arXiv subjects

Jialin Li

Publications and source records attributed to Jialin Li.

At least 19 recordsLinked to original sources

arXiv2026

Free modules of vector-valued modular forms and applications to Jacobi forms

The space $M(ρ;\mathcal{E})$ of weakly holomorphic modular forms for a dimension-$r$ complex representation $ρ$ of $\mathrm{SL}_2(\mathbb{Z})$ with growth prescribed by $\mathcal{E}$ at the cusp is a free module over $\mathbb{C}[E_4,E_6]$. We first prove that a $\mathbb{C}[E_4,E_6]$-linearly independent homogeneous family $F_1,\ldots,F_r$ in this module gives a basis if and only if $\sum_{i=1}^r \operatorname{wt}(F_i)=12\operatorname{tr}\mathcal{E}$. We then extend this criterion to Jacobi forms. As applications, we determine the generator weights of free modules of $\mathrm{O}(D_{8d}^+)$-invariant Jacobi forms of index $D_{8d}^{+}(p)$, where $D_{8d}^{+}$ is an even unimodular lattice extending $D_{8d}$, $d\ge2$, and $p=2,3,5$.

math.NT↗

arXiv2026

Progressively Refining SAM for Accurate Phytoplankton Segmentation

Accurate phytoplankton segmentation is essential for ecological monitoring, yet it remains largely underexplored due to the subtle visual characteristics of phytoplankton and the scarcity of pixel-level annotations. To address these challenges, we propose PhySeg, a SAM-based segmentation framework equipped with a lightweight refinement module that progressively restores fine-grained details through multi-scale feature fusion. In addition, we construct a phytoplankton segmentation dataset by manually annotating instance-level masks for an existing dataset, providing a valuable resource for future research. Experimental results demonstrate that PhySeg improves segmentation accuracy, validating the effectiveness of the proposed module. Code is available at https://github.com/lijialin404/PhySeg.

cs.CV↗

arXiv2026

SCI-AGENT-EXAM: Benchmarking Agents on Scientific Research Tasks

Large language model (LLM) agents are increasingly capable of supporting scientific workflows, but their capabilities across diverse research activities remain difficult to assess systematically. Existing benchmarks often focus on selected scientific capabilities or task types, limiting a comprehensive characterization of scientific agents' strengths and weaknesses in realistic research workflows. We introduce SCI-AGENT-EXAM, a workflow-oriented benchmark comprising 120 expert-curated tasks across 11 scientific fields and 7 research task types defined by the Scientific Workflow Task Taxonomy (SWTT). Each task requires both executable artifacts (e.g., code and scripts) and textual artifacts (e.g., reports and explanations). To evaluate the heterogeneous task outputs, we develop a hybrid evaluation protocol combining deterministic programmatic verification for executable artifacts with a calibrated, expert-validated LLM-based judge for textual artifacts. Beyond outcome assessment, we analyze progress and solution stability along execution trajectories to characterize how agents reach their final results. Experiments with 10 frontier LLM agents reveal substantial performance gaps: GPT-6 Astra achieves the highest overall score of 80.9, yet only 26.0% of its runs exceed the 0.95 success threshold. Further analyses uncover distinct capability profiles across scientific task types and substantial variation in execution efficiency and stability. Our code and data are available at https://github.com/SKYLENAGE-AI/Sci-Agent-Exam.

cs.CL↗

arXiv2026

Why Bother Me Again? Benchmarking Personal Agent Memory for Interaction Preferences

Personal intelligence systems are taking on increasingly complex everyday tasks, yet repeated questions, unnecessary confirmations, and unwanted outreach can turn capable assistance into interaction burden. Reducing this burden requires agents to remember how users want to be assisted. However, existing memory benchmarks focus on factual recall and personalized answers and rarely test how agents learn and act on interaction preferences. We introduce MemPABench, an interactive environment for on-policy evaluation of memory in personal agents. MemPABench poses long-horizon tasks requiring agents to decide when to clarify, confirm, or act autonomously. Four grounded simulated users hold 14 preferences across work and personal contexts. Their feedback spans 99 sessions over four months, with five event-driven changes per user testing memory updates. We evaluate preference selection, delivered behavior, and sensitivity to context and preference change. Evaluating four memory systems across four LLM backbones, we find that current memory brings limited gains in preference selection. Retrieval-based systems (i.e. Mem0, MemOS, simple RAG) do not outperform No Memory, while File-based memory improves accuracy but applies preferences beyond relevant contexts. Our analysis shows that current memory struggles to retrieve interaction preferences and adapt to preference changes. Even with the correct preference available, overriding the agent's default interaction habits remains challenging. The environment and dataset are available at https://mempabench.com.

cs.AI↗

arXiv2026

Omni-RS: Unifying Multi-Task Understanding and Generation in Remote Sensing

Practical remote sensing applications often require both understanding and generation capabilities for the same geographic scene. However, most existing methods address either understanding or generation in isolation, while existing unified remote sensing models' generative ability is limited to text-to-image. To bridge this gap, we propose Omni-RS, a unified framework supporting four remote sensing understanding tasks and five generation tasks. Specifically, we introduce the "Scene-Guided Two-Stage Training" strategy, which first adapts a multimodal large language model to remote sensing understanding and then uses its hidden states to guide a diffusion transformer for image generation. To enable joint learning across diverse generation tasks, we construct UniGen, a unified multi-task generation dataset, by organizing heterogeneous remote sensing data into a unified format. Furthermore, we introduce "Detailed Visual Conditioning", which uses visual conditions encoded by a VAE to guide the generation process, thereby supporting additional generation tasks that require reference images. Extensive experiments demonstrate that Omni-RS performs exceptionally well in understanding tasks and has achieved outstanding results in a variety of generative tasks.

cs.CV↗

arXiv2026

Averaged proximal reflected gradient method for monotone variational inequalities

Projected reflected gradient (PRG) method proposed by Malitsky is efficient for solving monotone variational inequality (MVI), while the existing upper bound of step size is not tight due to the inequality scaling in the theoretical analysis. In this paper, we construct an averaged variant of PRG method for more general MVI and present a novel Lyapunov function to establish convergent theory. This averaged PRG method provides an improvement of the golden ratio algorithm [Y, Malitsky, Math. Program., 184, 383-410, 2020], and the involved step size is compatible with that for the classical methods, such as Popov's extragradient and forward-reflected-backward methods. Moreover, a fully adaptive strategy without linesearch is presented to adjust step sizes, which generates closed-form and potentially much larger step sizes. Numerical experiments on the Nash-Cournot equilibrium, HpHard, and image reconstruction problems demonstrate that the proposed algorithm significantly outperforms existing state-of-the-art methods.

math.OC↗

arXiv2026

SpikeMoE: Brain-Inspired Competitive Routing for Flexible Spiking Mixture-of-Experts

Spiking Neural Networks (SNNs) enable event-driven computation through biologically inspired dynamics at the neuronal scale, while Mixture-of-Experts (MoE) perform conditional computation through expert selection at the model scale. Integrating their strengths offers potential for flexible neural architectures. A key challenge, however, lies in designing an expert selection mechanism based on spiking activity. To address this, we introduce a spike-based k-WTA Router inspired by competition-inhibition observed in the hippocampal CA1 region. The router incorporates lateral inhibition and refractory period to select Top-K experts according to discrete spike counts. Building on this, we present SpikeMoE, a framework that integrates neuronal-scale spiking dynamics with model-scale expert selection. To address incomplete multisensory inputs in multimodal tasks, we further equip SpikeMoE with a two-stage missing-modality modeling module that combines empirical prototypes from an observed-modality pool with modality-specific learnable embeddings to construct missing-modality representations. Experiments on vision, language, and multimodal benchmarks demonstrate that SpikeMoE achieves state-of-the-art performance among the SNN baselines, matches or exceeds the performance of ANN counterparts, and maintains robustness across diverse missing-modality conditions. These results demonstrate a favorable trade-off between performance and energy efficiency, validating the integration of spiking dynamics with sparse expert computation and highlighting SpikeMoE as a promising approach to energy-efficient brain-inspired computing.

cs.AI↗

arXiv2026

Breaking Fault Lines: Unifying TEE-Assisted BFT Consensus in Partially Trusted Worlds

This paper revisits TEE-assisted BFT under a universal partial-TEE model, where an arbitrary subset of replicas execute inside TEEs while the remaining replicas operate without hardware trust guarantees. We show that heterogeneous trust changes the structure of quorum formation and fault tolerance. In particular, we derive a tight resilience bound f < max {n/3, m/2}, where n is the total number of replicas and m is the number of TEE-enabled replicas. The result reveals a sharp threshold phenomenon: TEEs improve fault tolerance only once they exceed two-thirds of the deployment. Guided by this characterization, we introduce two protocol principles: (1) a dual-quorum construction that safely combines TEE-only and mixed quorums, and (2) a TEE-leader fast path that leverages hardware-enforced non-equivocation to reduce both consensus and view-change latency. We realize these ideas in Raftel, which is, to our knowledge, the first HotStuff-style BFT protocol designed explicitly for arbitrary partial-TEE deployments, and in chained-Raftel, a pipelined variant that further accelerates mixed-trust execution. We implement both protocols atop Intel SGX and evaluate them in LAN and WAN environments. Our results show that Raftel achieves up to 625 TPS with sub-670 ms latency in WAN settings, outperforming HotStuff by up to 308 TPS in throughput while approaching the performance of fully TEE-assisted protocols.

cs.DC↗

arXiv2026

Seamless Reconfiguration for DAG-BFT

Byzantine Atomic Broadcast in Asynchronous networks has been studied extensively for decades. The FLP impossibility result rules out deterministic consensus in a fully asynchronous setting, motivating randomized protocols that combine reliable broadcast with private coins to achieve termination with probability one. More recently, DAG-Rider popularized a new abstraction in which processes continuously reliably broadcast blocks to construct a directed acyclic graph (DAG) and then locally derive a total order using a randomized perfect coin. This separation of dissemination from ordering has renewed interest in asynchronous DAG-based Byzantine fault tolerance and inspired numerous protocols that improve expected latency or throughput. Reconfiguration, however, remains largely unexplored. Existing DAG-based BFT protocols generally assume a static set of processes, whereas deployed systems must occasionally add or remove nodes in response to operational demands. Existing solutions are not seamless; they either pause the system or delay membership changes until a predetermined epoch boundary. This may be impractical and can introduce downtime. However, a seamless solution is non-trivial; agreement on when to start utilizing a new configuration may yield a logical round number that was already surpassed by the concurrent asynchronous dissemination. In this work, we present a seamless reconfiguration protocol for asynchronous DAG-based Byzantine atomic broadcast. Our protocol allows the DAG to continue growing while the membership and corresponding quorum thresholds change, without halting the system. To the best of our knowledge, ours is the first protocol to provide seamless reconfiguration for DAG-based Byzantine atomic broadcast in a fully asynchronous network.

cs.DC↗

arXiv2026

Rankin-Cohen Pairings and Hilbert Hecke Eigenform Product Identities

We establish a Petersson-Rankin-Selberg identity for Hilbert Rankin-Cohen brackets over totally real fields of degree greater than one and arbitrary narrow class number. This extends results of Zhang-Zhang and Zhang-Zhou. For even $k,\ell\ge 2$, $r\in\mathbb{Z}_{\ge 0}$, $K=k+\ell+2r$, and normalized full-level cuspidal Hecke eigentuples $f\in S_K(ω_f)$ and $h\in S_\ell(ω_h)$, we show $\langle f,[E_k,h]_r\rangle=C_{F,k,\ell,r}L^S(k/2,Π(f)\timesΠ(h)^\vee)/L_F(k,ω_fω_h^{-1})\ne 0$, where $C_{F,k,\ell,r}>0$. Specializing to $k=2$, we remove the remaining GRH assumption in the product classification of Hao-Qin-Zhou. Hence, over real quadratic fields of narrow class number one, the only full-level product identities among Hecke eigenforms of even parallel weights at least two, up to interchanging the factors, are $E_4^{\mathrm{eig}}=60(E_2^{\mathrm{eig}})^2$ and $h_8=120E_2^{\mathrm{eig}}h_6$ over $\mathbb{Q}(\sqrt{5})$.

math.NT↗

arXiv2026

TokaGLINT: A Scalable GPU-Tailored Implicit Solver for Full 3D Tokamak Electromagnetic Simulations

We introduce TokaGLINT, a GPU-accelerated implicit solver for electromagnetic field computations in full 3D tokamak simulations, aimed at efficient large-scale parallel GPU computing. Its central innovation lies in the co-design of hierarchical domain decomposition and a fast exact local solver, where hierarchical partitioning is tailored to match fine-grained intra-card subdomains and exploit the tensor-based solver dedicated to curvilinear-coordinate symplectic CN-FDTD-discretized 3D Maxwell equations. Backed by automated operator fusion and batching customized for the intra-card multi-subdomain structure, the solver decouples unknowns through discrete transformations and leverages tensor-structured computations to achieve high hardware utilization, while preserving the long-time stability characteristic of symplectic discretizations. TokaGLINT scales the electromagnetic field solve beyond 10,000 GPUs, achieving 90.1% weak and 53.9% strong scaling efficiency, while delivering a 2.67X single-node speedup over an unpreconditioned BiCGStab baseline (HIP-enabled HYPRE). It is validated in EAST tokamak simulations within the SymPIC plasma simulation code, enabling high-fidelity long-duration modeling.

cs.DC↗

arXiv2026

Detecting Logic Vulnerabilities Across the Contract and Device Layers of Blockchain-Enabled IoT With Multi-Agent Heterogeneous Graph Attention

Blockchain-enabled Internet of Things (IoT) systems integrate smart contracts with embedded devices to support decentralized device management and access control. Their security therefore depends jointly on the logic of on-chain contracts and off-chain device firmware. Logic flaws in either layer can violate the same system invariants, such as unauthorized access, improper state changes, or unguarded privileged operations. Existing approaches rely on contract analysis, firmware analysis, and graph-based vulnerability detection. However, these methods typically focus on a single layer or artifact and often depend on predefined vulnerability patterns, emulation fidelity, or homogeneous representations that obscure security-relevant component roles. They also lack a unified architecture that supports different security tasks while remaining deployable on resource-constrained gateways. To address these limitations, we extend MA-HGAT into a cross-layer multi-agent heterogeneous graph attention framework that models contracts, firmware artifacts, device fleets, and transaction streams with a unified four-role, nine-relation schema. Role-aligned agents exchange heterogeneous evidence through cross-attention, while graph-, link-, and node-level heads support multiple detection tasks and a role-based gateway--cloud partition enables lightweight edge inference. MA-HGAT thus provides a unified and deployable framework for detecting logic vulnerabilities across the contract and device layers of blockchain-enabled IoT systems.

cs.CR↗

arXiv2026

Ossim: OS-Driven Live Simulation for Cluster-Scale Full-Stack Evaluation

Cluster-scale full-stack simulation is essential for evaluating distributed software stacks and emerging hardware components before deployment. Such simulation must achieve both full-stack fidelity for the unmodified production stack and the simulation performance required for iterative configuration exploration. However, no existing method achieves both. We present Ossim, an OS-level approach to cluster-scale full-stack simulation built on top of the Linux virtualization stack. Ossim comprises four subsystems: simulation-oriented scheduling, live memory hierarchy management, simulation-aware IPC, and distributed simulation orchestration. Together, they coordinate live and modeled components under shared simulated time while controlling interference among co-located live hosts. These mechanisms point toward simulation-native OS support, where simulation control and orchestration become core OS responsibilities.

cs.DC↗

arXiv2026

No need to modulate: On-sky results of a Neural Network enhanced pyramid wavefront sensor and prospects for the ELTs

One of the main limitations of ground-based extreme adaptive optics systems (XAO) is the balance between the temporal and photon noise error. The unmodulated Pyramid Wavefront Sensor (uPWFS) promises significant gains in sensitivity over its modulated counterpart, but its practical use is limited by its linearity range. Nonlinear reconstructors provide a pathway to recover this dynamic range while preserving the sensitivity of the uPWFS, thereby reducing photon noise and improving contrast. We present the real-time implementation of a Convolutional Neural Network (CNN) reconstructor and show on-sky results with MagAO-X, demonstrating robust and stable correction across diverse atmospheric conditions. Significant gains over default operation are seen in the low and moderate Strehl regimes, while the performance is slightly degraded in the high Strehl regime. We diagnose this in simulation and mainly attribute this to a non-optimized training dataset for the high-Strehl regime, rather than a fundamental limitation of the approach. Furthermore, initial simulations of the NN-enhanced uPWFS for a downscaled version of the Extremely Large Telescope (ELT) show substantial gains for fast petal-piston control. These results demonstrate that NN-enhanced wavefront sensing is a viable technology for future high-contrast instruments.

astro-ph.IM↗

arXiv2026

Demonstration of simultaneous PIAA- coronagraphy and wavefront sensing using a single metasurface-based focal-plane optic

Controlling residual wavefront aberrations downstream of an extreme adaptive optics (ExAO) system is a major challenge in high-contrast imaging. These aberrations produce quasi-static speckles due to differences between the wavefront-sensing and science paths. Highly sensitive wavefront sensors, such as Zernike wavefront sensors (ZWFSs), are used to mitigate these non-common path aberrations. High-performing coronagraphs, such as complex mask coronagraphs (CMCs), are also implemented in the focal plane. Both perform better with lossless apodization such as phase-induced amplitude apodization (PIAA) optics. Metasurfaces can have chromatic responses, allowing a single focal-plane optic to have different functionalities in different wavelength bands. We demonstrate such an optic by manufacturing a hybrid metasurface designed to function as a CMC and a ZWFS in two intermediate-band filters in the H band, each with a fractional bandwidth of approximately 1\%. We show measured optical responses with phases of $\sim π/2$ at shorter wavelengths and $π$ at longer wavelengths between $1500$ and $1700,\text{nm}$. This would allow for wavefront sensing at the shorter wavelength of $\sim1500\,\text{nm}$ and coronagraphy at the longer wavelength of $\sim1700\,\text{nm}$. Additionally, we tested the mask on-sky with the MagAO-X instrument at the Magellan Clay 6.5 m telescope at Las Campanas Observatory, Chile. On-sky results show a contrast of $\sim 10^{-1}$ at a non-ideal wavelength of $\sim 1600\,\text{nm}$. This is comparable to simulated contrast curves using the measured optical responses around that wavelength. Finally, we evaluated the wavefront-sensing performance of the metasurface using the MagAO-X internal source at $1300\,\mathrm{nm}$. The measured reconstruction error is consistent with simulations of an ideal Zernike wavefront sensor, confirming its wavefront-sensing functionality.

astro-ph.IM↗

arXiv2026

Sub-diffraction-limited coronagraphic imaging with nano-printed PIAACMC phase masks

Imaging Earth-like exoplanets in the habitable zone of their host star is among the main science objectives of future ground-based and space-based observatories. However, the extreme contrast and small separations needed to image such planets cannot be reached with current technology. The Phase-Induced Amplitude Apodization Complex Mask Coronagraph (PIAACMC) is a promising coronagraph to reach this goal. The PIAACMC uses a set of aspheric lenses to apodize the entrance pupil without throughput losses and a phase-shifting focal plane mask for starlight suppression. These allow us to maintain high throughput and achieve a small inner-working angle (IWA), unlocking the capability to observe exoplanets at the diffraction limit. The masks are manufactured in-house at Leiden University with Nanoscribe, a micro-3D-printer that uses two-photon polymerization to achieve sub-micron precision in height. We present the first scientific results with a focal plane mask for the PIAACMC on the Magellan Adaptive Optics eXtreme (MagAO-X) instrument for the 6.5-meter Magellan Clay telescope at Las Campanas Observatory, Chile. We show laboratory and on-sky contrast curves with a broadband z' filter centered at 908 nm with a 14% bandwidth. We use the PIAACMC to detect binary companions at separations ~0.8-5 lambda/D (~23-144 mas). This demonstrates the PIAACMC's capability to observe at the diffraction limit and below, with a sub-lambda/D IWA. Future work includes exploring new mask designs to improve the contrast in broadband light and performing active focal plane wavefront sensing and control.

astro-ph.IM↗

arXiv2026

Digging dark holes on-sky with the Self-Coherent Camera: Preliminary results

Current high-contrast imaging instruments are limited by wavefront errors originating from non-common path aberrations (NCPAs) due, for example, to manufacturing errors in the optics and temperature drifts in the system. These create quasi-static speckles in the final science image that are difficult to distinguish from companions. Therefore, focal plane wavefront sensing and control is needed to suppress speckles. The Self-Coherent Camera (SCC) is a wavefront sensor that allows us to estimate the stellar complex speckle field. In the Fast Atmospheric SCC Technique (FAST), the on-axis starlight hits a coronagraphic focal plane phase mask and is diffracted outside the Lyot stop where it is spatially filtered by a pinhole to create a reference beam. The reference beam and the leaked starlight are recombined on the science plane, creating interference fringes that do not affect the companion, because of incoherence. The focal plane mask was manufactured in-house at Leiden University with Nanoscribe, a micro-3D-printer that uses two-photon polymerization to achieve sub-micron precision in height. We present preliminary results of the first on-sky closed-loop SCC demonstration with the Magellan Adaptive Optics eXtreme (MagAO-X) instrument on the 6.5-meter Magellan Clay telescope at Las Campanas Observatory, Chile. We achieve a 1-sigma raw contrast improvement of a factor 10 in the desired dark hole region with FAST closed-loop control. In the future, we will show observations of stars with companions and use the SCC in post-processing as a Coherent Differential Imaging (CDI) technique to enhance the contrast even further.

astro-ph.IM↗

arXiv2026

The Roman Coronagraph Community Participation Program: pre-launch reference star list and impact of reference star properties on post-processing performance

The upcoming Roman Coronagraph will be the first high-contrast instrument in space capable of high-order wavefront sensing and control technologies, a critical technology demonstration for the proposed Habitable Worlds Observatory (HWO) that aims to directly image and characterize habitable exoEarths. The nominal Roman Coronagraph observing plan involves alternating observations of a science target and a bright, nearby reference star for both wavefront calibration and reference differential imaging post-processing. Reference star criteria for the most demanding coronagraph mode are restrictive, limiting the sample to only 40 candidates for which thorough observational vetting is needed to assess their suitability. Reference star properties such as resolved diameters, presence of circumstellar dust, and close point sources may also have more subtle impacts on post-processing efficacy that may inhibit final contrast performance. In this work, we describe the current progress of the CoronaGraph Instrument Reference stars for Exoplanets (CorGI-REx) observing campaign, a 300+-hour observing campaign that utilizes instruments from around the world to vet reference stars for high-order wavefront control suitability. We will present the pre-launch list of reference star candidates being utilized for the Roman Coronagraph Observation Phase constructed from a thorough analysis of high contrast and interferometric observations. We will also present the results of simulations investigating the impact of reference star resolved diameters and companions on post-processing performance. We conclude by discussing the importance of reference star selection for scheduling observations and optimizing contrast performance for the Roman Coronagraph along with implications for HWO coronagraph operations.

astro-ph.IM↗