Search arXiv⌕ Search

arXiv subjects

Ke Feng

Publications and source records attributed to Ke Feng.

At least 19 recordsLinked to original sources
arXiv2026

TactiNavi: Serving and Revising an LLM Game Coach with Explicit Strategy Conditions

A deployed LLM coaching agent answers requests whose correct response depends on player state and game version, so advice that is right for one player can be wrong for the next. We describe TactiNavi, the serving system behind the LLM coach of a commercial open-world game, which makes strategy applicability an explicit artifact shared by offline compilation, online execution, failure diagnosis, and release. An offline compiler turns versioned game evidence and expert guidance into strategy packages with explicit prerequisites, the serving plane evaluates those prerequisites as three-valued conditions over structured player state and records decision traces, and a control plane uses the resulting condition-rule dependencies to validate revisions on held-out traffic before release. In a retrospective evaluation of 1,122 logged coaching turns from three deployment periods, condition-bearing packages raise correctness from 57.9% to 82.0% over atomic facts and reduce strategy misapplication under changed player state from 18.1% to 10.4%, while trace-based claim verification reaches 86.4% precision at a median reply latency of 2.07 s. The deployment also showed that most apparent fixes are generation noise, because an unchanged-system rerun recovers 38.4% of failing turns and only one of 15 shipped revision units passes the release checks retrospectively. We distill these results into lessons on treating applicability as a first-class serving object, sizing traces for later diagnosis, and building the null model into the release path.

cs.AI↗
arXiv2026

Non-geometric ideal triangulations with all edge valences at least six

Costantino, Frigerio, Martelli and Petronio conjectured that an ideal triangulation in which every edge has valence at least six can be realized by hyperbolic partially truncated tetrahedra on the compact 3-manifold it defines. We show that this is false. Starting from the figure-eight knot complement, we construct an ideal triangulation with $2688$ tetrahedra and edge valences six and eighteen whose manifold is hyperbolic with totally geodesic boundary, but which admits no such realization. The obstruction is a critical point of the covolume energy of Luo and Yang at which three tetrahedra have collapsed: convexity then excludes every geometric competitor, an Euler characteristic count excludes edges of length zero, and the critical point itself is located by a finite certificate in exact integer arithmetic. The obstruction persists in every finite cover, while a complementary theorem shows that no counterexample can have only two edge lengths or only two orbits of edges.

math.GT↗
arXiv2026

A Stochastic Geometry Framework for Performance Analysis of RIS-assisted OFDM Cellular Networks

The reconfigurable intelligent surface (RIS) technology allows one to engineer spatial diversity in complex cellular networks. This paper provides a framework for the system-level performance assessment of RIS-assisted networks and in particular downlink coverage probability and ergodic rate. To account for the inherent randomness in the spatial deployments of base stations (BSs) and RISs, we model the placements of the RISs as point processes (PPs) conditioned on the associated BSs, which are modeled by a Poisson point process (PPP). These RIS PPs can be adapted based on the deployment strategy. We focus on modeling the RISs as a Matérn cluster process (MCP), where each RIS cluster is a finite PPP with support a disc centered on the association BS. We assume that the system uses the orthogonal frequency division multiplexing (OFDM) technique to exploit the multipath diversity provided by RISs. The coverage probability and the ergodic rate can be evaluated when RISs operate as batched powerless beamformers. The resulting analytical expressions provide a general methodology to evaluate the impact of key RIS-related parameters, such as the batch size and the density of RISs, on system-level performance. To demonstrate the framework's broad applicability, we also analyze a RIS placement variant where RISs are deployed around coverage holes. Numerical evaluations of the analytical expressions and Monte-Carlo simulations jointly validate the proposed analytical approach and provide valuable insights into the design of future RIS-assisted cellular networks.

cs.IT↗
arXiv2026

DAAF: From Failure Localization to Editable System Assets in LLM Agents

Deployed LLM agents increasingly rely on persistent, versioned system assets such as routing rules, knowledge segments, prompt instructions, and reusable skills. Failure-localization methods can identify where an error manifests in an agent or execution trace, but repair requires a different decision: which editable system asset should be changed, and is that change expected to improve the task outcome? We study this gap through component-attribute failure attribution, where diagnosis targets versioned, addressable items rather than execution locations. We propose the Detection-Aware Attribution Framework (DAAF), which learns the effects of valid attribute replacements and amortizes this intervention evidence into deployment-time diagnosis. DAAF combines sparse and noisy failure signals to decide whether intervention is warranted, learns component-type-conditioned replacement effects from controlled replays evaluated by executable task outcomes, and shares supervision across requests with compatible intervention responses. At diagnosis time, DAAF uses only the observed execution, registered candidates, and available failure signals; it requires neither counterfactual replay nor task reward and returns no_change, a repair target, or an unresolved decision when evidence is insufficient. On held-out tau^2-bench Telecom tasks, DAAF achieves 80.72% attribute Hit@1, recovers 62.65% of failed executions while limiting clean-task regression to 3.23%, and reaches 71.93% overall task success. These results show that intervention-grounded attribute attribution can connect failure localization to executable system repair.

cs.AI↗
arXiv2026

State-Grounded Conditioning: Wrapping User-Facing LLM Agents Where Direction Depends on Live State

We introduce State-Grounded Conditioning (SGC), a design principle for user-facing LLM agents that must condition on live user state (game state, session history, live inventory), and a distinct failure class we call direction drift: task-complete responses whose chosen direction misaligns with the current state. SGC externalises state-dependent control into rule kernels over structured inputs and three primary state slices, via Perception, Grounding, and Interaction wrappers with explicit conditioning dependencies. We evaluate SGC on a 200-session anonymised benchmark ($\approx$1,000 assistant model turns) from an in-game conversational coaching agent that guides players through consecutive competitive matches, reporting mean first-token latency and five human-annotated dialogue-quality metrics that jointly cover factual grounding and coach-like guidance progression. The Perception wrapper holds mean first-token latency at 1.5s (vs. 6.1s for PE-Agent inside a production tool-use harness); enabling all three wrappers lifts turn-level grounded accuracy from 61.1%/69.8% (Prompting / PE-Agent) to 96.7% and session-level grounded accuracy from 20.0%/26.5% to 83.5%; session-level grounding-failure incidents drop by $\approx$78% relative to the strongest baseline. A cumulative ablation shows complementary incremental gains as the wrappers are added. These results inform approximate state-slice orthogonality, without establishing independent per-wrapper effects.

cs.AI↗
arXiv2026

High-quality electron beam generation from laser wakefield accelerators for driving compact free electron lasers

Despite the successful demonstration of compact free electron lasers (FELs) driven by laser wakefield accelerators (LWFAs), the pursuit of further enhancements in high-gain compact FELs presents a challenge due to the limitations in electron beam quality. In this work, we pinpoint the pivotal physics and optimization strategies for high-quality single-stage LWFAs that are crucial for high-gain FELs. We have delved into the synergistic injection mechanism, where the self-evolution injection threshold is far from reached at the injection position, with both the shock front and self-evolution of the laser playing a role in the injection process. A thorough discussion has been provided on the beam-quality degradation and optimization strategies, in terms of global (slice) energy spread and projected (slice) emittance. With the goal of achieving high-gain FELs driven by LWFAs, we have also explored the synthesis quality of the electron beam to determine an optimized power gain length. A comprehensive start-to-end simulation has been conducted, demonstrating the effectiveness of compact FELs powered by these high-quality electron beams. The resulting radiation reaches the saturation regime after a 4.5-meter-long undulator, with an energy of 17.4 μJ and a power of 6.0 GW at a wavelength of 23.9 nm. This proposed scheme offers not only a framework for optimizing beam quality in LWFAs, but also a promising path for future compact LWFA-driven FELs to achieve saturated regimes, opening up new possibilities for widespread applications.

physics.plasm-ph↗
arXiv2026

A Temporal Retrieval Method for Modulated Electron Bunches via Adaptive Kernel Reconstruction

Femtosecond electron beams with complex modulation play a crucial role in applications such as X-ray Free Electron Lasers (XFELs) and plasma wakefield accelerators. However, diagnostics for the electron beam current profile still face challenges with complex structure. In this letter, we propose a novel temporal retrieval algorithm for the coherent transition radiation (CTR) diagnostics of complex modulated electron beams. Starting from the time-frequency analysis of the electron bunch train, the algorithm separates and reconstructs the high- and low-frequency components. A temporal kernel was derived from the inverse sampling of the measured spectrum to construct the high-frequency component, while the low-frequency envelope was composed of several basis functions. Tested on the electron bunch trains from the complex multi-gaussian model and bunching-enhanced coherent harmonic generation, the algorithm successfully reconstructed the temporal signals and achieves better performance than the Kramers-Kronig method. This method is expected to crucial provide temporal evidence for potential electron beam modulation schemes, and will enable broad prospects for future applications.

physics.acc-ph↗
arXiv2026

SVOM Ground Support System

The Ground Support System (GSS) is a critical component of the Space-based multi-band astronomical Variable Objects Monitor (SVOM) mission's ground segment. Its main tasks include organizing and implementing the operations and management of the SVOM payloads, receiving scientific data, as well as preprocessing and managing the scientific data. To address the specific requirements of the SVOM mission, including specialized mission planning, data reception, status monitoring, and product processing, a dedicated SVOM Ground Support System has been designed based on the integrated architecture of the Chinese Space Science Satellite Project Ground Support System. This system has successfully supported SVOM's launch and on-orbit operations.

astro-ph.IM↗
arXiv2026

Poisson Hail on a Wireless Ground

This paper defines a new model which incorporates three key ingredients of a large class of wireless communication systems: (1) spatial interactions through interference, (2) dynamics of the queueing type, with users joining and leaving, and (3) carrier sensing and collision avoidance as used in, e.g., WiFi. In systems using (3), rather than directly accessing the shared resources upon arrival, a customer is considerate and waits to access them until nearby users in service have left. This new model can be seen as a missing piece of a larger puzzle that contains such dynamics as spatial birth-and-death processes, the Poisson-Hail model, and wireless dynamics as key other pieces. It is shown that, under natural assumptions, this model can be represented as a Markov process on the space of counting measures. The main results are then two-fold. The first is on the shape of the stability region and, more precisely, on the characterization of the critical value of the arrival rate that separates stability from instability. The second is of a more qualitative or perhaps even ethical nature. There is evidence that for natural values of the system parameters, the implementation of sensing and collision avoidance stabilizes a system that would be unstable if immediate access to the shared resources would be granted. In other words, for these parameters, renouncing greedy access makes sharing sustainable, whereas indulging in greedy access kills the system.

cs.IT↗
arXiv2026

Performance Guarantees of Cellular Networks with Hardcore Regulation and Scheduling

Providing performance guarantees is one of the critical objectives of recent and future communication networks, toward which regulations, i.e., constraints on key system parameters, have played an indispensable role. This is the case for large wireless communication networks, where spatial regulations (e.g., constraints on intercell distance) have recently been shown, through a spatial network calculus, to be essential for establishing provable wireless link-level guarantees. In this work, we focus on performance guarantees for the downlink of cellular networks where we impose a hardcore (spatial) regulation on base station (BS) locations and evaluate how BS scheduling (which controls which BSs can transmit at a given time) impacts performance. Hardcore regulation is the simplest form of spatial regulation that enforces a minimal distance between any pair of transmitters in the network. Within this framework of spatial network calculus, we first provide an upper bound on the power of total interference for a spatially regulated cellular network, and then, identify the regimes where scheduling BSs yields better link-level rate guarantees compared to scenarios where base stations are always active. The hexagonal cellular network is analyzed as a special case. The results offer insights into what spatial regulations are needed, when to choose scheduling, and how to potentially reduce the network power consumption to provide a certain target performance guarantee.

cs.IT↗
arXiv2026

Robustness Optimization for Compact Free-electron Laser Driven by Laser Wakefield Accelerators

Despite the successful demonstration of compact free electron lasers (FELs) driven by laser wakefield accelerators (LWFAs), the inherent shot-to-shot fluctuations in LWFAs, including both laser and plasma instabilities, remain a primary obstacle to realizing LWFA-driven FELs with robust operation. Here, we present a conceptual design for LWFA-driven FELs with sufficient tolerance against shot-to-shot fluctuations using the Covariance Matrix Adaptation Evolution Strategy (CMA-ES). Start-to-end simulations demonstrated that this systematic optimization resulted in a significant improvement in the robustness of FELs. With the optimized configurations, the radiation energy can be maintained above 1 microjoule at a wavelength of approximately 25 nm, even when accounting for twice the root-mean-square (RMS) ranges of these instabilities. This proposed scheme represents a substantial advancement in the development of compact LWFA-driven FEL systems, enabling robust operation and paving the way for the realization of reliable and widely accessible sources.

physics.plasm-ph↗
arXiv2026

TF-CoDiT: Conditional Time Series Synthesis with Diffusion Transformers for Treasury Futures

Diffusion Transformers (DiT) have achieved milestones in synthesizing financial time-series data, such as stock prices and order flows. However, their performance in synthesizing treasury futures data is still underexplored. This work emphasizes the characteristics of treasury futures data, including its low volume, market dependencies, and the grouped correlations among multivariables. To overcome these challenges, we propose TF-CoDiT, the first DiT framework for language-controlled treasury futures synthesis. To facilitate low-data learning, TF-CoDiT adapts the standard DiT by transforming multi-channel 1-D time series into Discrete Wavelet Transform (DWT) coefficient matrices. A U-shape VAE is proposed to encode cross-channel dependencies hierarchically into a latent variable and bridge the latent and DWT spaces through decoding, thereby enabling latent diffusion generation. To derive prompts that cover essential conditions, we introduce the Financial Market Attribute Protocol (FinMAP) - a multi-level description system that standardizes daily$/$periodical market dynamics by recognizing 17$/$23 economic indicators from 7/8 perspectives. In our experiments, we gather four types of treasury futures data covering the period from 2015 to 2025, and define data synthesis tasks with durations ranging from one week to four months. Extensive evaluations demonstrate that TF-CoDiT can produce highly authentic data with errors at most 0.433 (MSE) and 0.453 (MAE) to the ground-truth. Further studies evidence the robustness of TF-CoDiT across contracts and temporal horizons.

cs.LG↗
arXiv2025

Achieving High Efficiency And Enhanced Beam Quality In Laser Wakefield Acceleration

Laser wakefield acceleration, characterized by the extremely high electric field gradient exceeding 100GV/m, is regarded as a compact and cost affordable technology for the next generation of particle colliders and light sources. However, it has always been a major challenge to effectively increase the energy transfer efficiency from the laser to the accelerated beam, while ensuring the beam quality remains suitable for practical applications. This study demonstrates that the laser with shorter pulse duration allows for a two-step dechirping process of the accelerated electron beam with charge of nanocoulomb level. The electron beams with an energy spread of 1% can be generated with the energy transfer efficiency of 10% to 30% in a large parameter space. For example, one electron beam with the energy of 420MeV, the charge of 5.5nC and the RMS energy spread of 2% can be produced using an 8.3J laser pulse with 7.2fs duration.

physics.acc-ph↗
arXiv2025

Biological Regulatory Network Inference through Circular Causal Structure Learning

Biological networks are pivotal in deciphering the complexity and functionality of biological systems. Causal inference, which focuses on determining the directionality and strength of interactions between variables rather than merely relying on correlations, is considered a logical approach for inferring biological networks. Existing methods for causal structure inference typically assume that causal relationships between variables can be represented by directed acyclic graphs (DAGs). However, this assumption is at odds with the reality of widespread feedback loops in biological systems, making these methods unsuitable for direct use in biological network inference. In this study, we propose a new framework named SCALD (Structural CAusal model for Loop Diagram), which employs a nonlinear structure equation model and a stable feedback loop conditional constraint through continuous optimization to infer causal regulatory relationships under feedback loops. We observe that SCALD outperforms state-of-the-art methods in inferring both transcriptional regulatory networks and signaling transduction networks. SCALD has irreplaceable advantages in identifying feedback regulation. Through transcription factor (TF) perturbation data analysis, we further validate the accuracy and sensitivity of SCALD. Additionally, SCALD facilitates the discovery of previously unknown regulatory relationships, which we have subsequently confirmed through ChIP-seq data analysis. Furthermore, by utilizing SCALD, we infer the key driver genes that facilitate the transformation from colon inflammation to cancer by examining the dynamic changes within regulatory networks during the process.

q-bio.MN↗
arXiv2025

OIG-Bench: A Multi-Agent Annotated Benchmark for Multimodal One-Image Guides Understanding

Recent advances in Multimodal Large Language Models (MLLMs) have demonstrated impressive capabilities. However, evaluating their capacity for human-like understanding in One-Image Guides remains insufficiently explored. One-Image Guides are a visual format combining text, imagery, and symbols to present reorganized and structured information for easier comprehension, which are specifically designed for human viewing and inherently embody the characteristics of human perception and understanding. Here, we present OIG-Bench, a comprehensive benchmark focused on One-Image Guide understanding across diverse domains. To reduce the cost of manual annotation, we developed a semi-automated annotation pipeline in which multiple intelligent agents collaborate to generate preliminary image descriptions, assisting humans in constructing image-text pairs. With OIG-Bench, we have conducted a comprehensive evaluation of 29 state-of-the-art MLLMs, including both proprietary and open-source models. The results show that Qwen2.5-VL-72B performs the best among the evaluated models, with an overall accuracy of 77%. Nevertheless, all models exhibit notable weaknesses in semantic understanding and logical reasoning, indicating that current MLLMs still struggle to accurately interpret complex visual-text relationships. In addition, we also demonstrate that the proposed multi-agent annotation system outperforms all MLLMs in image captioning, highlighting its potential as both a high-quality image description generator and a valuable tool for future dataset construction. Datasets are available at https://github.com/XiejcSYSU/OIG-Bench.

cs.CV↗
arXiv2025

Collimated Hard X-Rays from Hybrid Laser and Plasma Wakefield Accelerators

We report a synergistic enhancement of betatron radiation based on the hybrid laser and plasma wakefield acceleration scheme. Quasi-phase-stable acceleration in an up-ramp plasma density first generates GeV-energy electron beams that act as a drive beam for PWFA, which then further accelerates the witness beam to GeV energies, enhancing both photon energy and flux. A full width at half maximum divergence $(6.1 \pm 1.9)\times(5.8\pm 1.6) $ mrad$^2$ of betatron radiation, a critical energy of $71 \pm 8$ keV, and an average flux of more than $10^{14}$ photons per steradian above 5 keV were all experimentally obtained thanks to this scheme, which was an order of magnitude higher than the previous reports. Quasi-three-dimensional particle-in-cell simulations were used to model the acceleration and radiation of the electrons in our experimental conditions, establishing a new paradigm for compact collimated hard X-ray sources.

physics.plasm-ph↗
arXiv2025

Spatial Network Calculus: Toward Deterministic Wireless Networking

This paper extends the classical network calculus to spatial scenarios, focusing on wireless networks with differentiated services and varying transmit power levels. Building on a spatial network calculus, a prior extension of network calculus to spatial settings, we propose a generalized framework by introducing regulations for stationary marked point processes. The regulations correspond to two key constraints: the total transmit power of all transmitters within a spatial region and the cumulative received power at a receiver, which we refer to as ball regulation and shot-noise regulation, respectively. Then we prove the equivalence of ball regulation and shot-noise regulation for stationary marked point processes and establish a universal lower bound on the performance of all network links under these constraints. This framework is applicable to diverse network scenarios, as demonstrated by the analysis of performance guarantees for networks with multi-class users. In addition, we propose an SINR-based power control scheme adapted to user traffic, which ensures differentiated quality of service (QoS) for different user classes.

cs.NI↗
arXiv2025

Hyperbolization and geometric decomposition of a class of 3-manifolds

Thurston's triangulation conjecture asserts that every hyperbolic 3-manifold admits a geometric triangulation into hyper-ideal hyperbolic tetrahedra. So far, this conjecture had only been proven for a few special 3-manifolds. In this article, we confirm this conjecture for a class of 3-manifolds. To be precise, let $M$ be an oriented compact 3-manifold with boundary, no component of which is a 2-sphere, and $\mathcal{T}$ is an ideal triangulation of $M$. If $\mathcal{T}$ satisfies properly gluing condition, and the valence is at least 6 at each ideal edge and 11 at each hyper-ideal edge, then $M$ admits an unique complete hyperbolic metric with totally geodesic boundary, so that $\mathcal{T}$ is isotopic to a geometric ideal triangulation of $M$. We use analytical tools such as combinatorial Ricci flow (CRF, abbr.) to derive the conclusions. There are intrinsic difficulties in dealing with CRF. First, the CRF may collapse in a finite time, second, most of the smooth curvature flow methods are no longer applicable since there is no local coordinates in $\mathcal{T}$, and third, the evolution of CRF is affected by certain combinatorial obstacles in addition to topology. To this end, we introduce the ideas as ``extending CRF", ``tetrahedral comparison principles", and ``control CRF with edge valence" to solve the above difficulties. In addition, the presence of torus boundary adds substantial difficulties in this article, which we have solved by introducing the properly gluing conditions on $\mathcal{T}$ and reducing the ECRF to a flow relatively easy to handle.

math.GT↗