arXiv subjects

Zhen Zhang

Publications and source records attributed to Zhen Zhang.

At least 145 records · Page 8Linked to original sources

arXiv2026

CM2: Reinforcement Learning with Checklist Rewards for Multi-Turn and Multi-Step Agentic Tool Use

AI agents are increasingly used to solve real-world tasks by reasoning over multi-turn user interactions and invoking external tools. However, applying reinforcement learning to such settings remains difficult: realistic objectives often lack verifiable rewards and instead emphasize open-ended behaviors; moreover, RL for multi-turn, multi-step agentic tool use is still underexplored; and building and maintaining executable tool environments is costly, limiting scale and coverage. We propose CM2, an RL framework that replaces verifiable outcome rewards with checklist rewards. CM2 decomposes each turn's intended behavior into fine-grained binary criteria with explicit evidence grounding and structured metadata, turning open-ended judging into more stable classification-style decisions. To balance stability and informativeness, our method adopts a strategy of sparse reward assignment but dense evaluation criteria. Training is performed in a scalable LLM-simulated tool environment, avoiding heavy engineering for large tool sets. Experiments show that CM2 consistently improves over supervised fine-tuning. Starting from an 8B Base model and training on an 8k-example RL dataset, CM2 improves over the SFT counterpart by 8 points on tau^-Bench, by 10 points on BFCL-V4, and by 12 points on ToolSandbox. The results match or even outperform similarly sized open-source baselines, including the judging model. CM2 thus provides a scalable recipe for optimizing multi-turn, multi-step tool-using agents without relying on verifiable rewards. Code provided by the open-source community: https://github.com/namezhenzhang/CM2-RLCR-Tool-Agent.

cs.AI↗

arXiv2026

Anisotropic flows in Au+Au collisions at $\sqrt{s_{\rm{NN}}} = 2.4\,\text{GeV}$ with a Skyrme pseudopotential

Within the framework of the lattice Boltzmann-Uehling-Uhlenbeck transport model, we present a systematic study of proton anisotropic flow observables measured by the HADES collaboration, by utilizing the recently developed density-, momentum- and isospin-dependent N$5$LO Skyrme pseudopotential. In particular, we investigate the impacts of the momentum dependence of nucleon mean-field potentials, the stiffness of symmetric nuclear matter (SNM) EOS, the high-density behaviors of the symmetry energy and the in-medium modification of nucleon-nucleon elastic cross sections on proton $v_{1}$, $v_{2}$, $v_{3}$, and $v_{4}$ in Au+Au collisions at $\sqrt{s_{\rm{NN}}} = 2.4\,\text{GeV}$. Our results show that the proton anisotropic flows are strongly sensitive to the momentum dependence of nucleon mean-field potential as well as the incompressibility coefficient $K_0$ of SNM. In addition, the transverse momentum dependence of the proton $v_2$ exhibits a modest sensitivity to the higher-order skewness coefficient $J_0$ and kurtosis coefficient $I_0$ of SNM as well as the momentum dependence of the symmetry potential, while the transverse momentum dependence of the proton $v_1$ is shown to modestly depend on the in-medium modification of nucleon-nucleon elastic cross sections. Moreover, the high-density symmetry energy seems to have limited effects on the proton anisotropic flows. These findings highlight the necessity of considering the momentum dependence of nucleon mean-field potentials including the symmetry potential, the higher-order characteristic parameters of SNM EOS beyond $K_0$, and the in-medium modification of nucleon-nucleon elastic cross sections, in future Bayesian transport model analyses on proton anisotropic flows in heavy-ion collisions at HADES energies, thereby to extract information on nuclear matter EOS as well as the associated underlying nuclear effective interactions.

nucl-th↗

arXiv2026

Identifying Weight-Variant Latent Causal Models

The task of causal representation learning aims to uncover latent higher-level causal variables that affect lower-level observations. Identifying the true latent causal variables from observed data, while allowing instantaneous causal relations among latent variables, remains a challenge, however. To this end, we start with the analysis of three intrinsic indeterminacies in identifying latent variables from observations: transitivity, permutation indeterminacy, and scaling indeterminacy. We find that transitivity acts as a key role in impeding the identifiability of latent causal variables. To address the unidentifiable issue due to transitivity, we introduce a novel identifiability condition where the underlying latent causal model satisfies a linear-Gaussian model, in which the causal coefficients and the distribution of Gaussian noise are modulated by an additional observed variable. Under certain assumptions, including the existence of a reference condition under which latent causal influences vanish, we can show that the latent causal variables can be identified up to trivial permutation and scaling, and that partial identifiability results can still be obtained when this reference condition is violated for a subset of latent variables. Furthermore, based on these theoretical results, we propose a novel method, termed Structural caUsAl Variational autoEncoder (SuaVE), which directly learns causal representations and causal relationships among them, together with the mapping from the latent causal variables to the observed ones. Experimental results on synthetic and real data demonstrate the identifiability and consistency results and the efficacy of SuaVE in learning causal representations.

cs.LG↗

arXiv2026

Neural Evolutionary Kernel Method: A Knowledge-Guided Framework for Solving Evolutionary PDEs

Numerical solution of partial differential equations (PDEs) plays a vital role in various fields of science and engineering. In recent years, deep neural networks (DNNs) have emerged as a powerful tool for solving PDEs, leveraging their approximation capabilities to handle complex domains and high-dimensional problems. Among these, operator learning has gained increasing attention by learning mappings between function spaces using DNNs. This paper proposes a novel approach, termed the Neural Evolutionary Kernel Method (NEKM), for solving a class of time-dependent partial differential equations (PDEs) via deep neural network (DNN)-based kernel representations. By integrating boundary integral techniques with operator learning, prior mathematical information of time-dependent partial differential equations (PDEs) is embedded into the design of neural network architectures for predicting their solutions, enhancing both computational efficiency and solution accuracy. Numerical experiments on the heat, wave, and Schrödinger equations demonstrate that the Neural Evolutionary Kernel Method (NEKM) achieves high accuracy and favorable computational efficiency. Furthermore, the operator learning framework inherently supports the simultaneous prediction of solutions to multiple PDEs with different coefficients, rendering its capability for solving random PDEs.

math.NA↗

arXiv2026

GECAM discovery of the second FRB-associated Magnetar X-ray Burst from SGR J1935+2154

Fast radio burst (FRB) is mysterious phenomenon with millisecond-duration radio pulses observed mostly from cosmological distance. The association between FRB 200428 and a magnetar X-ray burst (MXB) from SGR J1935+2154 has significantly advanced the understanding of FRB and magnetar bursts. However, it is uncertain whether this association between MXB and FRB (i.e. MXB/FRB 200428) is genuine or just coincidental only based on this single event. Here we report the discovery of a bright ($\rm\sim7.6\times10^{-7}\,erg \cdot cm^{-2}$ in 1-250 keV) magnetar X-ray burst detected by GECAM on October 14th, 2022 (dubbed as MXB 221014) from SGR J1935+2154, which is associated with a FRB detected by CHIME and GBT. We conducted a detailed temporal and spectral analysis of MXB 221014 with GECAM data and find that it is a bright and typical ($T_{90}\sim$250 ms) X-ray burst from this magnetar. Interestingly, we find two narrow X-ray pulses in the MXB, one of which temporally aligns with the main pulse of the FRB 221014 $\sim5.70$ ms latter than the peak time of FRB 221014), resembling the feature found in MXB/FRB 200428. Furthermore, we did comprehensive comparison between MXB/FRB 221014 and MXB/FRB 200428, and find that while the two events share several common features, they also exhibit distinct differences, highlighting the variety of the MXB-FRB association morphology. This finding not only confirms the association between MXB and FRB but also provides new insights into the mechanism of and the relationship between FRB and MXB.

astro-ph.HE↗

arXiv2026

Systematic Study of the Simultaneous Events Detected by GECAM

GECAM is a constellation of all-sky monitors in hard X-ray and gamma-ray band primarily aimed at high energy transients such as gamma-ray bursts, soft gamma-ray repeaters, solar flares and terrestrial gamma-ray flashes. As GECAM has the highest temporal resolution (0.1~$μ$s) among instruments of its kind, it can identify the so-called simultaneous events (STE) that deposit signals in multiple detectors nearly at the same time (with a 0.3~$μ$s window). However, the properties and origin of STE have not yet been explored. In this work, we implemented, for the first time, a comprehensive analysis of the STE detected by GECAM, including their morphology, energy deposition, and the dependence on the geomagnetic coordinates. We find that these STE probably result from direct interactions between high-energy charged cosmic rays and satellite. These results demonstrate that GECAM can detect, identify, and characterize high-energy cosmic rays, making it a Micro Cosmic-Ray Observatory (MICRO) in low Earth orbit.

astro-ph.HE↗

arXiv2026

Theoretical Modeling of Large Language Model Self-Improvement Training Dynamics Through Solver-Verifier Gap

Self-improvement is a significant techniques within the realm of large language model (LLM), aiming to enhance the LLM performance without relying on external data. Despite its significance, generally how LLM performances evolve during the self-improvement process remains underexplored. In this paper, we theoretically model the training dynamics of self-improvement via the concept of solver-verifier gap. This is inspired by the conjecture that the performance enhancement of self-improvement stems from the gap between LLM's solver capability and verifier capability. Based on the theoretical framework, we further show how to model the entire training trajectory. This framework allows quantifying the capability limit of self-improvement by fitting the theoretical model to the experiment results. We validate the effectiveness of the theoretical framework on various LLMs and datasets. Beyond self-improvement, we extend our analysis to investigate how external data influences these dynamics within the framework. Notably, we find that under limited external data regimes, such external data can be utilized at any stage without significantly affecting final performances, which accords with the empirical observations.

cs.LG↗

arXiv2026

CA-YOLO: Cross Attention Empowered YOLO for Biomimetic Localization

In modern complex environments, achieving accurate and efficient target localization is essential in numerous fields. However, existing systems often face limitations in both accuracy and the ability to recognize small targets. In this study, we propose a bionic stabilized localization system based on CA-YOLO, designed to enhance both target localization accuracy and small target recognition capabilities. Acting as the "brain" of the system, the target detection algorithm emulates the visual focusing mechanism of animals by integrating bionic modules into the YOLO backbone network. These modules include the introduction of a small target detection head and the development of a Characteristic Fusion Attention Mechanism (CFAM). Furthermore, drawing inspiration from the human Vestibulo-Ocular Reflex (VOR), a bionic pan-tilt tracking control strategy is developed, which incorporates central positioning, stability optimization, adaptive control coefficient adjustment, and an intelligent recapture function. The experimental results show that CA-YOLO outperforms the original model on standard datasets (COCO and VisDrone), with average accuracy metrics improved by 3.94%and 4.90%, respectively.Further time-sensitive target localization experiments validate the effectiveness and practicality of this bionic stabilized localization system.

cs.CV↗

arXiv2026

Group-Evolving Agents: Open-Ended Self-Improvement via Experience Sharing

Open-ended self-improving agents can autonomously modify their own structural designs to advance their capabilities and overcome the limits of pre-defined architectures, thus reducing reliance on human intervention. We introduce Group-Evolving Agents (GEA), a new paradigm for open-ended self-improvements, which treats a group of agents as the fundamental evolutionary unit, enabling explicit experience sharing and reuse within the group throughout evolution. Unlike existing open-ended self-evolving paradigms that adopt tree-structured evolution, GEA overcomes the limitation of inefficient utilization of exploratory diversity caused by isolated evolutionary branches. We evaluate GEA on challenging coding benchmarks, where it significantly outperforms state-of-the-art self-evolving methods (71.0% vs. 56.7% on SWE-bench Verified, 88.3% vs. 68.3% on Polyglot) and matches or exceeds top human-designed agent frameworks (71.8% and 52.0% on two benchmarks, respectively). Analysis reveals that GEA more effectively converts early-stage exploratory diversity into sustained, long-term progress, achieving stronger performance under the same number of evolved agents. Furthermore, GEA exhibits consistent transferability across different coding models and greater robustness, fixing framework-level bugs in 1.4 iterations on average, versus 5 for self-evolving methods.

cs.AI↗

arXiv2026

Variational and Monte Carlo Methods for Bayesian Inversion of Dynamic Subsurface Flow Simulations Using Seismic and Fluid Pressure Data

In order to predict future performance of subsurface fluid reservoirs under possible operating scenarios, a dynamic, porous-medium flow simulation model must be tuned to include representative properties of the reservoir. Estimating subsurface reservoir properties given remotely sensed or borehole-based observations typically involves finding the solution to a challenging inverse problem. We compare Monte Carlo random sampling to variational inference methods which use optimisation to constrain parametrised uncertainties in nonlinear Bayesian inversions. We use them to estimate the posterior probability distribution of reservoir permeability given fluid pressure and seismic measurements. The methods include automatic differentiation variational inference (ADVI), Stein variational gradient descent (SVGD), and a Monte Carlo method called stochastic SVGD (sSVGD), all of which we benchmark against results from Metropolis-Hastings McMC. We also test an ADVI variant called physically structured variational inference (PSVI): in our implementation this method estimates only spatially-local correlations between model parameters based on the intuition that such correlations are strong in remote sensing problems in which data only inform about spatial-averages of local dynamics. We apply the methods to two- and three-dimensional inverse problems of carbon dioxide storage, inspired by the Endurance field, located in the UK North Sea. Results show that PSVI achieves a good balance between mean-field ADVI and full-rank ADVI in terms of accuracy of the posterior approximation and computational efficiency. SVGD and sSVGD offer more accurate approximations of the target posterior distribution, but at far higher computational cost. Between them, sSVGD outperforms SVGD, exhibiting better computational efficiency and mitigating the problems of mode collapse and spurious correlations.

physics.geo-ph↗

arXiv2026

Towards Identifiable Latent Additive Noise Models

Causal representation learning (CRL) offers the promise of uncovering the underlying causal model by which observed data was generated, but the practical applicability of existing methods remains limited by the strong assumptions required for identifiability and by challenges in applying them to real-world settings. Most current approaches are applicable only to relatively restrictive model classes, such as linear or polynomial models, which limits their flexibility and robustness in practice. One promising approach to this problem seeks to address these issues by leveraging changes in causal influences among latent variables. In this vein we propose a more general and relaxed framework than typically applied, formulated by imposing constraints on the function classes applied. Within this framework, we establish partial identifiability results under weaker conditions, including scenarios where only a subset of causal influences change. We then extend our analysis to a broader class of latent post-nonlinear models. Building on these theoretical insights, we develop a flexible method for learning latent causal representations. We demonstrate the effectiveness of our approach on synthetic and semi-synthetic datasets, and further showcase its applicability in a case study on human motion analysis, a complex real-world domain that also highlights the potential to broaden the practical reach of identifiable CRL models.

cs.LG↗

arXiv2026

Efficient Multimodal Planning Agent for Visual Question-Answering

Visual Question-Answering (VQA) is a challenging multimodal task that requires integrating visual and textual information to generate accurate responses. While multimodal Retrieval-Augmented Generation (mRAG) has shown promise in enhancing VQA systems by providing more evidence on both image and text sides, the default procedure that addresses VQA queries, especially the knowledge-intensive ones, often relies on multi-stage pipelines of mRAG with inherent dependencies. To mitigate the inefficiency limitations while maintaining VQA task performance, this paper proposes a method that trains a multimodal planning agent, dynamically decomposing the mRAG pipeline to solve the VQA task. Our method optimizes the trade-off between efficiency and effectiveness by training the agent to intelligently determine the necessity of each mRAG step. In our experiments, the agent can help reduce redundant computations, cutting search time by over 60\% compared to existing methods and decreasing costly tool calls. Meanwhile, experiments demonstrate that our method outperforms all baselines, including a Deep Research agent and a carefully designed prompt-based method, on average over six various datasets. Code will be released.

cs.CL↗

arXiv2026

Self-Supervised Path Planning in Unstructured Environments via Global-Guided Differentiable Hard Constraint Projection

Deploying deep learning agents for autonomous navigation in unstructured environments faces critical challenges regarding safety, data scarcity, and limited computational resources. Traditional solvers often suffer from high latency, while emerging learning-based approaches struggle to ensure deterministic feasibility. To bridge the gap from embodied to embedded intelligence, we propose a self-supervised framework incorporating a differentiable hard constraint projection layer for runtime assurance. To mitigate data scarcity, we construct a Global-Guided Artificial Potential Field (G-APF), which provides dense supervision signals without manual labeling. To enforce actuator limitations and geometric constraints efficiently, we employ an adaptive neural projection layer, which iteratively rectifies the coarse network output onto the feasible manifold. Extensive benchmarks on a test set of 20,000 scenarios demonstrate an 88.75\% success rate, substantiating the enhanced operational safety. Closed-loop experiments in CARLA further validate the physical realizability of the planned paths under dynamic constraints. Furthermore, deployment verification on an NVIDIA Jetson Orin NX confirms an inference latency of 94 ms, showing real-time feasibility on resource-constrained embedded hardware. This framework offers a generalized paradigm for embedding physical laws into neural architectures, providing a viable direction for solving constrained optimization in mechatronics. Source code is available at: https://github.com/wzq-13/SSHC.git.

cs.RO↗

arXiv2026

Brightest GRB flare observed in GRB 221009A: bridge the last gap between flare and prompt emission in GRB

Flares are usually observed during the afterglow phase of Gamma-Ray Bursts (GRBs) in soft X-ray, optical and radio bands, but rarely in gamma-ray band. Despite the extraordinary brightness, GECAM-C has accurately measured both the bright prompt emission and flare emission of GRB 221009A without instrumental effects, offering a good opportunity to study the relation between them. In this work, we present a comprehensive analysis of flare emission of GRB 221009A, which is composed of a series of flares. Among them, we identify an exceptionally bright flare with a record-breaking isotropic energy $E_{\rm iso} = 1.82 \times 10^{53}$ erg of GRB flares. It exhibits the highest peak energy ever detected in GRB flares, $E_{\rm peak} \sim 300$ keV, making it a genuine gamma-ray flare. It also shows rapid rise and decay timescales, significantly shorter than those of typical X-ray flares observed in soft X-ray or optical band, but comparable to those observed in prompt emissions. Despite these exceptional properties, the flare shares several common properties with typical GRB flares. We note that this is the first observation of a GRB flare in the keV-MeV band with sufficiently high temporal resolution and high statistics, which bridges the last gap between prompt emission and flare.

astro-ph.HE↗

arXiv2026

Communication-Efficient Federated Learning by Exploiting Spatio-Temporal Correlations of Gradients

Communication overhead is a critical challenge in federated learning, particularly in bandwidth-constrained networks. Although many methods have been proposed to reduce communication overhead, most focus solely on compressing individual gradients, overlooking the temporal correlations among them. Prior studies have shown that gradients exhibit spatial correlations, typically reflected in low-rank structures. Through empirical analysis, we further observe a strong temporal correlation between client gradients across adjacent rounds. Based on these observations, we propose GradESTC, a compression technique that exploits both spatial and temporal gradient correlations. GradESTC exploits spatial correlations to decompose each full gradient into a compact set of basis vectors and corresponding combination coefficients. By exploiting temporal correlations, only a small portion of the basis vectors need to be dynamically updated in each round. GradESTC significantly reduces communication overhead by transmitting lightweight combination coefficients and a limited number of updated basis vectors instead of the full gradients. Extensive experiments show that, upon reaching a target accuracy level near convergence, GradESTC reduces uplink communication by an average of 39.79% compared to the strongest baseline, while maintaining comparable convergence speed and final accuracy to uncompressed FedAvg. By effectively leveraging spatio-temporal gradient structures, GradESTC offers a practical and scalable solution for communication-efficient federated learning.

cs.LG↗

arXiv2026

GI-Bench: A Panoramic Benchmark Revealing the Knowledge-Experience Dissociation of Multimodal Large Language Models in Gastrointestinal Endoscopy Against Clinical Standards

Multimodal Large Language Models (MLLMs) show promise in gastroenterology, yet their performance against comprehensive clinical workflows and human benchmarks remains unverified. To systematically evaluate state-of-the-art MLLMs across a panoramic gastrointestinal endoscopy workflow and determine their clinical utility compared with human endoscopists. We constructed GI-Bench, a benchmark encompassing 20 fine-grained lesion categories. Twelve MLLMs were evaluated across a five-stage clinical workflow: anatomical localization, lesion identification, diagnosis, findings description, and management. Model performance was benchmarked against three junior endoscopists and three residency trainees using Macro-F1, mean Intersection-over-Union (mIoU), and multi-dimensional Likert scale. Gemini-3-Pro achieved state-of-the-art performance. In diagnostic reasoning, top-tier models (Macro-F1 0.641) outperformed trainees (0.492) and rivaled junior endoscopists (0.727; p>0.05). However, a critical "spatial grounding bottleneck" persisted; human lesion localization (mIoU >0.506) significantly outperformed the best model (0.345; p<0.05). Furthermore, qualitative analysis revealed a "fluency-accuracy paradox": models generated reports with superior linguistic readability compared with humans (p<0.05) but exhibited significantly lower factual correctness (p<0.05) due to "over-interpretation" and hallucination of visual features. GI-Bench maintains a dynamic leaderboard that tracks the evolving performance of MLLMs in clinical endoscopy. The current rankings and benchmark results are available at https://roterdl.github.io/GIBench/.

cs.CV↗

arXiv2026

Evidence for strong isovector nuclear spin-orbit interaction

The nucleon spin-orbit interaction is a cornerstone of nuclear structure theory, yet its isospin dependence remains elusive owing to the lack of clean experimental probes. Here we show that the charge-weak form factor difference in $^{48}$Ca, recently extracted in a model-independent manner by the CREX experiment, exhibits strong sensitivity to the isovector spin-orbit interaction. Using Skyrme-like energy density functionals, we demonstrate that a significantly enhanced isovector spin-orbit interaction, about four times stronger than conventional parametrizations, can resolve the PREX-CREX puzzle, which has challenged modern nuclear theories and our understanding of nuclear symmetry energy, while maintaining a good description of nuclear bulk properties and well-established shell structure of finite nuclei. This enhanced isovector spin-orbit interaction also provides a novel mechanism for the emergence of the $N = 14$, $16$, $32$ and $34$ magic numbers in neutron-rich nuclei on the mean-field level. These findings point to a strong isospin dependence of the nucleon spin-orbit interaction, which is expected to have important implications for nuclear structures, electroweak nuclear processes, and related problems in nuclear astrophysics.

nucl-th↗

arXiv2026

Upstream Laser-based Longitudinal Enhancement of Relativistic Photoelectrons

Controlling the longitudinal phase space of high-brightness relativistic electron beams is crucial for advancing a broad spectrum of charged-particle-based instrumentation and scientific frontiers. A generalized method for achieving this control involves manipulating the photoemission laser's temporal distribution at the picosecond level, a long-standing technical challenge. Recent developments in laser shaping have enabled the creation of high-power, picosecond-scale symmetrical and asymmetrical temporal profiles, capable of fine-tuning complex space-charge dynamics and external field effects in relativistic charged-particle beams. Here, we demonstrate that rather than deviations from theorized, idealized laser distributions, a controlled asymmetry can be harnessed to counteract accelerator-induced distortions. By implementing spatiotemporal shaping of the ultraviolet photocathode laser at the LCLS-II superconducting injector, we achieve deterministic control over the longitudinal phase space without downstream corrections. We find that this optical asymmetry induces a self-linearizing effect across both low (40 pC) and high (80 pC) charge regimes, effectively suppressing nonlinear compression and energy chirp. Consequently, this approach is expected to preserve a low emittance comparable to that of ideal flattop or regular Gaussian profiles, while delivering superior current uniformity and shot-to-shot stability. These results establish spatiotemporal laser shaping as a compact, generalizable tool for directly optimizing beam brightness at the source.

physics.optics↗