution of spatial focus. Additionally, our approach restructures positional indices to ensure a smooth transition between video and text tokens. Extensive experiments on different models demonstrate that VRoPE consistently outperforms previous RoPE variants, achieving significant improvements in video understanding, temporal reasoning, and retrieval tasks. Code is available at https://github.com/johncaged/VRoPE.

cs.AI↗