Search arXiv⌕ Search

arXiv subjects

Bowen Li

Publications and source records attributed to Bowen Li.

At least 55 records · Page 3Linked to original sources
arXiv2026

Space-Time Log-Sobolev Inequality and Hypocoercive Hypercontractivity for Underdamped Langevin Dynamics

We study hypercontractivity for underdamped Langevin dynamics with a convex confining potential whose spatial Gibbs marginal satisfies a logarithmic Sobolev inequality (LSI). Unlike in the overdamped case, the noise acts only on the velocity variable, so the usual LSI-based argument does not apply. Nevertheless, under an additional tame Hessian-growth assumption on the potential, we prove that, when the spatial LSI constant is $ρ$ and the friction parameter is of order $\sqrtρ$, the semigroup satisfies a Gross-type $L^p$-to-$L^q$ estimate in which the integrability exponent grows exponentially on the kinetic time scale $t \sim ρ^{-1/2}$. The central ingredient, of independent interest, is a space-time LSI, valid for controlled kinetic paths of finite action, which quantifies how dissipation in the velocity variable is transferred to the position variable. We derive it from a controlled version of the hypocoercive entropy-decay estimate, and we then restore the Gross hypercontractivity mechanism through a duality argument built on a non-reversible forward/backward interpolation of the underdamped Langevin semigroup. As a corollary, Rényi divergences of every order decay exponentially at the sharp hypocoercive rate $\mathcal{O}(\sqrtρ)$.

math.AP↗
arXiv2026

Loong: Synthesize Long Chain-of-Thoughts at Scale through Verifiers

Recent advances in Large Language Models (LLMs) have shown that their reasoning capabilities can be significantly improved through Reinforcement Learning with Verifiable Reward (RLVR), particularly in domains like mathematics and programming, where ground-truth correctness can be automatically evaluated. However, extending this success to other reasoning-intensive domains remains challenging due to the scarcity of high-quality, verifiable datasets and the high cost of human supervision. In this work, we introduce the Loong Project: an open-source framework for scalable synthetic data generation and verification across a diverse range of reasoning-intensive domains. The framework consists of two key components: (1) LoongBench, a curated seed dataset containing 8,729 human-vetted examples across 12 domains (e.g., Advanced Mathematics, Chemistry, Logic), each paired with executable code and rich metadata; and (2) LoongEnv, a modular synthetic data generation environment that supports multiple prompting strategies to produce new question-answer-code triples. Together, these components form an agent-environment loop that enables reinforcement learning, where an LLM-based agent is rewarded for generating Chain-of-Thought (CoT) solutions that align with code-executed answers. Empirically, we benchmark LoongBench on a broad suite of both open-source and proprietary LLMs to evaluate domain coverage and reveal performance bottlenecks. In addition, we conduct a comprehensive analysis of synthetic data generated by LoongEnv, examining correctness, difficulty, and diversity. Code and documentation are available at https://github.com/camel-ai/loong.

cs.LG↗
arXiv2026

The Missing Adapter Layer for Research Computing

Higher Degree by Research (HDR) candidates increasingly depend on cloud-provisioned virtual machines and local GPU hardware for their computational experiments, yet a persistent and under-addressed gap separates having compute resources from using them productively. Cloud and infrastructure teams can provision a virtual machine in minutes, but the path from a raw VM to a reproducible, GPU-ready research environment remains a significant barrier for researchers who are domain experts, not systems engineers. We argue that this gap is not a shortcoming of any particular tool but a missing architectural layer: an adapter layer that bridges cloud provisioning and interactive research work. We present a lightweight, open-source implementation of this layer, built on k3s and Coder and already in active use in our research workspace environment. A CI/CD pipeline connects GitHub directly to the local cluster, carrying a research project from commit to a running, accessible workspace in under five minutes. We then define a concrete metrics framework for evaluating any adapter layer covering deployment latency, environment reproducibility, onboarding friction, and resource utilisation and establish baselines against which improvements can be measured.

cs.CE↗
arXiv2026

VL-Nav: Neuro-Symbolic Reasoning-based Vision-Language Navigation

Navigating unseen, large-scale environments based on complex and abstract human instructions remains a formidable challenge for autonomous mobile robots. Addressing this requires robots to infer implicit semantics and efficiently explore large-scale task spaces. However, existing methods, ranging from end-to-end learning to foundation model-based modular architectures, often lack the capability to decompose complex tasks or employ efficient exploration strategies, leading to robot aimless wandering or target recognition failures. To address these limitations, we propose VL-Nav, a neuro-symbolic (NeSy) vision-language navigation system. The proposed system intertwines neural reasoning with symbolic guidance through two core components: (1) a NeSy task planner that leverages a symbolic 3D scene graph and image memory system to enhance the vision language models' (VLMs) neural reasoning capabilities for task decomposition and replanning; and (2) a NeSy exploration system that couples neural semantic cues with the symbolic heuristic function to efficiently gather the task-related information while minimizing unnecessary repeat travel during exploration. Validated on the DARPA TIAMAT Challenge navigation tasks, our system achieved an 83.4% success rate (SR) in indoor environments and 75% in outdoor scenarios. VL-Nav achieved an 86.3% SR in real-world experiments, including a challenging 483-meter run. Finally, we validate the system with complex instructions in a 3D multi-floor scenario.

cs.RO↗
arXiv2026

Neuro-Symbolic Learning for Long-Horizon Task Planning Under Complex Logical Constraints

Task planning often suffers from severe efficiency bottlenecks when robots must reason over long-horizon action sequences under complex logical constraints, including object affordances, spatial relationships, and sequential action dependencies. Recent neuro-symbolic methods improve planning efficiency by learning object-importance scores to prune task-irrelevant objects, but they typically rely on fixed offline supervision generated from full search spaces. This creates a train-test mismatch: at deployment, the planner operates in pruned search spaces induced by the model's own imperfect predictions, leading to exposure bias and degraded planning performance. To address this challenge, we formulate object-importance learning for task planning as an imperative learning-based bilevel optimization problem. The upper level optimizes a neural scorer, while the lower level solves a symbolic planning problem in the score-pruned search space. To stabilize this learning process, we introduce a 3R strategy into the lower-level planning, using parallel Repair, Restart, and Rollback recovery to provide reliable and adaptive feedback for upper-level learning. Experiments on three challenging benchmarks demonstrate state-of-the-art performance, including an 80.04% reduction in failure rate and a 57.14% reduction in planning time. We further validate the framework on a quadruped-based mobile manipulator in simulation and the real world, demonstrating its potential for efficient and deployable neuro-symbolic task planning.

cs.RO↗
arXiv2026

Passively synchronized dual-color mode-locked fiber lasers based on nonlinear amplifying loop mirrors

We have proposed and implemented a novel scheme for passive all-optical synchronization between erbium and ytterbium mode-locked fiber lasers. The passive locking of repetition rates for the dual-color pulses was realized by cross-phase modulation within phase-biased nonlinear amplifying loop mirrors. In contrast to previous demonstrations, the synchronization system was configured in an all-polarization-maintaining structure, thus gaining substantially improved stability and robustness. Consequently, the maximum tolerance of cavity-length mismatch of 16.2 mm was achieved unprecedentedly, which was at least one order of magnitude longer than previously reported results for comparable temporal durations of involved pulses. The corresponding relative timing jitter was measured to be 31 fs within 1-MHz bandwidth. Such tight and robust synchronization fiber laser system offers a great potential for various applications, such as pump-probe microscopy, Raman scattering spectroscopy and nonlinear frequency generation.

physics.optics↗
arXiv2026

Coincidence-pumping upconversion detector based on passively synchronized fiber laser system

We experimentally demonstrated a high-performance frequency upconversion detector for telecom-band photons based on a passively synchronized fiber laser system. The involved coincidence pumping technique enabled to spectrally convert the pulsed infrared photons into the visible regime with a conversion efficiency of 72\%. The overall detection efficiency of the upconversion detector reached to 30\% with a low noise equivalent power of $3\times10^{-17}\ \text{W/Hz}^{1/2}$. In contrast to previous demonstrations, the whole upconversion detection system was constructed in an all-polarization-maintaining fiber structure, thus favoring substantial improvement of compactness and robustness. Moreover, the long-term stability was manifested by at least ten-hour operation with a relative fluctuation of count rates as small as 0.26\%. The achieved features here would be desirable in many practical applications requiring efficient and robust coherent manipulation of pulsed optical fields by nonlinear frequency conversion.

physics.optics↗
arXiv2026

Thermal expectation estimation via single-trajectory Gibbs sampling with non-destructive measurements

Estimating thermal expectation values of quantum many-body systems is a central challenge in physics, chemistry, and materials science. Standard quantum Gibbs sampling protocols address this task by preparing the Gibbs state from scratch after every measurement, incurring a full mixing time cost at each step. Recent advances in single-trajectory Gibbs sampling [Jiang et al. 2026] substantially reduce this overhead: once stationarity is reached, measurements can be collected along a single trajectory without re-thermalizing, provided the measurement channel preserves the Gibbs ensemble. However, explicit constructions of such non-destructive measurements have been limited primarily to observables that commute with the Hamiltonian. In this work, we fundamentally extend the single-trajectory framework to arbitrary, non-commuting observables. We provide two measurement constructions that extract measurement information without fully destroying the Gibbs state, thereby eliminating the need for full re-mixing between samples. First, we construct a measurement that satisfies exact detailed balance. This ensures the system remains in equilibrium throughout the trajectory, allowing measurement outcomes to decorrelate in an autocorrelation time that could be significantly shorter than the global mixing time. Second, assuming the underlying quantum Gibbs sampler has a positive spectral gap, we design a simplified measurement scheme that ensures the post-selected state serves as a warm start for rapid re-mixing. This approach successfully decouples the resampling cost from the global mixing time. Both measurement schemes admit efficient quantum circuit implementations, requiring only polylogarithmic Hamiltonian simulation time.

quant-ph↗
arXiv2026

PilotTTS: A Disciplined Modular Recipe for Competitive Speech Synthesis

Building state-of-the-art text-to-speech (TTS) systems typically demands millions of hours of proprietary data and complex multi-stage architectures, creating substantial barriers for resource-constrained research teams. In this report, we present PilotTTS, a lightweight autoregressive TTS system that achieves competitive performance through minimalist architecture and rigorous data engineering. PilotTTS is trained on only 200K hours of data processed entirely with open-source tools. Specifically, our contributions are: (1) a reproducible multi-stage data processing pipeline covering quality assessment, label annotation, and filtering, and (2) a compact model architecture that employs Q-Former-based conditioning to decouple speaker identity from speaking style via cross-sample paired training. Within a unified framework, PilotTTS supports zero-shot voice cloning, emotion synthesis (11 categories), paralinguistic synthesis (4 categories), and Chinese dialect synthesis (14 dialects). On the Seed-TTS Eval benchmark, PilotTTS achieves the lowest WER of 1.50% on test-en, a CER of 0.87% on test-zh, and the highest speaker similarity on both test sets (0.862 and 0.815), outperforming systems trained on significantly larger datasets. We release the complete data pipeline recipe, pretrained weights, and code at https://github.com/AMAPVOICE/PilotTTS.

cs.SD↗
arXiv2026

Joint CFO-Channel Estimation under Strong Inter-Cell Interference for Low-Altitude Radio Mapping

Extending terrestrial networks into low-altitude airspace is a practical way to support aerial services, and accurate low-altitude radio maps are essential for characterizing terrestrial base station (BS) coverage and guiding system design. This work targets per-cell per-beam radio mapping from 5G new radio (NR) synchronization signal (SS) burst sets. Conventional processing treats interference as noise and focuses on the strongest link, which is insufficient to comprehensive awareness of the radio environment and ineffective in dense multi-cell low-altitude scenarios. We propose a successive waveform reconstruction and cancellation framework that iteratively estimates, reconstructs, and subtracts the SSs of stronger BSs, thereby enabling reliable detection and estimation of ultra-weak signals. To support this, we introduce the notion of a carrier frequency offset (CFO)-coherent block within which a common-CFO/per-synchronization signal block (SSB)-channel model holds and design a joint CFO-channel estimator that coherently aggregates multiple SSBs within each CFO-coherent block. We further derive closed-form scaling laws that relate estimation accuracy to unmanned aerial vehicle (UAV) speed, motion geometry, burst periodicity, and the length of the CFO-coherent block. Simulations show that the proposed framework can detect and estimate SSs at signal-to-interference-and-noise ratio (SINR) levels down to -30 dB. Field tests at 150 m altitude demonstrate per-beam coverage maps for more than ten overlapping BSs and reveal that, despite strong received power, the measured SINR rarely exceeds 10 dB, underscoring the need for careful interference management in low-altitude airspace.

eess.SP↗
arXiv2026

A Performance Bound for the Greedy Algorithm in a Generalized Class of String Optimization Problems

We present a simple performance bound for the greedy scheme in string optimization problems that obtains strong results. Our approach vastly generalizes the group of previously established greedy curvature bounds by Conforti and Cornuéjols (1984). We consider three constants, $α_G$, $α_G'$, and $α_G''$ introduced by Conforti and Cornuéjols (1984), that are used in performance bounds of greedy schemes in submodular set optimization. We first generalize both of the $α_G$ and $α_G''$ bounds to string optimization problems in a manner that includes maximizing submodular set functions over matroids as a special case. We then derive a much simpler and computable bound that allows for applications to a far more general class of functions with string domains. We prove that our bound is superior to both the $α_G$ and $α_G''$ bounds and provide a counterexample to show that the $α_G'$ bound is incorrect under the assumptions in Conforti and Cornuéjols (1984). We conclude with two applications. The first is an application of our result to sensor coverage problems. We demonstrate our performance bound in cases where the objective function is set submodular and string submodular. The second is an application to a social welfare maximization problem with black-box utility functions.

eess.SY↗
arXiv2026

KinDER: A Physical Reasoning Benchmark for Robot Learning and Planning

Robotic systems that interact with the physical world must reason about kinematic and dynamic constraints imposed by their own embodiment, their environment, and the task at hand. We introduce KinDER, a benchmark for Kinematic and Dynamic Embodied Reasoning that targets physical reasoning challenges arising in robot learning and planning. KinDER comprises 25 procedurally generated environments, a Gymnasium-compatible Python library with parameterized skills and demonstrations, and a standardized evaluation suite with 13 implemented baselines spanning task and motion planning, imitation learning, reinforcement learning, and foundation-model-based approaches. The environments are designed to isolate five core physical reasoning challenges: basic spatial relations, nonprehensile multi-object manipulation, tool use, combinatorial geometric constraints, and dynamic constraints, disentangled from perception, language understanding, and application-specific complexity. Empirical evaluation shows that existing methods struggle to solve many of the environments, indicating substantial gaps in current approaches to physical reasoning. We additionally include real-to-sim-to-real experiments on a mobile manipulator to assess the correspondence between simulation and real-world physical interaction. KinDER is fully open-sourced and intended to enable systematic comparison across diverse paradigms for advancing physical reasoning in robotics. Website and code: https://prpl-group.com/kinder-site/

cs.RO↗
arXiv2026

Kernel-Smith: A Unified Recipe for Evolutionary Kernel Optimization

We present Kernel-Smith, a framework for high-performance GPU kernel and operator generation that combines a stable evaluation-driven evolutionary agent with an evolution-oriented post-training recipe. On the agent side, Kernel-Smith maintains a population of executable candidates and iteratively improves them using an archive of top-performing and diverse programs together with structured execution feedback on compilation, correctness, and speedup. To make this search reliable, we build backend-specific evaluation services for Triton on NVIDIA GPUs and Maca on MetaX GPUs. On the training side, we convert long-horizon evolution trajectories into step-centric supervision and reinforcement learning signals by retaining correctness-preserving, high-gain revisions, so that the model is optimized as a strong local improver inside the evolutionary loop rather than as a one-shot generator. Under a unified evolutionary protocol, Kernel-Smith-235B-RL achieves state-of-the-art overall performance on KernelBench with Nvidia Triton backend, attaining the best average speedup ratio and outperforming frontier proprietary models including Gemini-3.0-pro and Claude-4.6-opus. We further validate the framework on the MetaX MACA backend, where our Kernel-Smith-MACA-30B surpasses large-scale counterparts such as DeepSeek-V3.2-think and Qwen3-235B-2507-think, highlighting potential for seamless adaptation across heterogeneous platforms. Beyond benchmark results, the same workflow produces upstream contributions to production systems including SGLang and LMDeploy, demonstrating that LLM-driven kernel optimization can transfer from controlled evaluation to practical deployment.

cs.CL↗
arXiv2026

TREX: Automating LLM Fine-tuning via Agent-Driven Tree-based Exploration

While Large Language Models (LLMs) have empowered AI research agents to perform isolated scientific tasks, automating complex, real-world workflows, such as LLM training, remains a significant challenge. In this paper, we introduce TREX, a multi-agent system that automates the entire LLM training life-cycle. By orchestrating collaboration between two core modules-the Researcher and the Executor-the system seamlessly performs requirement analysis, open-domain literature and data research, formulation of training strategies, preparation of data recipes, and model training and evaluation. The multi-round experimental process is modeled as a search tree, enabling the system to efficiently plan exploration paths, reuse historical results, and distill high-level insights from iterative trials. To evaluate the capability of automated LLM training, we construct FT-Bench, a benchmark comprising 10 tasks derived from real-world scenarios, ranging from optimizing fundamental model capabilities to enhancing performance on domain-specific tasks. Experimental results demonstrate that the TREX agent consistently optimizes model performance on target tasks.

cs.AI↗
arXiv2026

Beyond Rating: A Comprehensive Evaluation and Benchmark for AI Reviews

The rapid adoption of Large Language Models (LLMs) has spurred interest in automated peer review; however, progress is currently stifled by benchmarks that treat reviewing primarily as a rating prediction task. We argue that the utility of a review lies in its textual justification--its arguments, questions, and critique--rather than a scalar score. To address this, we introduce Beyond Rating, a holistic evaluation framework that assesses AI reviewers across five dimensions: Content Faithfulness, Argumentative Alignment, Focus Consistency, Question Constructiveness, and AI-Likelihood. Notably, we propose a Max-Recall strategy to accommodate valid expert disagreement and introduce a curated dataset of paper with high-confidence reviews, rigorously filtered to remove procedural noise. Extensive experiments demonstrate that while traditional n-gram metrics fail to reflect human preferences, our proposed text-centric metrics--particularly the recall of weakness arguments--correlate strongly with rating accuracy. These findings establish that aligning AI critique focus with human experts is a prerequisite for reliable automated scoring, offering a robust standard for future research.

cs.CL↗
arXiv2026

A Digital Pathology Resource for Liver Cancer Quantification with Datasets, Benchmarks, and Tools

Liver cancer, especially hepatocellular carcinoma (HCC), imposes a substantial global disease burden. Accurate diagnosis and prognostic assessment directly influence treatment selection and patient survival, and pathological examination remains the gold standard for liver cancer diagnosis. Identifying diverse tissue components and pathological subtypes on histopathology slides is crucial for estimating postoperative recurrence risk and overall prognosis. However, most publicly available resources are still provided at the whole-slide image (WSI) level, and well-annotated datasets for fine-grained tissue component identification in liver cancer are scarce, which hinders reproducible model development and the deployment of quantitative analysis tools. To address this gap, we release HepatoBench, a patch-level image database for liver cancer with annotations for seven key tissue categories. Based on HepatoBench, we train and open-source a deep learning classification model as a tissue recognition tool. Furthermore, we train a WSI-level tumor/non-tumor segmentation model to automatically localize lesion regions across entire slides. By integrating the patch-level tissue classifier with the WSI-level segmentation model, we build HepatoQuant, an end-to-end, disease-specific regional quantification tool for liver cancer, enabling a unified workflow from WSIs to tissue composition parsing and quantitative statistics. We also open-source HepatoBench, the benchmarking protocol, and supporting tools, providing a solid foundation for automated regional quantification and fair method comparison in liver cancer pathology.

cs.CV↗
arXiv2026

How to Fine-Tune a Reasoning Model? A Teacher-Student Cooperation Framework to Synthesize Student-Consistent SFT Data

A widely adopted strategy for model enhancement is to use synthetic data generated by a stronger model for supervised fine-tuning (SFT). However, for emerging reasoning models like Qwen3-8B, this approach often fails to improve reasoning capabilities and can even lead to a substantial drop in performance. In this work, we identify substantial stylistic divergence between teacher generated data and the distribution of student as a major factor impacting SFT. To bridge this gap, we propose a Teacher-Student Cooperation Data Synthesis framework (TESSY), which interleaves teacher and student models to alternately generate style and non-style tokens. Consequently, TESSY produces synthetic sequences that inherit the advanced reasoning capabilities of the teacher while maintaining stylistic consistency with the distribution of the student. In experiments on code generation using GPT-OSS-120B as the teacher, fine-tuning Qwen3-8B on teacher-generated data leads to performance drops of 3.25% on LiveCodeBench-Pro and 10.02% on OJBench, whereas TESSY achieves improvements of 11.25% and 6.68%.

cs.CL↗
arXiv2026

Forbidding Exactly One Hamming Distance

Addressing questions raised in recent papers, we study the $r$-distance graph $H_r(n)$ on the Boolean cube $\{0,1\}^n$, where two vertices are adjacent if their Hamming distance is exactly $r$. For fixed integers $s \ge 2$ and even $r \ge 2$, we determine the asymptotic order of the $s$-independence number $α_s(H_r(n))$, showing that \[ α_s\left(H_r(n)\right)=Θ\left(\frac{2^n}{n^{r/2}}\right). \] The upper bound is derived via a reduction to extremal problems for sunflower-free set systems, while the lower bound is obtained using algebraic constructions based on BCH codes and constant-weight codes.

math.CO↗