arXiv subjects

Bowen Li

Publications and source records attributed to Bowen Li.

At least 19 recordsLinked to original sources

arXiv2026

HEART-Bench: Do LLMs Make Persona-Consistent Decisions?

LLM agents in interactive entertainment and social applications, including companions, role-play characters, and persona clones, need decisions grounded in personality and past experiences. We introduce HEART-Bench to evaluate memory-grounded persona-consistent response selection. The benchmark combines 11 Big Five-anchored synthetic characters, each with 1,000 autobiographical memories, and 64 DIAMONDS-guided decision scenarios into 673 multiple-choice questions. Domain experts independently compare candidate responses, with a senior expert resolving disagreements and revising responses when needed; other characters' adjudicated responses to the same scenario serve as distractors. In the main evaluation, models receive character IDs and occupations, scenarios, and options, with retrieved memories added in memory-augmented settings. Detailed personality-and-value profiles are withheld. We evaluate 16 LLMs under No-Retrieval, Naive-RAG, Mem0, and PersonaDB. Accuracy without memories stays near the 25% chance level, and the best memory-augmented configuration reaches 64.9%. On a matched 150-item subset, human annotators reach 80.8%, compared with 64.0% for the best model. Memory-content ablations show the importance of retrieving experiences that match both the target character and the decision context.

cs.CL↗

arXiv2026

Why Are LLMs Confidently Wrong? Correcting Overconfident Errors via Causal Head Intervention

Modern large language models (LLMs) often exhibit overconfident errors, assigning high confidence to incorrect predictions and thereby undermining reliability in real-world use. Existing calibration methods typically rely on post-hoc adjustment or retraining, operating at the output level without addressing the underlying causes of miscalibration. We propose CoCHI, (Class-oriented Causal Head Intervention), a framework that improves calibration by identifying and intervening on internal components responsible for distinct confidence behaviours. By analysing predictions through correctness-confidence patterns, CoCHI enables targeted intervention on model internals, reducing overconfident errors while preserving correct predictions. Our approach provides a mechanistic perspective on calibration, linking reliability to identifiable structures within the model rather than treating it as an output-level issue. Experiments across multiple LLMs show consistent improvements in calibration metrics, including Expected Calibration Error, Negative Log-Likelihood, and Brier Score, while maintaining accuracy on five multiple-choice question answering benchmarks. These results suggest that effective calibration requires not only adjusting outputs, but also addressing the underlying causal mechanisms within the model.

cs.AI↗

arXiv2026

Entropic hypocoercivity via the Polchinski flow: an application to the lattice sine-Gordon model

We develop a multiscale criterion for exponential decay of relative entropy in the canonical stochastic quantization of lattice field theories, the underdamped Langevin dynamics of the field and its momentum. For smooth bounded perturbations of Gaussian fields, the criterion requires the uniform operator-norm bound on the smoothed interaction Hessian along the Polchinski flow of Bauerschmidt and Bodineau to have a small $L^2$ norm in the scale variable. The decay constants depend only on this $L^2$ norm and the friction. The proof measures the transfer of momentum dissipation to position on the same scales that decompose entropy. A phase-space extension of the Polchinski flow gives Gaussian coarse-graining that commutes with free kinetic evolution. Integrating position--momentum cross terms over these scales yields a corrector bounded by entropy. A force defect accounts for changes in the averaged interaction force away from equilibrium; its evolution in scale controls the errors in the kinetic estimate. For the massive two-dimensional lattice sine--Gordon model with $0<β<4π$ and small dimensionless local coupling, the criterion gives entropy decay for all finite-entropy initial laws, uniformly in the lattice spacing and the volume. For friction proportional to the mass $m$, the rate has optimal order $m$.

math-ph↗

arXiv2026

Rethinking Scientific Discovery in the Agentic Era

Artificial intelligence has advanced scientific discovery, but most AI4Science systems remain fragmented tools that rely on humans to coordinate problem formulation, literature grounding, model use, simulation, validation, and knowledge reuse. This paper presents \textbf{SCION (Scientific Collaborative Innovation with Agentic Organizational Nexus)}, an agentic scientific operating system that acts as an \textbf{organizational nexus}. Through a Science Agent serving as a \textbf{Meta-Harness}, SCION connects scientific tasks, tools, agents, artifacts, and memory, transforming research into an executable, auditable, and reusable operational process. At its core is the \textbf{Research Execution Plan (REP)}, which compiles high-level scientific intent into staged objectives, dependencies, verification checkpoints, tool requirements, expected artifacts, and fallback conditions. SCION further integrates hierarchical multi-agent execution, profile-driven specialization, selective context construction, governed delegation, and layered epistemic memory to support long-horizon scientific work. We formulate discovery under SCION as \textbf{Target-conditioned Inverse Search} and extend it to hidden-target settings through batch active search under finite experimental budgets. Applications in materials analysis, molecule design, and protein or antibody screening, together with experiments on scientific reading, idea generation, molecule generation, and antibody screening, show that SCION outperforms existing autonomous research-agent baselines, especially in decomposition, verification, refinement, and memory reuse. Overall, SCION shifts AI from isolated tools toward a coordinated operational layer for traceable and reusable scientific innovation.

cs.CL↗

arXiv2026

Classification of finite-dimensional Nichols algebras of rank two in twisted Yetter--Drinfeld categories

Let $G$ be a finite non-abelian group, let $Φ\in Z^3(G,\mathbb C^\times)$ be normalized, and let $V,W$ be finite-dimensional simple objects of ${}_G^G\mathcal{YD}^Φ$. We classify, up to interchange, the braided-indecomposable pairs $(V,W)$ whose supports generate $G$ and for which $\mathcal B(V\oplus W)$ is finite-dimensional. The classification consists of eight cases with five possible support quandles. In every case the Cartan graph is standard of type $A_2$, $B_2$, or $G_2$, and the dimension is determined explicitly. A new phenomenon occurs {in the $Γ_2$ case}: the twisted setting admits a family of type $G_2$ absent from the ordinary $Γ_2$ classification and we construct an explicit example over a non-abelian group of order $16$.

math.QA↗

arXiv2026

Near-Optimal Quantum Algorithm and Complexity Analysis for Riccati Problems

We develop quantum algorithms that construct block-encodings of solution matrices for continuous- and discrete-time algebraic Riccati equations (CAREs and DAREs), differential Riccati equations (DREs), and finite Riccati recursions. The Quantum Weighted Riesz Method, unifying four kinds of Riccati problems, constructs projectors onto the solution graphs using weighted Riesz branching operators and recovers the solution matrices. Under the stated access and normalization assumptions, our algorithms for CAREs and DAREs have near-optimal query complexity $Θ(\mathcal Rα)$ up to logarithmic factors, with the generalized singularity factor $\mathcal R$ that reflects spectral separation, and the output normalization $α$. For DREs and finite-horizon problems, the query complexity has an additional initialization conditioning factor $κ_{\rm init}$. We establish product query lower bounds for specified input-oracle families, demonstrating necessary joint dependence on spectral separation and initialization or output scale. We also prove that selected-value promise problems are BQP-complete on explicit circuit-generated families of single-control discrete-time algebraic and zero-terminal differential Riccati equations. Applications include classical feedback evaluation for linear-quadratic control, a query-complexity comparison for stable random-phase approximation Riccati equations in quantum chemistry, and heated boundary control with numerical validation.

quant-ph↗

arXiv2026

OncoReg: Medical Image Registration for Oncological Challenges

In modern cancer research, the vast volume of medical data generated is often underutilised due to challenges related to patient privacy. The OncoReg Challenge addresses this issue by enabling researchers to develop and validate image registration methods through a two-phase framework that ensures patient privacy while fostering the development of more generalisable AI models. Phase one involves working with a publicly available dataset, while phase two focuses on training models on a private dataset within secure hospital networks. OncoReg builds upon the foundation established by the Learn2Reg Challenge by incorporating the registration of interventional cone-beam computed tomography with standard planning fan-beam CT images in radiotherapy. Accurate image registration is crucial in oncology, particularly for dynamic treatment adjustments in image-guided radiotherapy, where precise alignment is necessary to minimise radiation exposure to healthy tissues while effectively targeting tumours. This work details the methodology and data behind the OncoReg Challenge and provides a comprehensive analysis of the competition entries and results. Findings reveal that feature extraction plays a pivotal role in this registration task. A new method emerging from this challenge demonstrated its versatility, while established approaches continue to perform comparably to newer techniques. Both deep learning and classical approaches still play significant roles in image registration, with the combination of methods, particularly in feature extraction, proving most effective.

eess.IV↗

arXiv2026

Legitimate Reconstruction Advantage in Wiretapped Semantics-Aware Remote Monitoring

This paper investigates joint accuracy and security in semantics-aware remote monitoring, where source updates intended for a legitimate receiver may also be intercepted by an eavesdropper. We define legitimate reconstruction advantage (LRA) as the joint event that the intended receiver correctly reconstructs the current source state using received updates while the eavesdropper reconstructs it incorrectly. Source evolution changes the relevance of received updates at the two receivers, so their marginal reconstruction accuracies do not, in general, determine the probability of this joint event. Under randomized stationary (RS) transmission and zero-order-hold reconstruction at both receivers, we derive an explicit joint stationary distribution and closed-form expressions for LRA probability and persistence for general finite-state ergodic Markov sources. For two-state and uniform-transition sources, we obtain algebraic expressions that enable low-complexity optimization of RS policies. Analytical and numerical results show that maximizing the probability of legitimate advantage need not maximize its persistence. For two-state sources, rare-transmission analysis reveals that, at a fixed sum of transition probabilities, greater transition asymmetry can increase persistence while reducing LRA probability. Moreover, optimizing marginal accuracy and security metrics can select policies that are suboptimal for LRA. Counterintuitively, more frequent transmission or improved legitimate-channel reliability can increase legitimate reconstruction accuracy while reducing LRA probability. Finally, a geofencing application translates prescribed LRA requirements into spatial constraints on eavesdropper locations.

cs.IT↗

arXiv2026

Dirac Points in High-Contrast Dielectric Honeycomb Structures

We study Dirac points for a two-dimensional scalar Helmholtz equation in a periodic dielectric medium with two penetrable inclusions per honeycomb cell. In a joint small-inclusion and high-contrast regime, we reduce the Bloch eigenvalue problem to a two-dimensional interaction matrix and derive a two-term asymptotic expansion for the two relevant eigenvalues, uniformly for quasi-momenta separated from the reciprocal lattice. At a vertex K of the Brillouin zone, rotation and inversion-conjugation symmetries force these eigenvalues to coincide. A Schur-complement analysis of the exact operator pencil, together with a nonvanishing lattice Green-function coefficient, proves that the resulting double eigenvalue has a nonzero conical splitting and gives an asymptotic formula for the conical slope in the scaled eigenvalue. For fixed sufficiently large contrast, symmetry-preserving changes of variables and analytic perturbation theory then continue this Dirac point as the inclusion size varies: it remains nondegenerate except at a relatively closed discrete set of admissible sizes.

math.AP↗

arXiv2026

Frequency-dependent capacitance matrix formulation for Fabry--Pérot resonances. Part II: One-dimensional periodic media

We study Bloch resonances of one-dimensional periodic systems of high-contrast resonators beyond the subwavelength regime. An exact transfer-matrix formulation identifies the limiting wavenumbers with the union of the interior Neumann frequencies and the Dirichlet frequencies of the separating intervals. At each fixed reference wavenumber, we introduce a periodic frequency-dependent capacitance matrix whose positive eigenvalues determine all leading square-root splittings in the contrast parameter. A direct cyclic matching expansion proves the approximation, including multiple eigenvalues and small complex contrasts. A spectral exclusion result outside the limiting set yields a direct bandgap statement. At a common resonance of all intervals, we also derive uniform eigenmode approximations and distinguish the zero-frequency mode from nonzero double Bloch eigenvalues. For positive contrast, exact linear crossings are characterized by the monodromy matrix being proportional to the identity. A complete local classification is obtained for two resonators per cell and, through band folding, for fully symmetric arrays with any number of resonators per cell. We further formulate the weighted Zak phase for isolated bands and its quantization under reflection symmetry. Explicit two-resonator formulas and numerical illustrations complement the analysis.

math.AP↗

arXiv2026

A nearly linear bound for the Lovász conjecture

The celebrated conjecture of Lovász from 1969 asks whether every connected vertex-transitive graph has a Hamiltonian path. Bucić, Christoph, Pokrovskiy and Steiner recently proved that every such graph on $n$ vertices contains a cycle of length $n^{2/3-o(1)}$. In this paper, we improve this bound to $n^{1-o(1)}$. Our proof uses a structure theorem of Tessera and Tointon to first obtain a partition of the vertex set into sets of small diameter in the original graph. When the parts are large, we repeatedly traverse a spanning tree of maximum degree at most three in the quotient graph, and use the Lovász local lemma to join random short paths along this traversal and extract a long path in the original graph. When the parts are small, we apply Babai's contraction lemma to reduce the problem to finding a long path in a connected Cayley graph of a nilpotent group with boundedly many generators and bounded nilpotency class, and then show that such a Cayley graph on $m$ vertices contains a path on $m^{1-o(1)}$ vertices.

math.CO↗

arXiv2026

Vesta: Queueing-Aware LLM Scheduling for Real-World Workloads

Agentic applications generate multi-turn sequences of LLM requests, with each new turn extending the context accumulated from prior interactions and tool calls. Prefix caching reuses the KV cache of earlier turns and avoids redundant prefill, but routing every turn to its cache-affine instance can create severe local queues while capacity remains idle elsewhere. Remote KV cache access relaxes this tension, yet migrating execution adds transfer work and is harmful when the cluster is already congested. Thus, deciding where to execute a request requires reasoning jointly about cache locality, queueing, and the current operating point of the cluster. We present Vesta, a queueing-aware scheduler for distributed agentic LLM serving. Vesta assigns each session a stable holder for its authoritative KV cache while independently selecting an executor for every request. Grounded in classical queueing theory, its objective accounts for backlog, reusable prefixes, KV cache transfer, and the queueing externality of additional service demand; it migrates execution to recover idle capacity when beneficial and favors placements that add the least prefill and transfer work as congestion rises. Vesta also migrates reusable prefixes across length buckets to avoid cold starts. On real-world agentic workloads, Vesta improves throughput by 14.0\% and reduces mean TTFT by 57.8\% over the state-of-the-art scheduler in a PD-colocated deployment. In PD-disaggregated deployments, it reduces mean TTFT by up to 18.1\% and P99 TTFT by up to 29.0\% at matched throughput.

cs.DC↗

arXiv2026

Recover, Discover, Plan: Learning Skills and Concepts from Robot Failures

Intelligent robots should not only recover from failures, but also acquire the abstract knowledge needed to avoid them in the future. While reinforcement learning (RL) can learn reactive recovery behaviors, training a separate policy for every distinct failure mode is highly inefficient. We introduce Recovery-Driven Synthesis of Relational Concepts (ReSYNC), the first approach that progressively discovers and refines state abstractions (relational predicates) from failure-recovery experience to support abstract planning. Unlike purely reactive methods, ReSYNC jointly learns skills and concepts through an incremental dual-learning process. In the skill-learning phase, the robot uses RL to learn to recover from failures seen in training tasks. In the concept-learning phase, the robot discovers new relational predicates and refines its abstract planning model to explain and generalize the learned recovery behaviors. This interaction enables ReSYNC to convert local recoveries seen during training into global failure avoidance at test time. Across four simulated domains, we show that ReSYNC's ability to continually expand and refine its abstraction library allows it to solve long-horizon, previously unseen problems, outperforming strong baselines by over 50%. Additionally, we demonstrate sim-to-real transfer of ReSYNC, where it performs real-world non-prehensile manipulation skills and generalizes to unseen scenarios through abstract planning. Overall, ReSYNC represents a significant step toward robots that autonomously acquire abstractions for scalable, failure-aware planning in the physical world.

cs.RO↗

arXiv2026

RepoLaunch: Automating Build and Management of Code Repositories across Languages and Platforms

Language model (LM) agents have driven substantial progress in automated software engineering (SWE), yet building and testing software repositories at scale remains a largely manual and labor-intensive bottleneck. In this work, we introduce RepoLaunch, a novel agentic framework that automatically resolves dependencies, compiles source code, and extracts test results across diverse programming languages and operating systems. RepoLaunch achieves a 78% build success rate, outperforming the Python/Linux-only prior system by 18%. To demonstrate its application, we further present a fully automated pipeline for SWE dataset creation driven by RepoLaunch, which only requires human input at the task-design stage. RepoLaunch is open-sourced, and its automated task-generation pipeline has been adopted by several recent works on agentic benchmarking and training.

cs.SE↗

arXiv2026

Model Predictive Communication for Timely Status Updates in Low-Altitude Networks

Timely information delivery in low-altitude networks is critical for many time-sensitive applications, such as unmanned aerial vehicle (UAV) navigation, inspection, and surveillance. The key challenge lies in balancing three competing factors: stringent data freshness requirements, UAV onboard energy consumption, and interference with terrestrial services. Addressing this challenge requires not only efficient power and channel allocation strategies but also effective communication timing over the entire operation horizon. In this work, we propose a model predictive communication (MPComm) framework, enabled by advanced channel sensing techniques, in which the channel conditions that the UAV will experience are largely predictable. Within this framework, we formulate a constrained bi-objective optimization problem to achieve a desired trade-off between energy consumption and terrestrial channel occupation, subject to a strict timeliness constraint. We solve this problem using Pareto analysis and show that the original non-convex, mixed-integer problem can be decomposed into a two-layer structure: the outer layer determines the optimal communication timing, while the inner layer determines the optimal power and channel allocation for each communication interval. An efficient algorithm for the inner problem is developed using non-convex analysis, with asymptotic optimality guarantees, while the outer problem is solved optimally via a simple graph search, with edges characterized by inner solutions. The proposed approach applies to a broad class of problem variants, including objective transformations and single-objective specializations. Numerical results demonstrate the efficiency of the proposed solution, achieving up to a six-fold reduction in terrestrial channel occupation and a 6dB energy saving compared to benchmark schemes.

eess.SY↗

arXiv2026

EMPIRIC: Experiment-Driven Learning of Residual World Models for Robot Planning

A robot should be able to learn through experiments how unfamiliar objects behave and interact, then plan with that knowledge. It need not start from scratch: physics engines supply knowledge of motion and contact, but can omit entire mechanisms, such as glue curing, water heating, or wind. We present EMPIRIC, an agent that learns a residual world model: a physics engine extended with code for the missing mechanisms. The learned programs can introduce new forces, constraints, and hidden state, and Bayesian inference estimates their parameters and states from noisy observations. The resulting model lets the agent predict the outcomes of actions, choose informative experiments, and revise its hypotheses when predictions fail. Across five simulated domains, EMPIRIC learns interpretable, reusable models, and solves more tasks with fewer environment interactions than all three baselines. On a physical robot, it learns wind forces and domino masses to solve a manipulation task. Website and code: https://yichao-liang.github.io/empiric

cs.RO↗

arXiv2026

Unifying Deep Predicate Invention with Pre-trained Foundation Models

Long-horizon robotic tasks are hard due to continuous state-action spaces and sparse feedback. Symbolic world models help by decomposing tasks into discrete predicates that capture object properties and relations. Existing methods learn predicates either top-down, by prompting foundation models without data grounding, or bottom-up, from demonstrations without high-level priors. We introduce UniPred, a bilevel learning framework that unifies both. UniPred uses large language models (LLMs) to propose predicate effect distributions that supervise neural predicate learning from low-level data, while learned feedback iteratively refines the LLM hypotheses. Leveraging strong visual foundation model features, UniPred learns robust predicate classifiers in cluttered scenes. We further propose a predicate evaluation method that supports symbolic models beyond STRIPS assumptions. Across five simulated and two real-robot domains, UniPred achieves 2-4 times higher success rates than top-down methods and 3-4 times faster learning than bottom-up approaches, advancing scalable and flexible symbolic world modeling for robotics. For more information and project materials (video, code...) visit https://unipred.github.io/

cs.RO↗

arXiv2026

Coding Agents for Generalized Task and Motion Planning Problems

Task and motion planning (TAMP) problems remain difficult even with full observability and object-centric states because discrete decisions are tightly coupled to geometric, kinematic, and dynamic constraints. Generalized TAMP addresses this difficulty by exploiting regularities across problem instances to reduce planning effort on new instances. However, existing methods require substantial TAMP-specific engineering. We investigate whether coding agents can automate this process by synthesizing programs that generalize across instances. Given a task description and simulator access, each agent chooses how to interact with the environment while developing a program within a fixed synthesis budget. The program is then frozen and evaluated on unseen instances. We evaluate Claude Code (Opus 5) and Codex (GPT-5.6 Sol and GPT-6 Astra) on 28 simulated environments from KinDER and PDDLStream, with object counts beyond those evaluated in the original benchmark. Across all program synthesis methods, we evaluate 980 generated programs on 100 held-out instances each, 98,000 evaluation episodes in total. Overall, we find that coding agents are surprisingly effective at generalized TAMP: all three agent configurations outperform hand-engineered planners, one-shot generation, and an LLM-based generalized planning baseline in mean success (56% to 95% versus 47% for the planners, on the 16 environments where a planner is available). As object counts grow, the agents' programs maintain higher success than the planner, using an order of magnitude less computation per instance on average. Logs show agents using interaction to calibrate physical models, test edge cases, and refine strategies. We release all code, including the full prompts given to the agents. These findings suggest that coding agents are a strong baseline for generalized TAMP.

cs.RO↗