Search arXiv⌕ Search

arXiv subjects

Peng Lu

Publications and source records attributed to Peng Lu.

At least 19 recordsLinked to original sources
arXiv2026

Predicting protein structures at any folding stage from sequence

Protein structure predictors reliably identify native conformations, but do not describe the continuum connecting unfolded ensembles to folded states. Here we present FoldScape, a sequence-conditioned deep learning framework that generates C$α$ conformational ensembles at a specified folding fraction $Q$. FoldScape reconstructs three-dimensional chains through hierarchical, sliding-window regressions of one-dimensional inter-residue distances, followed by chain assembly and sampling with a rectified-flow decoder. It is trained on multi-scale coarse-grained folding trajectories and all-atom mechanical-unfolding trajectories. Across 20 human proteins (36-857 residues), at matched sample size, FoldScape covers about an order of magnitude more of the folding coordinate than seed-varied AlphaFold 3 ensembles, robustly across proteins, order parameters and binning choices. AlphaFold 3 better anchors the immediate native vicinity, which FoldScape does not reach, making the approaches complementary. FoldScape reconstruction accuracy remains approximately constant along $Q$, from collapsed coil to near-native globule, rather than deteriorating away from the native state. A stride-one crop-and-superpose scheme extends inference to chains several times longer than a single chain-level forward pass. We also show that training trajectories resolve mechanical unfolding at residue resolution, that elongation-conditioned ensembles exhibit reproducible continuous observables across pulling velocities, and that partially unfolded chains form geometrically coherent segments. FoldScape thus combines broad coverage of folding stages with stable reconstruction accuracy, without per-protein simulation.

q-bio.BM↗
arXiv2026

ROOF: Learning Humanoid Fall Recovery under Overhead Constraints

A fallen humanoid beneath an overhead obstacle faces a coupled recovery problem. It must create clearance while organizing support for a stable rise, since attempting to stand before making room can further increase obstruction. We introduce a reinforcement learning method with contact-supported geometric escape guidance, which rewards incremental reduction of robot-obstacle overlap without prescribing an escape path. To achieve this, a learned motion prior, together with posture-dependent support and joint-stall costs, is proposed to shape a single policy that learns when to clear the obstacle, how to establish support, and when to complete the rise, all using proprioception alone. In simulation, it recovers in 88.5% of trials beneath plates that can only be lifted, compared with 18.9% without geometric guidance, and in 98.6% of trials beneath free plates with randomized size, mass and placement. On a Unitree G1, the deployed policy recovered in 95 of 150 attempts across seven real-world obstacle scenes, including furniture and multi-object arrangements absent from training. Project page: https://lu-yidan.github.io/ROOF/

cs.RO↗
arXiv2026

Global well-posedness and asymptotic behavior of large strong solutions to the 3D full compressible Navier-Stokes equations with temperature-dependent coefficients

It is well known that the global well-posedness of the Navier-Stokes equations with temperature-dependent coefficients is a challenging problem, especially in multi-dimensional space. In this paper, we study the 3D Navier-Stokes equations with temperature-dependent coefficients in the whole space. When the initial density and the initial temperature are linearly equivalent to some large constant states, we establish the first result on the global existence of large strong solution. Moreover, the optimal decay rates of the solution to its associated equilibrium are established when the initial data belong to $L^{p_0}(\mathbb{R}^3)$ for some $p_0\in[1,2]$.

math.AP↗
arXiv2026

From Structural Feedback to Prompt Policies: Learning Faithful Text-to-Image Prompt Editors

Prompt optimization should improve the image a user receives without changing the scene the user requested. Yet stylistic rewriting can weaken object attributes and relations even when global text--image similarity remains high. We propose StruPO, a structure-aware prompt editor that uses a calibrated scene graph as a shared representation for policy conditioning and image-grounded supervision. The graph exposes what should be preserved, while deletion, reordering, and insertion provide a token-aligned way to modify the prompt. This turns structural feedback from an external critique into a training signal for a reusable editing policy. Learning from that signal requires handling both uncertain grounding and stochastic image generation. StruPO retains grounding confidence through smooth reward shaping and averages multiple images for each edit before computing group-relative advantages. Experiments span three disjoint 2K-prompt test sets and four frozen generation backbones, with comparisons to learned prompt optimizers, iterative structural feedback, diffusion-model fine-tuning, and inference-time attention control. StruPO improves the faithfulness--aesthetics trade-off, with consistent gains in relational fidelity. Independent automatic evaluators and a 15-annotator study provide complementary evidence, and human preference margins increase with relation complexity. Controlled ablations connect these gains to graph conditioning, local editing, and image-grounded learning. The resulting policy produces reusable optimized text in one pass after graph calibration, without an iterative critique--rewrite loop.

cs.AI↗
arXiv2026

Herculean: An Agentic Benchmark for Financial Intelligence

As AI agents advance, the central question is no longer whether they can solve isolated financial tasks, but whether their capabilities generalize across complete professional workflows. Existing benchmarks cannot answer this question because they evaluate either static competencies or agents within a single interaction structure. We introduce Herculean, a benchmark of end-to-end agentic financial intelligence that crosses agent architecture, backbone model, and four structurally distinct workflows: Trading, Hedging, Market Insights, and Auditing. Each workflow is implemented as an MCP skill that preserves its native state, actions, constraints, and outcome criteria, allowing intact agent systems to be evaluated under matched external conditions. Across 30 architecture--backbone configurations, we find that agent capability is not globally ordered. Workflow rankings are weakly aligned, changing the architecture frequently reverses performance under the same backbone, and each architecture concentrates failure in different forms of workflow control. These results establish that agent performance is an interaction between architecture and workflow demand, rather than an intrinsic property of the underlying model. Consequently, success on a single financial workflow provides little evidence of dependable, general financial intelligence.

cs.AI↗
arXiv2026

Steady gradient Ricci Yang-Mills solitons on 2-orbifolds

Following the recent work of M. Womack \cite{Wo26} we consider the steady Ricci Yang-Mills solitons on 2d surfaces containing orbifold point $\mathbb{R}^2/\mathbb{Z}_p$. We show that there are a family of solitons depending parameter $λ$ which approaches to $\mathbb{Z}_p$-quotient of Hamilton cigar soliton as $λ\to (-2/p)^+$ and approaches to (after rescaling) $\mathbb{Z}_p$-quotient of round sphere as $λ\to \infty$. For any integer $q >p$ there is a $λ$ whose corresponding soliton metric is defined on football orbifold $S_{p,q}^2$.

math.DG↗
arXiv2026

Global strong solutions to the 1D full compressible Navier--Stokes equations with temperature-dependent coefficients and vacuum

We study the initial--boundary value problem for the one-dimensional full compressible Navier--Stokes system whose viscosity and heat conductivity are given by $μ(θ)=μθ^α$ and $κ(θ)=κθ^β$, respectively. The initial density is nonnegative and is allowed to vanish. For $α>2$ and $β>2α+4$, we prove the existence of a global strong solution for arbitrarily large compatible $H^2$ data, provided that the prescribed boundary temperature is sufficiently high relative to their size. No velocity or temperature perturbation is required to shrink as the boundary temperature tends to infinity. A shifted initial-layer weight and an exact symmetrization of the two material-derivative equations separate the fast thermal mode from the momentum mode. The resulting estimates are uniform with respect to the positive lower bound used in the approximation and therefore survive the limit to genuine vacuum. The proof further combines a relative-entropy identity centered at the conserved mean density, the physical effective viscous flux, a Zlotnik-type estimate along particle trajectories, and finite-time transport estimates for the density derivatives. The density remains uniformly bounded and the temperature stays uniformly comparable with the prescribed boundary temperature. After the initial layer we establish vacuum-compatible exponential relaxation: the density converges to its conserved mean in every finite $L^p$, the velocity and temperature perturbation decay in $H^1$, and the weighted material-derivative energy decays at rate $c\barθ^{1-α}$.

math.AP↗
arXiv2026

Global solutions to the two-dimensional compressible MHD equations with density-dependent bulk viscosity

We study the two-dimensional periodic resistive compressible magnetohydrodynamic equations with constant shear viscosity and magnetic diffusivity, bulk viscosity $ρ^β$, and pressure $ρ^γ$. For every $β>1$ and $γ>1$, we construct global finite-energy renormalized weak solutions from strictly positive bounded $W^{1,q}$ densities, $q>2$, and arbitrary $H^1$ velocity and magnetic data. The construction combines approximation-independent finite density moments, generalized effective-flux identities, and a joint logarithmic renormalization of the density and its entropy defect. An augmented relative energy containing the $β$-power density potential yields weak--strong uniqueness against positive $H^2$ strong solutions for the same full pressure range. We also establish global strong existence for a large-background family when $γ>\max\{5-2β,β+1\}$. Writing $m=\intρ_0$, the density perturbation is bounded in the scaled norm $m^{(γ-3)/2}\|ρ_0-m\|_{H^2}$, while the fixed $H^2$ velocity and magnetic norms need not be small. For sufficiently large $m$, frequency-dependent acoustic dissipation, staged vorticity weights, and a separate initial-layer flux estimate close a simultaneous bootstrap. The admissible pressure exponents have infimum $7/3$, with the endpoint excluded. For $β>2$ the relevant damping rate is the minimum of the viscous and pressure-relaxation rates; keeping both rates extends the argument without an upper restriction on $β$. The strong theorem has no upper restriction on $γ$ under this scaled density hypothesis; for fixed additive density perturbations it applies in the stated lower range up to $γ=3$.

math.AP↗
arXiv2026

Semantically Similar, Yet Not Answerable: Diagnosing the Semantic-Answerability Gap in Table RAG

In retrieval-augmented generation (RAG), semantic relevance asks whether a source matches a query in meaning, while answerability asks whether it contains sufficient information to answer the query. A Semantic-Answerability Gap (SAG) may arise in retrieval when a retriever can reach semantically relevant sources yet fail to identify those that are uniquely answerable. We uncover this gap using tables as a controlled setting, where shared schemas and entities provide strong semantic signals while localized content and row-column bindings distinguish answerable from non-answerable sources. Using TCR-Bench, a controlled sibling-table benchmark, we find that dense retrievers achieve only 18.2% top-1 target retrieval, reducing QA F1 from 0.755 with the oracle table to 0.330 with retrieved top-5 tables. Controlled diagnostics show that retrievers favor semantic volume over sufficiency, respond weakly to row-column binding disruptions, and struggle to distinguish Targets from Siblings. Explicit answerability assessment substantially improves target identification, while fine-tuning shows that answerability is learnable but difficult to transfer without compromising broad semantic retrieval.

cs.AI↗
arXiv2026

An Improved Lower Bound for Packing 203 Equal Spheres in a Unit Cube

This paper reports an improved numerical lower bound for packing 203 equal spheres in a unit cube. Let r denote the maximum common radius of equal non-overlapping spheres that can be packed in a unit cube. For N = 203, the configuration currently listed in Packomania has a feasible radius of approximately r = 0.088145402159. Starting from this public configuration, we perform further numerical optimization and obtain a new feasible configuration with r > 0.088145634693. The final configuration is checked by an independent verification program using 100-digit arithmetic. All 20503 sphere-pair constrains and all 1218 sphere-wall constraints are satisfied. The coordinates of the new configuration and the related programs are provided as supplementary material.

math.MG↗
arXiv2026

ARM: Attention with Routed-Memory for Learnable Sparse Control

Despite advances in long-context inference, large language models (LLMs) remain fundamentally limited by the key-value (KV) caching mechanisms that are necessary for stable computation. Techniques such as selective token eviction and pruning have vastly mitigated these issues, but often discard core information to manage the growing cache. In this paper, we propose Attention with Routed Memory (ARM) a novel KV caching structure that introduces a fully differentiable, fixed-size memory system organized as a hierarchical router. Via a Gumbel-Softmax, ARM learns to select memory slots and perform sigmoid-gated updates that softly combine new and stored information, avoiding hard eviction and reducing information loss. By further training a policy to dynamically select varying amounts of memory at inference, ARM adapts its accesses for both simple contexts and inputs that require deeper reasoning, enabling more scalable and effective retrieval on both short- and long-contexts. Experimental results on standard commonsense and long-context reasoning benchmarks demonstrate that ARM achieves superior performance and efficiency compared to fixed KV-caching approaches, while remaining efficient and scalable in terms of both memory and generation latency.

cs.LG↗
arXiv2026

NeurIDA: Dynamic Modeling for Effective In-Database Analytics

Relational Database Management Systems (RDBMS) manage complex, interrelated data and support a broad spectrum of analytical tasks. With the growing demand for predictive analytics, the deep integration of machine learning (ML) into RDBMS has become critical. However, a fundamental challenge hinders this evolution: conventional ML models are static and task-specific, whereas RDBMS environments are dynamic and must support diverse analytical queries. Each analytical task entails constructing a bespoke pipeline from scratch, which incurs significant development overhead and hence limits wide adoption of ML in analytics. We present NeurIDA, an autonomous end-to-end system for in-database analytics that dynamically "tweaks" the best available base model to better serve a given analytical task. In particular, we propose a novel paradigm of dynamic in-database modeling to pre-train a composable base model architecture over the relational data. Upon receiving a task, NeurIDA formulates the task and data profile to dynamically select and configure relevant components from the pool of base models and shared model components for prediction. For friendly user experience, NeurIDA supports natural language queries; it interprets user intent to construct structured task profiles, and generates analytical reports with dedicated LLM agents. By design, NeurIDA enables ease-of-use and yet effective and efficient in-database AI analytics. Extensive experiment study shows that NeurIDA consistently delivers up to 12% improvement in AUC-ROC and 25% relative reduction in MAE across ten tasks on five real-world datasets. The source code is available at https://github.com/Zrealshadow/NeurIDA

cs.DB↗
arXiv2026

FINSKILLOPS: A Self-Evolving Multi-Agent System for SEC Filing QA

Financial QA systems are typically improved before deployment through better retrieval, prompting, or agent coordination, leaving their reliability behavior fixed thereafter. In practice, new SEC-filing questions repeatedly expose heterogeneous errors in period, entity, evidence use, and calculation. Existing self-improvement methods can turn failures into new behaviors, but offer limited control over where a correction should apply or which previously correct answers it may break. We therefore frame post-deployment improvement as controlled behavioral maintenance: recurring failures should become scoped skill patches, and each patch should earn deployment with- out introducing regressions. We instantiate this view in FINSKILLOPS, a multi-agent system for SEC filing QA. FINSKILLOPS derives reusable skills from evidence-grounded, typed failure diagnoses and governs them through targeted validation, protected-case regression checks, negative controls, and versioned replacement or retirement. Across six financial QA benchmarks, a single frozen skill registry achieves the highest verdict-weighted correctness and reference consistency among the evaluated systems. Evolved skills raise correctness from 3.70 to 4.55 on our enhanced benchmark. In a separate 12-round operational study, only six of 33 proposed skills are promoted, while the monitoring non-correct rate falls from 20.0% to 12.5%. These results establish controlled skill scope, admission, and lifecycle management as the foundation for reliable self-improvement.

cs.AI↗
arXiv2026

BENCHCOMPASS: From Scores to Signals for Training and Harness Decisions in Payment-Domain LLMs

Payment operations are a critical financial infrastructure, but the value of large language models in this domain remains unclear because payment rules change quickly, evidence is fragmented, and decisions depend on transaction state, participant role, region, and payment rail. Existing benchmarks do not isolate whether failures come from missing payment-rule knowledge, poor use of supplied evidence, or brittleness under imperfect harness inputs. We introduce BENCHCOMPASS, a payment-domain benchmark whose construction pipeline builds scenario-grounded tasks from typed evidence packs, applies LLM-based quality checks, creates task-input attack variants, and reserves final item admission for domain experts. The release contains an expert-reviewed Pro benchmark covering payment knowledge, context-grounded scenario reasoning, and Attacked Open robustness, plus a lower-assurance Normal pool for inspection and future curation. Across 16 model variants, BENCHCOMPASS shows qualitatively different failure modes: missing parametric payment knowledge, incomplete reasoning over supplied rules, and failure to reject plausible but invalid workflows. The benchmark remains unsaturated: the best frontier model reaches 89.6% on Open Context-Grounded Reasoning and 81.7% under attacked inputs, while a representative 32B open-weight model reaches 69.8% and 42.6%. Benchmark data and code are available at https://github.com/ant-intl/BenchCompass.

cs.AI↗
arXiv2026

Learning Human-Like Badminton Skills for Humanoid Robots

Realizing versatile and human-like performance in high-demand sports like badminton remains a formidable challenge for humanoid robotics. Unlike standard locomotion or static manipulation, this task demands a seamless integration of explosive whole-body coordination and precise, timing-critical interception. While recent advances have achieved lifelike motion mimicry, bridging the gap between kinematic imitation and functional, physics-aware striking without compromising stylistic naturalness is non-trivial. To address this, we propose Imitation-to-Interaction, a progressive reinforcement learning framework designed to evolve a robot from a "mimic" to a capable "striker." Our approach establishes a robust motor prior from human data, distills it into a compact, model-based state representation, and stabilizes dynamics via adversarial priors. Crucially, to overcome the sparsity of expert demonstrations, we introduce a manifold expansion strategy that generalizes discrete strike points into a dense interaction volume. We validate our framework through the mastery of diverse skills, including lifts and drop shots, in simulation. Furthermore, we demonstrate the first zero-shot sim-to-real transfer of anthropomorphic badminton skills to a humanoid robot, successfully replicating the kinetic elegance and functional precision of human athletes in the physical world.

cs.RO↗
arXiv2026

Bridging Language and Spherical Space: Object-Centric Control for Text-to-Panorama Generation

Panoramic image generation is increasingly important for immersive applications such as virtual reality, augmented reality, and 3D content creation. Unlike perspective images, panoramic images represent a viewer-centered $360^\circ$ surrounding space, where directional expressions such as left, right, front, and behind play a central role in spatial understanding. However, existing text-to-panorama methods largely rely on implicit spatial reasoning and often fail to faithfully ground object-level directional descriptions in spherical panoramic scenes. A straightforward alternative is to introduce explicit layouts, but requiring manually specified spatial conditions reduces the flexibility of language-based interaction and does not directly resolve the misalignment between egocentric directional language and panoramic image space. To address this issue, we propose PanoCtrl, an object-centric framework for controllable text-to-panorama generation. Our method explicitly bridges natural language and spherical panoramic space by converting textual descriptions into structured object-level spherical conditions and integrating them into the diffusion process. Specifically, we introduce PanoParse, a text-conditioned parser that predicts object semantics and spherical bounding field-of-view (BFoV) parameters, and \textbf{PanoControl}, which injects object-level semantic and spatial guidance into the diffusion transformer through object-aware attention and spatial residual enhancement. To support this task, we construct PanoGround, a dataset with object-level spherical annotations and diverse directional descriptions for controllable panoramic generation. Extensive experiments demonstrate that PanoCtrl achieves state-of-the-art performance in both spatial alignment and image quality.

cs.CV↗
arXiv2026

MILD: Tractable Terrain Modeling for Learning Improved Bipedal Locomotion on Deformable Surfaces

Enabling robots to walk on yielding terrain is vital for applications ranging from disaster response to planetary exploration. While bipedal robots hold immense potential, their locomotion on deformable surfaces remains limited as current simulators fail to capture the spatiotemporal heterogeneity of such yielding substrates. We present MILD, featuring a physics-grounded discrete-element contact solver that accurately simulates spatially varying foot-terrain interactions. Complementing this model, we train a terrain-aware locomotion controller via deep reinforcement learning with latent modulation and proprioceptive estimation. Quantitative comparisons against state-of-the-art methods show our approach generates more diverse and realistic contact scenarios during training, resulting in controllers that exhibit natural adaptation on real deformable surfaces. Through hardware experiments, we demonstrate the system's capability for online terrain identification and adaptation across a wide range of surface stiffness.

cs.RO↗