arXiv subjects

Fan Guo

Publications and source records attributed to Fan Guo.

At least 19 recordsLinked to original sources

arXiv2026

Energetic Nonthermal Electrons within the Above-the-looptop Regions in Solar Flares: Acceleration, Feedback, and Quasi-Periodic Pulsations

Solar flares are among the most dramatic events in the solar system, releasing substantial magnetic energy and accelerating a large number of electrons to high energies. Notably, in certain events, the above-the-looptop region may contain a significant population of nonthermal electrons, both in number and energy. For the first time, we adopt a novel numerical method that combines magnetohydrodynamics with energetic particles incorporating feedback from nonthermal electrons to investigate electron acceleration and transport in solar flares. We find that a large fraction of energetic electrons are accelerated via the current sheet and termination shock regions. Most energetic electrons are concentrated in the above-the-looptop region, carrying a sizable amount of the released energy. We observe that greater feedback of nonthermal electrons leads to steeper energy spectra. The energy density of the nonthermal electrons oscillates due to the periodic impact of magnetic islands into the above-the-looptop, which may help explain the observed quasi-periodic pulsations (QPPs). Our simulations provide new insights into the origin of nonthermal electrons and associated emissions in the above-the-looptop region.

astro-ph.HE↗

arXiv2026

Can Magnetic Reconnection Accelerate Energetic Particles in the Heliosheath?

The mechanism for accelerating energetic particles in the outer heliosphere is still under debate. The termination shock, turbulence, and magnetic reconnection have all been proposed as viable mechanisms. We investigate compression-driven particle acceleration during magnetic reconnection in the heliospheric current sheet using the MHD--SDE framework that incorporates energetic-particle feedback into the MHD evolution. For the heliosheath conditions considered here, we find that reconnection-driven acceleration is inefficient downstream of the termination shock, whereas reconnection near the heliopause can generate stronger compression and produce a power-law differential intensity spectrum $J(\varepsilon)\propto\varepsilon^{-5.4}$. However, even the hardest spectrum achieved during magnetic reconnection remains much steeper than the typical $J(\varepsilon)\propto\varepsilon^{-1.5}$ spectrum observed in the heliosheath. This indicates that magnetic reconnection alone is insufficient to explain the acceleration of energetic particles in the heliosheath.

astro-ph.HE↗

arXiv2026

gKSM: GPU Memory Deduplication at Driver Layer for Serverless LLM Inference

In serverless LLM inference, cloud providers commonly locate multiple instances on a single GPU to improve utilization. However, each instance independently maintains its model weights and runtime state, making duplicate GPU memory the primary bottleneck. Memory deduplication is a widely-used technique for eliminating duplicate data in CPU systems, but extending it to GPU memory faces fundamental challenges. Effective GPU memory deduplication must avoid redundant data movement between CPU and GPU while preserving write isolation for shared pages without native copy-on-write (COW) support. This paper presents gKSM, a transparent GPU memory deduplication framework for serverless LLM inference. gKSM addresses these challenges through CPU-GPU collaborative hash management that offloads hash computation to GPU kernels, zero-copy model weight deduplication that eliminates duplication before model weight transfer, and a cross-privilege page fault handler that combines a lightweight kernel hook and CUDA Virtual Memory Management (VMM) remapping in the user space. gKSM also designs a stable page scanning algorithm and a reuse-aware placement policy in cluster scale. Experiments show that gKSM reduces GPU memory usage by 59.8% while maintaining high throughput compared to ServerlessLLM. Under memory constrained workloads, gKSM improves aggregate throughput by 2.15X and supports five instances concurrently. In cluster level simulation, gKSM achieves a service rate of 99.97%, compared to 78.2% for SLInfer and 26.9% for ServerlessLLM.

cs.OS↗

arXiv2026

Are $E>B$ regions essential for particle injection in relativistic magnetic reconnection?

During relativistic magnetic reconnection, X-points ($E>B$ regions at vanishing guide field) have been proposed to play an essential role in injecting particles into the nonthermal population. Using fully kinetic particle-in-cell simulations at a range of magnetizations $σ\equiv B_0^2/(4πn_0 m_e c^2) = 12$--$800$, we study in detail how the particle energy gains up to the injection threshold $γ_{\rm inj} = σ/4$ are partitioned into work done inside $E>B$ regions (unmagnetized), within one gyration after exit (re-magnetizing), and at all other times (magnetized). The magnetized channel supplies the largest part of the energy gain for $\sim 95\%$ of injected tracers at $σ= 12$ and $\sim 65\%$ at $σ= 800$. Most of the tracers in this magnetized majority never enter an $E>B$ region before injection. We also compare these injection shares with the earlier studies that proposed an essential role for $E>B$ regions. We conclude that $E>B$ regions are neither necessary for injection nor the dominant direct contributors to pre-injection work by injection share in two-dimensional, nonradiative pair-plasma reconnection at vanishing guide field.

astro-ph.HE↗

arXiv2026

Effects of Coulomb Collisions on the Structures of Fast Magnetosonic Shocks

We present fully kinetic particle-in-cell simulations to demonstrate qualitative effects of binary Coulomb collisions on fast magnetosonic shocks. We find that with sufficiently high collisional frequencies, shock rippling and reformation can be inhibited, leading to stationary, laminar shock fronts. We observe that ion-ion collisions disrupt the motion of particles reflected back upstream that is the key for cyclical reformation. Meanwhile, the electron-ion collisions provide a Spitzer resistivity that can dissipate energy from the instability that causes shock rippling. This result helps us understand the role of collisions in laboratory shocks and possibly the outer heliosphere.

physics.plasm-ph↗

arXiv2026

GDEPO: Group Dual-dynamic and Equal-right Advantage Policy Optimization for Sample-Efficient Reinforcement Learning

Automated Theorem Proving (ATP) is a fundamental AI challenge, requiring machine-verifiable proofs in formal languages like Lean to evaluate reasoning. Reinforcement Learning (RL), especially Group Relative Policy Optimization (GRPO), is a mainstream approach. However, GRPO faces two critical issues in ATP: its relative advantage estimation may conflict with binary verifier feedback under composite rewards, and its static sampling strategy discards batches without valid proofs, causing data waste. Repeated training on mastered problems also wastes computation. To address this, we propose Group Dual-dynamic and Equal-right Advantage Policy Optimization (GDEPO), incorporating three core mechanisms: 1) dynamic additional sampling, resampling ineffective batches until a valid proof is found; 2) equal-right advantage, decoupling the advantage function's sign (correctness) from its magnitude (auxiliary rewards) to ensure stable policy updates while increasing effective computation; and 3) dynamic additional iterations, applying extra gradient steps to initially failed but eventually successful samples to accelerate learning. Experiments on MinF2F-test, MathOlympiadBench, and PutnamBench confirm GDEPO's effectiveness, with ablation studies validating its synergistic components. The method enhances data utilization and optimization efficiency, offering a novel training paradigm for ATP.

cs.AI↗

arXiv2026

Shared Worlds, Private Minds: Structured Memory for Long-Form Writing as World Creation

LLM agents that write long-form fiction need an explicit memory of the evolving storyworld to keep new events consistent with established facts. Such memory must keep heterogeneous narrative information distinct, integrate story developments across granularities, and recover dependencies that a writing request leaves implicit. We present NarraWorld, a structured memory system for long-form writing that treats memory construction as world creation. From a shared evidence-grounded graph, NarraWorld derives four connected views: world facts, per-character beliefs, open developments, and hypothetical branches (possible-world continuations). Hierarchical aggregation with atomic closure consolidates events into scenes, plotlines, and plots, keeping each higher-level node traceable to its constituent source spans. For retrieval, planned reconstruction infers a query's dependencies from the current narrative situation and a preview of memory, then assembles the relevant records within a token budget. Across three writing benchmarks, NarraWorld achieves the strongest aggregate results. Its memory also transfers to situated role-playing and largely preserves recall on a general-purpose long-term memory benchmark, paving the way for agents that sustain coherent storyworlds across diverse narrative tasks.

cs.CL↗

arXiv2026

STAGE: A Full-Screenplay Benchmark for Reasoning over Evolving Stories

Movie screenplays are a demanding testbed for long-form narrative understanding, as characters' goals, beliefs, knowledge, and relationships evolve continuously across scenes. However, existing benchmarks primarily evaluate isolated facts from the completed screenplay, leaving unassessed whether models can track the evolving state of characters as the story unfolds. We introduce STAGE, a benchmark over 151 English and Chinese full-length screenplays, built on a provenance-linked narrative backbone that recovers the state and epistemic access of each character at every point along its timeline. Three tasks derived from the backbone jointly probe whether models can maintain, explain, and act on evolving narrative state: Character Development Tracking updates a focal character's state between checkpoints, Cross-Scene Narrative Evolution Reasoning targets cross-scene state transitions, and In-Script Character Role-Playing requires responses bounded by the character's state and knowledge at a specified point. We identify three failure modes of current LLMs: silent forgetting under recursive state updating, limited cross-scene reasoning even when all relevant evidence is supplied, and a trade-off in role-playing where stylistic character fidelity and screenplay-grounded memory faithfulness are optimized by different memory-access strategies. STAGE thus provides a unified framework for diagnosing how current models fail to track, reason about, and enact story evolution.

cs.CL↗

arXiv2026

Narrative Knowledge Weaver: Narrative-Centric Retrieval-Augmented Reasoning for Long-Form Text Understanding

Long-form narrative QA requires reasoning over evolving story worlds rather than isolated passages: answers may depend on earlier goals, changing character states, social relations, causal triggers, temporal position, and later consequences. Existing retrieval and graph-augmented generation methods improve evidence access, but their units--chunks, entities, relations, summaries, or tool actions--do not directly encode how evidence functions in a story. We introduce Narrative Knowledge Weaver(NKW), a source-grounded framework that aligns textual evidence, atomic facts, canonical graph structure, entity profiles, interactions, episodes, and storylines. At query time, NKW uses text, graph, and narrative tools with post-retrieval reading skills to assemble evidence and audit actor, scope, polarity, state, and temporal constraints. Across STAGE, FairytaleQA, and QuALITY, NKW is strongest on screenplay-level story-world QA while remaining competitive on more passage-centered benchmarks. Ablations, question-type analyses, graph-asset statistics, and case studies show complementary benefits for character, scene, temporal, causal, and narrative-progression reasoning.

cs.CL↗

arXiv2026

On the Nature of Candle-Flame-Shaped Solar Flares and Sub-Alfvénic Supra-Arcade Plasma Downflows

Certain solar flares exhibit a distinctive candle-flame or cusp-shaped feature above the bright flare arcade visible in extreme ultraviolet (EUV) and X-ray channels sensitive to high-temperature plasma. The presence of a cusp-like structure is generally regarded as a key piece of morphological evidence for magnetic reconnection to power explosive energy release in solar flares. In addition, downward-propagating plasma flows above the flare arcade have often been interpreted as outflows driven by magnetic reconnection. However, the relationship between the observed candle-flame-shaped morphology and the underlying magnetic field geometry for reconnection remains unclear. Likewise, the observed speed of the plasma downflows has been found to be too low compared to the upstream Alfvén speed predicted by reconnection theories. With the help of a recently developed three-dimensional magnetohydrodynamics (MHD) model, we examine the locations where magnetic topology changes from antiparallel to closed (Y-points) in a candle-flame-shaped flare, compare the observational emission features with synthetic EUV images generated from the model, and analyze their time evolutions. We also investigate the role of projection effects and line-of-sight integration in the measurements of plasma downflow speeds. Our analysis reveals that the Y-points do not necessarily coincide with the apparent cusp tip. Also, the apparent speeds of the supra-arcade downflows, as derived from tracks in the time-distance plots, underestimate the true Alfvén speeds in the reconnection inflow region by at least a factor of 2 up to an order of magnitude.

astro-ph.SR↗

arXiv2026

The Role of Magnetic Reconnection in Energizing Protons and Heavier Ions at the Heliospheric Current Sheet

During near-Sun crossings of the heliospheric current sheet (HCS), Parker Solar Probe (PSP) observed populations of high-energy protons and heavier ions indicating possible energization by magnetic reconnection up to 10s -- 100s keV nucleon$^{-1}$. Here we study ion acceleration by magnetic reconnection at the HCS. To estimate ion energization, we solve the Parker transport equation coupled to a large-scale 2D MHD reconnection simulation. We find that multiple ion species develop power-law distributions with both spectral index and high-energy cutoff $E_{\text{max}}$ consistent with in-situ data. By accounting for the injection physics determined by kinetic simulations, we confirm that the charge-to-mass ratio scales as $E_{\text{max}} \propto (Q/M)^α$ with $α\sim 0.8-1.1$, approximately consistent with PSP measurements in the broader range $α\sim 0.6-1.7$. In the limit where ions are injected at the same energy per nucleon, $α$ can be as low as $\sim 0.3$. These findings further support the role of magnetic reconnection in producing high-energy heavy ions at the HCS.

astro-ph.SR↗

arXiv2026

Turbulence and its Potential Impact on Solar Chromospheric and Coronal Heating

Low-frequency turbulence in the solar chromosphere remains poorly understood. We address 1) the sources of low-frequency turbulence that potentially heat the chromosphere, and 2) how turbulence is transported and dissipated throughout the chromosphere and lower corona. We use particle-in-cell simulations to investigate mixed polarity magnetic fields corresponding to emergent magnetic carpet field in coronal holes or quiet Sun regions for strong (imbalanced) and weak (balanced) guide magnetic fields. The initial mixed polarity magnetic field transitions rapidly to a turbulent state dominated by advected small-scale nonlinear structures, with a minority slab turbulence population and the emergent field is largely annihilated. Turbulence is anisotropic for imbalanced magnetic field and more isotropic for balanced cases. We develop a transport model for turbulence advected and dissipated throughout the chromosphere by randomly distributed energy-containing scale dynamical flows described by log-normal statistics. We compute the expectations for the total energy per unit volume (h) J m^{-3}, the Elsasser specific energy (h) m^2 s^{-2}, the heating rate <\cdot{H}>(h) J m^{-3} s^{-1}, and the correlation length <λ>(h) km as functions of height h above the photosphere. Turbulent energy is injected into the low corona by a random "patchwork" of sites across the transition region surface. The expected energy injection rates <\cdot{S}> J m^{-2} s^{-1} for the chromosphere and at the base of the corona exceed the estimated energy requirements needed to heat both the chromosphere and corona. Similarly, we show that spicules can be heated gradually with increasing height by entrained magnetic carpet and photospheric turbulence.

astro-ph.SR↗

arXiv2026

Energy-Dependent Polarization Angle Variability as a Robust Diagnostic for Blazar Flaring Mechanisms

Identifying the physical mechanism driving blazar flares remains a central challenge in high-energy astrophysics. We show that the energy dependence of the standard deviation of the polarization angle variability ($σ_\text{PA}$) provides a powerful and robust discriminator of blazar flaring mechanisms. Using particle-in-cell-integrated polarized radiative transfer simulations, we perform to-date the most rigorous statistical analyses of polarization variability. We demonstrate that magnetic reconnection and magnetized turbulence imprint qualitatively distinct energy dependence of $σ_\text{PA}$ that directly reflect their different magnetic field evolution and particle transport. Reconnection predicts higher $σ_\text{PA}$ with higher photon energy till the synchrotron spectral peak, whereas turbulence produces nearly flat $σ_\text{PA}$ across the synchrotron spectral component. These trends are resilient to realistic observational limitations. Applying our results to optical and IXPE data of Mrk~421 and 1ES~1959+650, we find strong evidence for reconnection-driven flares embedded in a turbulent blazar zone. Energy-dependent $σ_\text{PA}$ emerges as a decisive new probe of particle acceleration in relativistic jets.

astro-ph.HE↗

arXiv2026

Vanishing Watermarks: Diffusion-Based Image Editing Undermines Robust Invisible Watermarking

Robust invisible watermarking schemes aim to embed hidden information into images such that the watermark survives common manipulations. However, powerful diffusion-based image generation and editing techniques now pose a new threat to these watermarks. In this paper, we present a comprehensive theoretical and empirical analysis demonstrating that diffusion models can effectively erase robust watermarks even when those watermarks were designed to withstand conventional distortions. We show that a diffusion-driven image regeneration process, which leverages generative models to recreate an image, can remove embedded watermarks while preserving the image's perceptual content. Furthermore, we introduce a guided diffusion-based attack that explicitly targets the embedded watermark signal during generation, significantly degrading watermark detectability. Theoretically, we prove that as an image undergoes sufficient diffusion transformations, the mutual information between the watermarked image and the hidden payload approaches zero, leading to inevitable decoding failure. Experimentally, we evaluate multiple state-of-the-art watermarking methods (including deep learning-based schemes like StegaStamp, TrustMark, and VINE) and demonstrate that diffusion edits yield near-zero watermark recovery rates after attack, while maintaining high visual fidelity of the regenerated images. Our findings reveal a fundamental vulnerability in current robust watermarking techniques against generative model-based edits, underscoring the need for new strategies to ensure watermark resilience in the era of powerful diffusion models.

cs.CR↗

arXiv2026

Heterogeneous Multi-Expert Reinforcement Learning for Long-Horizon Multi-Goal Tasks in Autonomous Forklifts

Autonomous mobile manipulation in unstructured warehouses requires a balance between efficient large-scale navigation and high-precision object interaction. Traditional end-to-end learning approaches often struggle to handle the conflicting demands of these distinct phases. Navigation relies on robust decision-making over large spaces, while manipulation needs high sensitivity to fine local details. Forcing a single network to learn these different objectives simultaneously often causes optimization interference, where improving one task degrades the other. To address these limitations, we propose a Heterogeneous Multi-Expert Reinforcement Learning (HMER) framework tailored for autonomous forklifts. HMER decomposes long-horizon tasks into specialized sub-policies controlled by a Semantic Task Planner. This structure separates macro-level navigation from micro-level manipulation, allowing each expert to focus on its specific action space without interference. The planner coordinates the sequential execution of these experts, bridging the gap between task planning and continuous control. Furthermore, to solve the problem of sparse exploration, we introduce a Hybrid Imitation-Reinforcement Training Strategy. This method uses expert demonstrations to initialize the policy and Reinforcement Learning for fine-tuning. Experiments in Gazebo simulations show that HMER significantly outperforms sequential and end-to-end baselines. Our method achieves a task success rate of 94.2\% (compared to 62.5\% for baselines), reduces operation time by 21.4\%, and maintains placement error within 1.5 cm, validating its efficacy for precise material handling.

cs.RO↗

arXiv2025

Evidence of Time-Dependent Diffusive Shock Acceleration in the 2022 September 5 Solar Energetic Particle Event

On 2022 September 5, a large solar energetic particle (SEP) event was detected by Parker Solar Probe (PSP) and Solar Orbiter (SolO), at heliocentric distances of 0.07 and 0.71 au, respectively. PSP observed an unusual velocity-dispersion signature: particles below $\sim$1 MeV exhibited a normal velocity dispersion, while higher-energy particles displayed an inverse velocity arrival feature, with the most energetic particles arriving later than those at lower energies. The maximum energy increased from about 20-30 MeV upstream to over 60 MeV downstream of the shock. The arrival of SEPs at PSP was significantly delayed relative to the expected onset of the eruption. In contrast, SolO detected a typical large SEP event characterized by a regular velocity dispersion at all energies up to 100 MeV. To understand these features, we simulate particle acceleration and transport from the shock to the observers with our newly developed SEP model - Particle ARizona and MIchigan Solver on Advected Nodes (PARMISAN). Our results reveal that the inverse velocity arrival and delayed particle onset detected by PSP originate from the time-dependent diffusive shock acceleration processes. After shock passage, PSP's magnetic connectivity gradually shifted due to its high velocity near perihelion, detecting high-energy SEPs streaming sunward. Conversely, SolO maintained a stable magnetic connection to the strong shock region where efficient acceleration was achieved. These results underscore the importance of spatial and temporal dependence in SEP acceleration at interplanetary shocks, and provide new insights to understand SEP variations in the inner heliosphere.

astro-ph.SR↗

arXiv2025

Bayesian Statistical Inversion for High-Dimensional Computer Model Output and Spatially Distributed Counts

Data collected by the Interstellar Boundary Explorer (IBEX) satellite, recording heliospheric energetic neutral atoms (ENAs), exhibit a phenomenon that has caused space scientists to revise hypotheses about the physical processes, and computer simulations under those models, in play at the boundary of our solar system. Evaluating the fit of these computer models involves tuning their parameters to observational data from IBEX. This would be a classic (Bayesian) inverse problem if not for three challenges: (1) the computer simulations are slow, limiting the size of campaigns of runs; so (2) surrogate modeling is essential, but outputs are high-resolution images, thwarting conventional methods; and (3) IBEX observations are counts, whereas most inverse problem techniques assume Gaussian field data. To fill that gap we propose a novel approach to Bayesian inverse problems coupling a Poisson response with a sparse Gaussian process surrogate using the Vecchia approximation. We demonstrate the capabilities of our proposed framework, which compare favorably to alternatives, through multiple simulated examples in terms of recovering "true" computer model parameters and accurate out-of-sample prediction. We then apply this new technology to IBEX satellite data and associated computer models developed at Los Alamos National Laboratory.

stat.AP↗

arXiv2025

Effects of Multi-scale Coupling on Particle Acceleration and Energy Partition in Magnetic Reconnection

The interplay between kinetic and macroscopic scales during magnetic reconnection is investigated using particle-in-cell simulations of magnetic island coalescence in the strongly-magnetized, relativistic pair plasma regime. For large system sizes, secondary current sheet formation and downstream turbulence driven by the reconnection outflows dominate the global energy dissipation so that it is causally connected, but spatially and temporally de-coupled from the primary reconnecting current sheet. When compared to simulations of an isolated, force-free current sheet, these dynamics activate additional particle acceleration channels which are responsible for a significant population of the non-thermal particles, modifying the particle energy spectra.

physics.plasm-ph↗