arXiv subjects

Jiaqi Fu

Publications and source records attributed to Jiaqi Fu.

15 recordsLinked to original sources

arXiv2026

Don't Wait to Reply: Towards Responsive yet Thoughtful Dialogue through Proactive Thinking

Thinking has emerged as a critical capability for Large Language Models (LLMs) tackling complex tasks. However, its reactive nature, where reasoning is passively triggered only upon receiving a user response, inevitably introduces latency that compromises conversational fluidity. This stands in sharp contrast to human dialogue, where speakers proactively anticipate and plan future content during natural pauses to ensure seamless interaction. To bridge this gap, we propose Proactive Thinking, a framework that empowers models to pre-compute potential response elements during conversational downtime instead of waiting idly for the next input. We then introduce a training-free baseline that can think ahead by anticipating future states, balancing efficiency and quality through speculative continual thinking. To evaluate this approach in practice, we adapt three benchmarks of varying complexity into time-aware environments that simulate real-time conversational flow. We demonstrate that proactive thinking effectively improves interaction efficiency without compromising performance. Ultimately, this work advocates for a fundamental shift toward more intelligent, anticipatory, and real-time conversational AI.

cs.CL↗

arXiv2026

Bounds on the radius of outermost photon sphere and black hole shadow in $n$-dimensional Einstein gravity

The photon sphere, which determines the contour of the black hole shadow, provides a direct probe of strong-field gravity. In this work, we derive model-independent bounds on both the outermost photon sphere radius $r_{\gamma,\mathrm{out}}$ and the shadow radius $r_{\mathrm{sh}}$ for static, spherically symmetric, asymptotically flat black holes in $n$-dimensional ($n\ge 4$) Einstein gravity, supported by an anisotropic matter field. We first establish a rigorous upper bound on $r_{\gamma,\mathrm{out}}$ under the Weak Energy Condition (WEC), the Strong Energy Condition (SEC), and an asymptotic decay condition on the matter fields, proving $r_{\gamma,\mathrm{out}} \le \bigl[(n-1)M\bigr]^{\frac{1}{n-3}}$, where $M$ is the ADM mass, with the bound saturated by the vacuum Schwarzschild-Tangherlini black hole. We further discuss the lower bound on $r_{\gamma,\mathrm{out}}$, clarifying that a conditional bound can be obtained from the innermost photon sphere under an extra monotonicity assumption. Turning to the black hole shadow, we derive both upper and lower bounds on $r_{\mathrm{sh}}$. Making use of the photon sphere upper bound and the fact that, under the WEC and SEC, the effective potential for null geodesics outside the outermost photon sphere is bounded below by that of the vacuum solution with the same ADM mass, we obtain $r_{\mathrm{sh}}\le\sqrt{\frac{n-1}{n-3}}\,\bigl[(n-1)M\bigr]^{\frac{1}{n-3}}$, while for the lower bound, using only the WEC, we prove $r_{\mathrm{sh}} \ge \left( \frac{n-1}{2} \right)^{\frac{1}{n-3}} \sqrt{\frac{n-1}{n-3}}r_H$, where $r_H$ is the horizon radius. Our results provide clear geometric constraints on the observable signatures of higher-dimensional black holes and underline the special role of the shadow as a robust observable.

gr-qc↗

arXiv2026

From Bootstrapping to Sequence Modeling: A Unified Generative Framework for Personalized Landing-Page Modeling

Modern online platforms increasingly adopt multi-page architectures to accommodate diverse user needs. On these platforms, page navigation (the process of directing users to specific functional pages upon app entry) serves as a critical gateway that shapes user's first impression and significantly influences subsequent engagement. To optimize this process, Kuaishou formulated the task of Personalized Landing Page Modeling (PLPM) and proposed KLAN, a reinforcement learning framework built upon Conservative Q-Learning (CQL). However, CQL-based approaches suffer from two fundamental limitations: (1) the Markov assumption fails to capture the strong non-Markovian temporal dependencies inherent in real-world user behaviors, and (2) TD learning with bootstrapping incurs severe cumulative errors and credit assignment difficulties under delayed rewards, particularly in long-horizon settings where users enter the app multiple times daily. To address these limitations, we propose GLAN (Generative Landing-page Adaptive Navigator), a sequence modeling framework built on Decision Transformer to tackle PLPM from a unified global-local perspective. Specifically, GLAN incorporates two key modules. First, we design the L-RTG module that captures users' inter-day consumption dynamics to provide accurate global guidance for all page assignments within a day. Furthermore, we propose the HRM module that decomposes session-level feedback into fine-grained signals, enabling precise local supervision for each page assignment. Extensive online experiments conducted on the Kuaishou platform demonstrate the effectiveness of GLAN, achieving +0.158\% and +0.108\% improvements on Daily Active Users (DAU) and user Lifetime (LT) respectively.

cs.IR↗

arXiv2026

Video2Sim2Real: Full-Stack Autonomous Dexterous Skill Acquisition from a Single Human Video

Human manipulation videos are a convenient and intuitive source for robot learning. However, directly transferring human dexterity to robots remains challenging due to perception errors and embodiment gap. To address this, we introduce Video2Sim2Real, a full-stack framework for autonomous skill acquisition from a single human manipulation video. Our framework first uses off-the-shelf foundation models to reconstruct a simulator-ready digital twin and extract robot and object motion priors. Rather than treating the extracted robot motion as a reliable reference throughout execution, our key idea is to recover and leverage the most fundamental sources of supervision from the demonstrated skill: We identify object-centric keyframes to optimize the corresponding robot configurations using object information from the simulator, and use these configurations as anchors that refine the robot motion such that it ultimately has the desired impact on the environment. To bridge the remaining sim-to-real gap, we introduce a sim-to-real strategy that decouples robustness to noisy and incomplete perception from variations in hand-object interaction dynamics. Specifically, we learn to recalibrate robot configurations from noisy real-world point clouds via IL, and leverage residual RL to perform local finger-level adaptations to ensure for robust and effective interactions. Finally, a collision-aware motion planning module enables spatial generalization to novel object configurations. Across several everyday manipulation tasks, Video2Sim2Real improves simulated task success, safety, and trajectory coherence over numerous baselines, and achieves better sim-to-real transfer than existing techniques. These results demonstrate a promising path toward autonomous dexterous skill acquisition from human videos.

cs.RO↗

arXiv2026

Going with the Flow: Koopman Behavioral Models as Pseudo Planners for Visuo-Motor Dexterity

Contemporary visuo-motor dexterity models often rely on expressive policy classes with diffusion and transformer backbones to achieve strong performance. However, these architectures require significant data and computational resources, and remain far from reliable, particularly for multi-fingered dexterity. Importantly, they model skills as reactive mappings and rely on fixed-horizon action chunking, creating a rigid trade-off between temporal coherence and reactivity. To address these issues, we first introduce Unified Behavioral Models (UBMs), a framework to represent dexterous skills as coupled dynamical systems that capture how visual features of the environment (visual flow) and proprioceptive states of the robot (action flow) co-evolve. As such, UBMs ensure temporal coherence by construction rather than heuristic averaging. Unlike world models that attempt to predict the impact of arbitrary robot actions on the environment, UBMs target behavioral dynamics that encode how demonstrated robot behavior is related to desired impacts on the environment. A UBM can be viewed as a pseudo planner: given an initial condition, it computes the desired robot behavior over the entire skill horizon, while simultaneously ``imagining" the resulting flow of visual features. To operationalize UBMs, we propose Koopman-UBM, a first instantiation of UBMs as a structured latent linear system. K-UBM is computationally efficient, enabling reactivity and adaptation via an online replanning strategy: the model acts as its own runtime monitor, automatically triggering replanning when predicted and observed visual flow diverge beyond a threshold. Across seven simulated tasks and four real-world tasks, our approach matches or exceeds the performance of state-of-the-art baselines, while offering considerably faster inference, smooth execution, robustness to occlusions, and flexible replanning.

cs.RO↗

arXiv2026

A universal lower bound on the photon sphere radius in higher-dimensional black holes

The photon sphere, a hypersurface of circular null geodesics, plays a fundamental role in characterizing black hole spacetimes, influencing phenomena such as black hole shadows, gravitational lensing, and quasinormal modes. While universal upper bounds on the photon sphere radius have been established for both four-dimensional and higher-dimensional black holes, the question of a corresponding lower bound in higher-dimensional black holes remains less explored. In this work, we derive a universal lower bound for the photon sphere radius in static, spherically symmetric, asymptotically flat black hole spacetimes of arbitrary dimension $n\ge 4$. Under the assumptions of the weak energy condition, a non-positive trace of the energy-momentum tensor, and a monotonicity condition on the radial pressure function $|r^{n-1}p_r(r)|$, we prove that the photon sphere radius $r_\gamma$ satisfies $r_\gamma\ge (\frac{n-1}{2})^{1/(n-3)}r_H$, where $r_H$ is the radius of the outer event horizon. For $n=4$, this reduces to the known result $r_\gamma\ge \frac{3}{2}r_H$. Our result generalizes Hod's four-dimensional theorem to higher dimensions, and provides a new geometric constraint on the structure of black holes in extended theories of gravity.

gr-qc↗

arXiv2026

Bounds on the photon sphere radius for spherically symmetric black holes in n-dimensional Einstein gravity

The photon sphere, a hypersurface of null circular geodesics, plays a fundamental role in characterizing black hole spacetimes, influencing phenomena such as black hole shadows, gravitational lensing, and quasinormal modes. In this work, we derive both upper and lower bounds on the photon sphere radius for static, spherically symmetric, asymptotically flat black holes within $n$-dimensional Einstein gravity ($n\ge 4$), assuming an anisotropic matter field satisfying the weak energy condition and a non-positive trace of the energy-momentum tensor. For the upper bound, we obtain $r_\gamma\le [(n-1)M]^{\frac{1}{n-3}}$, where $M$ is the ADM mass. In the four-dimensional case ($n=4$), this reduces to $r_\gamma\le 3M$, in agreement with previous results. For the lower bound, under the additional assumption that $|r^{n-1}p_r(r)|$ is monotonically decreasing, we prove $r_\gamma\ge (\frac{n-1}{2})^{1/(n-3)}r_H$, where $r_H$ is the radius of the outer event horizon; for $n=4$ this gives $r_\gamma\ge \frac{3}{2}r_H$, also consistent with previous four-dimensional result. These results provide dimension-dependent geometric constraints that generalize well-known four-dimensional bounds to a specific class of higher-dimensional black holes (described by a Tangherlini-type metric) and deepen our understanding of spacetime structure in higher-dimensional gravitational theories.

gr-qc↗

arXiv2025

$(-1)$-Shifted Darboux theorem of derived schemes in characteristic $p>2$

The derived geometry approach to Donaldson--Thomas theory (over $\mathbb{C}$) is built on Pantev--To\"en--Vezzosi--Vaqui\'e's existence theorem of $(-1)$-shifted symplectic forms \cite{pantev2013shifted} and Brav--Bussi--Joyce's shifted Darboux theorem \cite{brav2019darboux}. In this paper, we prove a Darboux theorem in characteristic $p>2$ for the $(-1)$-shifted symplectic forms endowed with an \textit{infinitesimal structure}. A key ingredient is Antieau's derived infinitesimal cohomology \cite{antieau2025filtrations}, which enjoys a Poincar\'e-type lemma. Our argument is in fact characteristic-free and provides a conceptual understanding of the Brav--Bussi--Joyce theorem. Moreover, we extend the existence theorem of Pantev--To\"en--Vaqui\'e--Vezzosi by constructing a de Rham $(-1)$-shifted symplectic form on $\operatorname{Map}_k(X,\underline{\operatorname{Perf}})$, where $X$ is a Calabi--Yau $3$-fold over a field $k$ in characteristic $p>2$. We conjecture that this $(-1)$-shifted symplectic form admits an infinitesimal structure.

math.AG↗

arXiv2025

Robust topological invariants of timelike circular orbits for spinning test particles in black hole spacetimes

The spin-curvature coupling in the Mathisson-Papapetrou-Dixon (MPD) formalism induces non-geodesic motion, shifting the orbital parameters of spinning test particles in black hole spacetimes. We investigate whether these quantitative shifts alter the qualitative, global structure of the orbit manifold. Using a topological approach, we study timelike circular orbits (TCOs) for spinning particles in static, spherically symmetric spacetimes. By constructing an auxiliary vector field, we compute the topological winding number $W$ in horizon-bounded regions of asymptotically flat, anti-de Sitter (AdS), and de Sitter (dS) backgrounds. We find that $W$ is robust against both the magnitude and direction of the particle's spin: between two horizons, $W = -1$, guaranteeing at least one unstable TCO; outside the outermost horizon in asymptotically flat and AdS spacetimes, $W = 0$, enforcing that TCOs must appear in stable-unstable pairs or be absent. This spin independence reveals that the fundamental orbital structure is a property of spacetime geometry itself, not of the particle's spin. We validate this with quantitative examples in Schwarzschild, Schwarzschild-AdS, and Schwarzschild-dS spacetimes, showing explicit spin-induced TCO shifts while confirming the invariant topology. This result provides a topological foundation for interpreting gravitational waveforms from extreme mass-ratio inspirals involving spinning secondaries.

gr-qc↗

arXiv2025

MobileRL: Online Agentic Reinforcement Learning for Mobile GUI Agents

Building general-purpose graphical user interface (GUI) agents has become increasingly promising with the progress in vision language models. However, developing effective mobile GUI agents with reinforcement learning (RL) remains challenging due to the heavy-tailed distribution of task difficulty and the inefficiency of large-scale environment sampling. We present an online agentic reinforcement learning framework MobileRL to enhance GUI agents in mobile environments. Its core component is the Difficulty-ADAptive GRPO (ADAGRPO) algorithm. In ADAGRPO, we design difficulty-adaptive positive replay and failure curriculum filtering to adapt the model to different task difficulties. We introduce the shortest-path reward adjustment strategy to reshape rewards concerning the task length in multi-turn agentic tasks. Those strategies jointly stabilize RL training, improve sample efficiency, and generate strong performance across diverse mobile apps and tasks. We apply MOBILERL to two open models (Qwen2.5-VL-7B-Instruct and GLM-4.1V-9B-Base). The resultant MOBILERL-9B model achieves state-of-the-art results in terms of success rates on both AndroidWorld (80.2%) and AndroidLab (53.6%). The MOBILERL framework is open-sourced at: https://github.com/THUDM/MobileRL.

cs.LG↗

arXiv2025

The existence and upper bound for stable photon spheres in static spherically symmetric black holes

In this work, we establish the existence conditions and a universal upper bound on the radius of stable photon spheres (SPS) outside the event horizons of static, spherically symmetric, asymptotically flat black holes surrounded by matter fields. We prove that stable photon spheres exist if the external matter satisfies specific conditions. Furthermore, under the additional assumption of a monotonically decreasing mass-radius ratio $m(r)/r^3$ outside the horizon, we derive a strict upper bound on the radius $r_{\mathrm{sps}}$ of any stable photon sphere: $r_{\mathrm{sps}}<6M$, where $M$ is the asymptotic mass of the black hole. This bound is independent of specific black hole solutions and broadly applies to hairy black holes and other configurations with external matter fields meeting the stated energy conditions. Our results resolve fundamental questions regarding the existence and spatial constraints on stable photon orbits, with implications for gravitational lensing, accretion disk dynamics (e.g., the Aschenbach effect), and black hole shadow observations.

gr-qc↗

arXiv2025

KLAN: Kuaishou Landing-page Adaptive Navigator

Modern online platforms configure multiple pages to accommodate diverse user needs. This multi-page architecture inherently establishes a two-stage interaction paradigm between the user and the platform: (1) Stage I: page navigation, navigating users to a specific page and (2) Stage II: in-page interaction, where users engage with customized content within the specific page. While the majority of research has been focusing on the sequential recommendation task that improves users' feedback in Stage II, there has been little investigation on how to achieve better page navigation in Stage I. To fill this gap, we formally define the task of Personalized Landing Page Modeling (PLPM) into the field of recommender systems: Given a user upon app entry, the goal of PLPM is to proactively select the most suitable landing page from a set of candidates (e.g., functional tabs, content channels, or aggregation pages) to optimize the short-term PDR metric and the long-term user engagement and satisfaction metrics, while adhering to industrial constraints. Additionally, we propose KLAN (Kuaishou Landing-page Adaptive Navigator), a hierarchical solution framework designed to provide personalized landing pages under the formulation of PLPM. KLAN comprises three key components: (1) KLAN-ISP captures inter-day static page preference; (2) KLAN-IIT captures intra-day dynamic interest transitions and (3) KLAN-AM adaptively integrates both components for optimal navigation decisions. Extensive online experiments conducted on the Kuaishou platform demonstrate the effectiveness of KLAN, obtaining +0.205% and +0.192% improvements on in Daily Active Users (DAU) and user Lifetime (LT). Our KLAN is ultimately deployed on the online platform at full traffic, serving hundreds of millions of users. To promote further research in this important area, we will release our dataset and code upon paper acceptance.

cs.IR↗

arXiv2025

Design of a Formation Control System to Assist Human Operators in Flying a Swarm of Robotic Blimps

Formation control is essential for swarm robotics, enabling coordinated behavior in complex environments. In this paper, we introduce a novel formation control system for an indoor blimp swarm using a specialized leader-follower approach enhanced with a dynamic leader-switching mechanism. This strategy allows any blimp to take on the leader role, distributing maneuvering demands across the swarm and enhancing overall formation stability. Only the leader blimp is manually controlled by a human operator, while follower blimps use onboard monocular cameras and a laser altimeter for relative position and altitude estimation. A leader-switching scheme is proposed to assist the human operator to maintain stability of the swarm, especially when a sharp turn is performed. Experimental results confirm that the leader-switching mechanism effectively maintains stable formations and adapts to dynamic indoor environments while assisting human operator.

cs.RO↗

arXiv2024

A duality between Lie algebroids and infinitesimal foliations

There are two natural analogues of algebraic foliations in derived algebraic geometry, called partition Lie algebroids and infinitesimal derived foliations, and both make sense in general characteristics. We construct an equivalence between these two notions under some finiteness conditions. Our method is refining the PD Koszul duality in \cite{BM}\cite{BCN} using the (completed) Hodge filtration.

math.AG↗

arXiv2024

MineLand: Simulating Large-Scale Multi-Agent Interactions with Limited Multimodal Senses and Physical Needs

While Vision-Language Models (VLMs) hold promise for tasks requiring extensive collaboration, traditional multi-agent simulators have facilitated rich explorations of an interactive artificial society that reflects collective behavior. However, these existing simulators face significant limitations. Firstly, they struggle with handling large numbers of agents due to high resource demands. Secondly, they often assume agents possess perfect information and limitless capabilities, hindering the ecological validity of simulated social interactions. To bridge this gap, we propose a multi-agent Minecraft simulator, MineLand, that bridges this gap by introducing three key features: large-scale scalability, limited multimodal senses, and physical needs. Our simulator supports 64 or more agents. Agents have limited visual, auditory, and environmental awareness, forcing them to actively communicate and collaborate to fulfill physical needs like food and resources. Additionally, we further introduce an AI agent framework, Alex, inspired by multitasking theory, enabling agents to handle intricate coordination and scheduling. Our experiments demonstrate that the simulator, the corresponding benchmark, and the AI agent framework contribute to more ecological and nuanced collective behavior.The source code of MineLand and Alex is openly available at https://github.com/cocacola-lab/MineLand.

cs.CL↗