arXiv subjects

Jiajun Chen

Publications and source records attributed to Jiajun Chen.

At least 19 recordsLinked to original sources

arXiv2026

Taming the Implicit: Dual-Channel Risk-Aware Reinforcement Fine-Tuning for Continual Multimodal Post-Training

Reinforcement fine-tuning (RFT) is widely believed to inherently resist catastrophic forgetting in continual post-training of multimodal large language models. Under pronounced task distributional shifts, however, forgetting across representative RFT algorithms escalates sharply. This stems from the implicit reward-variance regularization inherent to RFT, which proves incapable of suppressing uncontrolled optimization risk. We propose Risk-Aware Policy Optimization (RAPO), the first dual-channel framework for explicit risk governance in continual RFT. On the policy channel, Risk-Aware Policy Scaling adaptively calibrates per-sample update magnitude via rollout reliability and Fisher-inspired local predictive sensitivity; on the data channel, Risk-Aware Dynamic Bucket Sampling reorganizes training batches through dynamic risk stratification, steering optimization toward informative yet stable samples. As a plug-and-play strategy requiring no cross-task memory, RAPO generalizes to any RFT algorithm without modification. On the public MLLM-CL benchmark, RAPO reduces final forgetting by 79.8% relative to its RLOO backbone while retaining new-task competitiveness.

cs.AI↗

arXiv2026

Core-Halo Mass Relation in Cosmological Vector Dark Matter

We study the cosmological core-halo relation in vector dark matter using three-component Schr\"odinger-Poisson simulations. Starting from cosmological vector-field initial conditions, which due to the evolution of the vector field during inflation are enhanced on small scales, we find that nonlinear evolution begins almost immediately following matter-radiation equality and produces compact self-gravitating Proca-star condensates at the centers of halos. After confirming the central condensates through their radial density profiles, we find the empirical relation $\widetilde M_\star\propto \widetilde M_{\rm h}^{0.6403}$ between the Proca star mass and halo mass, although interestingly we find that we are only able to confirm Proca stars in $\mathcal{O}(10\%)$ of halos. This serves as important input for future studies of the abundance and merger rates of Proca stars in models of vector dark matter. We also examine the vector-field structure of the objects through global longitudinal and transverse polarization fractions and local spin density inside halos, which increases over cosmic time.

astro-ph.CO↗

arXiv2026

Spin Polarization of Proca Stars Formed by Gravitational Bose--Einstein Condensation

We study the internal spin polarization of Proca stars formed by gravitational Bose--Einstein condensation of a three-component nonrelativistic vector field. In idealized periodic-box simulations, we decompose the aperture-averaged spin into a coherent net fraction, a local polarization fraction, and their ratio, thereby distinguishing genuine coherent core polarization from local spin density whose direction cancels inside the aperture. For independent vector components, condensation produces Proca stars that are sizably but not maximally polarized. Across an independent-component simulation ensemble, the coherent core-spin fraction has mean $\langle\chi_{\rm net}\rangle\simeq0.62$, with substantial realization-to-realization scatter. We interpret this scatter as the outcome of random elliptical polarization of the dominant component-space mode, rather than as evidence for a universal Proca-star spin fraction. This interpretation is supported by the core polarization matrix: its leading eigenvector provides an estimate of the ideal single-mode spin fraction, while the difference between this estimate and the directly integrated coherent spin tracks the departure of the core from a rank-one component-space state. The measured leading-eigenvector spin fractions are broadly compatible with an isotropic random-complex-vector model and less compatible with an equal-amplitude random-phase model. Correlated and circular initial data drive the dominant component-space mode toward the circular-polarization bound, giving the ordering independent $\rightarrow$ correlated $\rightarrow$ circular. These results show that internal polarization is a genuine vector degree of freedom of gravitationally condensed nonrelativistic Proca stars, and that the resulting core spin is controlled by the polarization of the dominant condensed mode rather than by a fixed universal value.

hep-ph↗

arXiv2026

Rethinking Scientific Discovery in the Agentic Era

Artificial intelligence has advanced scientific discovery, but most AI4Science systems remain fragmented tools that rely on humans to coordinate problem formulation, literature grounding, model use, simulation, validation, and knowledge reuse. This paper presents \textbf{SCION (Scientific Collaborative Innovation with Agentic Organizational Nexus)}, an agentic scientific operating system that acts as an \textbf{organizational nexus}. Through a Science Agent serving as a \textbf{Meta-Harness}, SCION connects scientific tasks, tools, agents, artifacts, and memory, transforming research into an executable, auditable, and reusable operational process. At its core is the \textbf{Research Execution Plan (REP)}, which compiles high-level scientific intent into staged objectives, dependencies, verification checkpoints, tool requirements, expected artifacts, and fallback conditions. SCION further integrates hierarchical multi-agent execution, profile-driven specialization, selective context construction, governed delegation, and layered epistemic memory to support long-horizon scientific work. We formulate discovery under SCION as \textbf{Target-conditioned Inverse Search} and extend it to hidden-target settings through batch active search under finite experimental budgets. Applications in materials analysis, molecule design, and protein or antibody screening, together with experiments on scientific reading, idea generation, molecule generation, and antibody screening, show that SCION outperforms existing autonomous research-agent baselines, especially in decomposition, verification, refinement, and memory reuse. Overall, SCION shifts AI from isolated tools toward a coordinated operational layer for traceable and reusable scientific innovation.

cs.CL↗

arXiv2026

Yukawa-Screened Bose-Star Condensation

We study Bose-star formation in a Yukawa-Schr\"odinger-Poisson (YSP) system. A finite interaction range suppresses the infrared kinetic relaxation responsible for Bose-star condensation, modifying both the equilibrium Bose-star structure and the condensation timescale. We derive a screened kinetic condensation formula in which the ordinary gravitational Coulomb logarithm is replaced by a finite Yukawa transport logarithm. Static YSP solutions show that Yukawa screening broadens the Bose-star density profile relative to the ordinary Newtonian soliton. Fully dynamical pseudospectral simulations with homogeneous and isotropic initial conditions demonstrate that Yukawa screening systematically delays Bose-star condensation, in good agreement with the screened kinetic prediction after fitting a single overall normalization parameter.

hep-ph↗

arXiv2026

MuKV: Multi-Grained KV Cache Compression for Long Streaming Video Question-Answering

Long streaming video QA remains challenging due to growing visual tokens and limited reasoning length of large language models (LLMs). KV-caching stores the Key-Value (KV) of the historical tokens via LLM prefill and enables more efficient streaming QA. However, existing methods cache every one or two frames, causing redundant memory usage and losing fine-grained spatial details within frame or temporal contexts across frames. This paper proposes MuKV, a method that features a multi-grained KV cache compression module and a semi-hierarchical retrieval approach to improve both efficiency and accuracy for long streaming VideoQA. For the offline KV cache, MuKV extracts visual representations at patch-, frame-, and segment-levels. The multiple levels of granularity preserve both local cues and global temporal context, while maintaining efficiency with a dual signal token compression mechanism guided by self-attention and frequency. For online QA, MuKV designs a semi-hierarchical retrieval method to retrieve relevant KV caches for answer generation. Experiments on long-streaming VideoQA benchmarks show that MuKV significantly improves answer accuracy, without sacrificing memory and online QA efficiency. Moreover, our compression mechanism alone brings consistent benefits across answer accuracy, memory, and QA efficiency over baselines, showcasing highly effective contribution.

cs.CV↗

arXiv2026

A Data-Efficient Path to Multilingual LLMs: Language Expansion via Post-training PARAM$\Delta$ Integration into Upcycled MoE

Expanding Large Language Models~(LLMs) to new languages is a costly endeavor, demanding extensive Continued Pre-Training~(CPT) and data-intensive alignment. While recent data-free merging techniques attempt to bypass alignment by fusing a multilingual CPT-enhanced model with its instruct counterpart, they are plagued by a critical trade-off: mitigating parameter conflicts to preserve original abilities inevitably dilutes new language acquisition, and vice-versa. To resolve this conflict, we introduce \method, which upcycles a dense model into a Mixture-of-Experts~(MoE) architecture, allocating different experts to different languages. Alignment ability is then transferred by grafting a MoE-expanded parameter delta~($\Delta_{\text{post}}$) to the CPT-enhanced base model, bypassing the complex alignment phase. Experiments demonstrate \method's superiority even against baselines with similar FLOPs or number of parameters; it improves performance on expanded languages while effectively preserving original capabilities. We further show our approach is highly applicable across different models and Post-training deltas.

cs.CL↗

arXiv2026

ClawMark: A Living-World Benchmark for Multi-Turn, Multi-Day, Multimodal Coworker Agents

Language-model agents are increasingly used as persistent coworkers that assist users across multiple working days. During such workflows, the surrounding environment may change independently of the agent: new emails arrive, calendar entries shift, knowledge-base records are updated, and evidence appears across images, scanned PDFs, audio, video, and spreadsheets. Existing benchmarks do not adequately evaluate this setting because they typically run within a single static episode and remain largely text-centric. We introduce \bench{}, a benchmark for coworker agents built around multi-turn multi-day tasks, a stateful sandboxed service environment whose state evolves between turns, and rule-based verification. The current release contains 100 tasks across 13 professional scenarios, executed against five stateful sandboxed services (filesystem, email, calendar, knowledge base, spreadsheet) and scored by 1537 deterministic Python checkers over post-execution service state; no LLM-as-judge is invoked during scoring. We benchmark seven frontier agent systems. The strongest model reaches 75.8 weighted score, but the best strict Task Success is only 20.0\%, indicating that partial progress is common while complete end-to-end workflow completion remains rare. Turn-level analysis shows that performance drops after the first exogenous environment update, highlighting adaptation to changing state as a key open challenge. We release the benchmark, evaluation harness, and construction pipeline to support reproducible coworker-agent evaluation.

cs.CV↗

arXiv2026

OpenMobile: Building Open Mobile Agents with Task and Trajectory Synthesis

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

cs.AI↗

arXiv2026

TAPO: Translation Augmented Policy Optimization for Multilingual Mathematical Reasoning

Large Language Models (LLMs) have demonstrated remarkable proficiency in English mathematical reasoning, yet a significant performance disparity persists in multilingual contexts, largely attributed to deficiencies in language understanding. To bridge this gap, we introduce Translation-Augmented Policy Optimization (TAPO), a novel reinforcement learning framework built upon GRPO. TAPO enforces an explicit alignment strategy where the model leverages English as a pivot and follows an understand-then-reason paradigm. Crucially, we employ a step-level relative advantage mechanism that decouples understanding from reasoning, allowing the integration of translation quality rewards without introducing optimization conflicts. Extensive experiments reveal that TAPO effectively synergizes language understanding with reasoning capabilities and is compatible with various models. It outperforms baseline methods in both multilingual mathematical reasoning and translation tasks, while generalizing well to unseen languages and out-of-domain tasks.

cs.CL↗

arXiv2026

Ostwald's Rule of Stages in One-Dimension

Ostwald's Rule of Stages, which is one of the most widely observed phenomena associated with crystallization of polymorphs, follows naturally from the thermodynamics of nucleation. However, most observations of its manifestations have been limited to three-dimensional crystals and its validity in one-dimension, where no nucleation barrier exists, remains unclear. Here we investigate the two-dimensional assemblies and phase transformation mechanisms of a peptide that forms two distinct phases on graphite via one-dimensional nucleation using in situ atomic force microscopy. We find that the evolution of phases illustrates Ostwald's Rule, but does so for purely kinetic reasons, and that the stable phase replaces the metastable via a dissolution-reprecipitation mechanism enabled by inherent fluctuations of the phase boundary. The findings provide general insights into the growth and transformation mechanisms of coexisting two-dimensional phases and thus delineate a strategy for capturing transient two-dimensional structures.

cond-mat.soft↗

arXiv2026

Quantum Confocal Microscopy in Fock Space with a 19 dB Metrological Gain

Quantum metrology promises measurement precision beyond classical limits by exploiting large-scale quantum states, yet realizing this advantage faces two fundamental challenges: the deterministic preparation of non-trivial quantum probes and the efficient extraction of metrological information in high-dimensional Hilbert spaces. Here, we introduce quantum confocal microscopy in Fock space that simultaneously resolves both challenges. Drawing a direct analogy between classical wave optics and quantum state evolution in a bosonic mode, we construct a confocal system with two Fock-space lenses. The first lens deterministically focuses a coherent state into a quantum probe with a tightly concentrated photon-number distribution, while the second lens maps the metrological information back to the vacuum state for efficient readout. Using a superconducting circuit QED platform, we prepare focused probe states with mean photon numbers up to ${N} = 500$, achieving a 21.5$\pm$1.1 dB compression of the photon-number uncertainty relative to a coherent state, with a scalable quantum circuit of $\mathcal{O}(1)$ operational depth. We demonstrate a displacement sensitivity scaling as $N^{-0.416}$, approaching the Heisenberg scaling ($N^{-0.5}$), and achieve a record metrological gain of 19.06$\pm$0.13 dB beyond the standard quantum limit. This work establishes quantum confocal microscopy as a scalable and practical framework for quantum-enhanced precision measurement, readily extendable to other bosonic platforms and high-dimensional quantum many-body systems.

quant-ph↗

arXiv2026

ExpLang: Improved Exploration and Exploitation in LLM Reasoning with On-Policy Thinking Language Selection

Current large reasoning models (LRMs) have shown strong ability on challenging tasks after reinforcement learning (RL) based post-training. However, previous work mainly focuses on English reasoning in expectation of the strongest performance, despite the demonstrated potential advantage of multilingual thinking, as well as the requirement for native thinking traces by global users. In this paper, we propose ExpLang, a novel LLM post-training pipeline that enables on-policy thinking language selection to improve exploration and exploitation during RL with the use of multiple languages. The results show that our method steadily outperforms English-only training with the same training budget, while showing high thinking language compliance for both seen and unseen languages. Analysis shows that, by enabling on-policy thinking language selection as an action during RL, ExpLang effectively extends the RL exploration space with diversified language preference and improves the RL exploitation outcome with leveraged non-English advantage. The method is orthogonal to most RL algorithms and opens up a new perspective on using multilinguality to improve LRMs.

cs.CL↗

arXiv2026

Value Alignment Tax: Measuring Value Trade-offs in LLM Alignment

Existing work on value alignment typically characterizes value relations statically, ignoring how alignment interventions, such as prompting, fine-tuning, or preference optimization, reshape the broader value system. In practice, aligning a target value can implicitly shift other values, creating value trade-offs that remain largely unmeasured. We introduce VAT, a framework that quantifies value trade-offs by measuring how alignment-induced changes propagate across interconnected values relative to achieved on-target gain. VAT captures the system-level dynamics of value expression under alignment intervention, enabling evaluation of both intended improvements and unintended side effects. Using a controlled scenario-action dataset grounded in Schwartz value theory, we collect paired pre-post normative judgments and analyze alignment effects across models, values, and interventions. Results show that alignment often produces uneven and structured co-movement among values, revealing systematic trade-offs between target and non-target values. These effects are largely invisible under conventional target-only evaluation, but become evident via VAT, highlighting process-level alignment risks and offering new insights into the dynamic nature of value alignment in LLMs. Dataset and code are open-sourced.

cs.AI↗

arXiv2026

E2PL: Effective and Efficient Prompt Learning for Incomplete Multi-view Multi-Label Class Incremental Learning

Multi-view multi-label classification (MvMLC) is indispensable for modern web applications aggregating information from diverse sources. However, real-world web-scale settings are rife with missing views and continuously emerging classes, which pose significant obstacles to robust learning. Prevailing methods are ill-equipped for this reality, as they either lack adaptability to new classes or incur exponential parameter growth when handling all possible missing-view patterns, severely limiting their scalability in web environments. To systematically address this gap, we formally introduce a novel task, termed \emph{incomplete multi-view multi-label class incremental learning} (IMvMLCIL), which requires models to simultaneously address heterogeneous missing views and dynamic class expansion. To tackle this task, we propose \textsf{E2PL}, an Effective and Efficient Prompt Learning framework for IMvMLCIL. \textsf{E2PL} unifies two novel prompt designs: \emph{task-tailored prompts} for class-incremental adaptation and \emph{missing-aware prompts} for the flexible integration of arbitrary view-missing scenarios. To fundamentally address the exponential parameter explosion inherent in missing-aware prompts, we devise an \emph{efficient prototype tensorization} module, which leverages atomic tensor decomposition to elegantly reduce the prompt parameter complexity from exponential to linear w.r.t. the number of views. We further incorporate a \emph{dynamic contrastive learning} strategy explicitly model the complex dependencies among diverse missing-view patterns, thus enhancing the model's robustness. Extensive experiments on three benchmarks demonstrate that \textsf{E2PL} consistently outperforms state-of-the-art methods in both effectiveness and efficiency. The codes and datasets are available at https://anonymous.4open.science/r/code-for-E2PL.

cs.LG↗

arXiv2026

DeTracker: Motion-decoupled Vehicle Detection and Tracking in Unstabilized Satellite Videos

Satellite videos provide continuous observations of surface dynamics but pose significant challenges for multi-object tracking (MOT), especially under unstabilized conditions where platform jitter and the weak appearance of tiny objects jointly degrade tracking performance. To address this problem, we propose DeTracker, a joint-detection-and-tracking framework tailored for unstabilized satellite videos. DeTracker introduces a task-driven Global-Local Motion Decoupling (GLMD) module to address the motion imbalance between dominant platform motion and weak target motion. It suppresses background-dominated motion via global semantic alignment at the feature level and captures target-specific motion through local refinement, improving trajectory stability and identity consistency. In addition, a Temporal Dependency Feature Pyramid (TDFP) module is developed to perform cross-frame temporal feature fusion, enhancing the continuity and discriminability of tiny-object representations. We further construct a new benchmark dataset, SDM-Car-SU, which simulates multi-directional and multi-speed platform motions to enable systematic evaluation of tracking robustness under varying motion perturbations. Extensive experiments on both simulated and real unstabilized satellite videos demonstrate that DeTracker significantly outperforms existing methods, achieving 61.1% MOTA on SDM-Car-SU and 45.3% MOTA on real satellite video data. The code and dataset will be publicly available at https://github.com/alex-chenjiajun/DeTracker.

cs.CV↗

arXiv2025

All-optical control and multiplexed readout of multiple superconducting qubits

Superconducting quantum circuits operate at millikelvin temperatures, typically requiring independent microwave cables for each qubit for connecting room-temperature control and readout electronics. However, scaling to large-scale processors hosting hundreds of qubits faces a severe input/output (I/O) bottleneck, as the dense cable arrays impose prohibitive constraints on physical footprint, thermal load, wiring complexity, and cost. Here we demonstrate a complete optical I/O architecture for superconducting quantum circuits, in which all control and readout signals are transmitted exclusively via optical photons. Employing a broadband traveling-wave Brillouin microwave-to-optical transducer, we achieve simultaneous frequency-multiplexed optical readout of two qubits. Combined with fiber-integrated photodiode arrays for control signal delivery, this closed-loop optical I/O introduces no measurable degradation to qubit coherence times, with an optically driven single-qubit gate fidelity showing only a 0.19% reduction relative to standard microwave operation. These results establish optical interconnects as a viable path toward large-scale superconducting quantum processors, and open the possibility of networking multiple superconducting quantum computers housed in separate dilution refrigerators through a centralized room-temperature control infrastructure.

quant-ph↗

arXiv2025

Scalable Optical Links for Controlling Bosonic Quantum Processors

Superconducting quantum computing has the potential to revolutionize computational capabilities. However, scaling up large quantum processors is limited by the cumbersome and heat-conductive electronic cables that connect room-temperature control electronics to quantum processors, leading to significant signal attenuation. Optical fibers provide a promising solution, but their use has been restricted to controlling simple two-level quantum systems over short distances. Here, we demonstrate optical control of a bosonic quantum processor, achieving universal operations on the joint Hilbert space of a transmon qubit and a storage cavity. Using an array of cryogenic fiber-integrated uni-traveling-carrier photodiodes, we prepare Fock states containing up to ten photons. Additionally, remote control of bosonic modes over a transmission distance of 15 km has been achieved, with fidelities exceeding 95%. The combination of high-dimensional quantum control, multi-channel operation, and long-distance transmission addresses the key requirements for scaling superconducting quantum computers and enables architectures for distributed quantum data centers.

quant-ph↗