arXiv subjects

Chunxiao Liu

Publications and source records attributed to Chunxiao Liu.

At least 19 recordsLinked to original sources

arXiv2026

Quasi-flat and exact flat bands in patterned two-dimensional electron gases and honeycomb wire networks

We investigate the formation of flat bands in a two-dimensional electron gas patterned by a periodic triangular array of repulsive antidots. In the high-barrier, narrow-trench regime, the spectrum develops kagome-like band triplets, each comprising a nearly flat band and two dispersive bands, with wave functions concentrated along the potential-free trenches. The two lowest energy bands have a different character with graphene-like dispersion and wave functions localized near the junction centers. The appearance of the trench-concentrated nearly flat bands further motivates a honeycomb quantum graph model as an effective description of the narrow-trench limit. In the single-channel graph, destructive interference produces exact flat bands with compact localized eigenstates protected by the $D_3$ symmetry. Their energies obey a quantization condition determined by propagation along the links and scattering at the junctions. We then extend the analysis to a multichannel quantum graph and derive the most general junction scattering matrix compatible with $D_3$ symmetry. Even- and odd-parity transverse channels contain equivalent symmetry sectors, and $D_3$ symmetry therefore allows them to mix. This mode conversion can disrupt the destructive interference responsible for flat bands and generically gives those bands a finite width. Exact flatness is nevertheless retained when only one transverse-parity sector is active, or when an additional symmetry forbids mixing between the two sectors. These results connect the continuum antidot model to its quantum-graph limit and identify transverse-mode mixing as a mechanism that limits flatness in multichannel networks.

cond-mat.mes-hall↗

arXiv2026

DynaWeave: Learning to Adapt through Dynamic Scenario Synthesis

Large language model (LLM) agents increasingly use tools to complete tasks in stateful environments. However, many training environments remain largely static apart from changes caused by the agent's own actions, whereas external events introduced by other users, services, or sudden contingencies in real environments can invalidate ongoing plans. To address this gap, we introduce DynaWeave, an automated synthesis framework that transforms existing static tasks into verifiable dynamic scenarios. The framework generates exogenous events relevant to each task, covering changes in task requirements, environment state, and external information. It uses directed acyclic graphs (DAGs) to represent dependencies between events and an execution engine to activate them during agent interaction. A dynamic verifier revises only the evaluation checks affected by events that actually occur and retains all other checks. Using DynaWeave, we collect 2,095 verified interaction trajectories across 36 environments and use them for supervised fine-tuning of Qwen3.5 models. Experiments on Gaia2, CostBench, and ToolMaze show that DynaWeave achieves consistent improvements in overall performance across all three evaluated model sizes, increasing the task completion rate on CostBench by 67.34 percentage points and even exceeding the performance of Gemini-3.1-Pro-Preview on ToolMaze. These results support dynamic scenario synthesis as a practical approach to training agents to adapt to changing environments. Our code is available at https://github.com/Miaow-Lab/DynaWeave.

cs.CL↗

arXiv2026

Projective symmetry groups beyond spin-$1/2$: Classifying integer-spin quantum spin liquids

Wen's projective symmetry group (PSG), introduced a quarter of a century ago, reduced the classification of symmetric quantum spin liquids to an algebraic problem and has guided the study of spin-$1/2$ models and materials ever since. Many current candidate materials and quantum simulators, however, have spin-$1$ or larger moments, and no comparable classification exists for them. We formulate the PSG for fermionic partons of arbitrary spin $S$. Integer and half-odd-integer spins already differ in their gauge structure. Half-odd-integer spins give the familiar $\mathrm{SU}(2)$, $\mathrm{U}(1)$, and $\mathbb{Z}_2$ invariant gauge groups, while integer spins give a different hierarchy that includes $\mathrm{U}(1)\rtimes\mathbb{Z}_2$ and $\mathbb{Z}_2\times\mathbb{Z}_2$; the latter has no spin-$1/2$ counterpart. Three lattice-independent results follow for integer spin. Singlet pairing of the partons is odd under site exchange. A lattice symmetry can exchange the particle and hole representations of the local constraint, which forces symmetry-related sites into opposite constraint sectors. And a time-reversal-symmetric $\mathrm{U}(1)$ state with the same constraint on every site can be gapped only if the number of sites per unit cell is a multiple of $2S+1$. We apply the formalism to the spin-$1$ diamond lattice, realized in frustrated A-site spinels, and find ten fully symmetric Ansätze up to second neighbors. In the $J_1$-$J_2$ Heisenberg model at the mean-field level, the ground state depends on how the constraint is imposed. With the particle and hole representations, the lowest states are fully gapped $\mathrm{U}(1)$ states whose two sublattices sit in opposite constraint sectors, and one of them is gauge equivalent to a paired Ansatz. The classification provides a starting point for integer-spin quantum spin liquids on other lattices and in other materials.

cond-mat.str-el↗

arXiv2026

Has GPT-6-Astra Achieved Embodied Intelligence?

GPT-6-Astra demonstrates a level of competence in robotic manipulation and navigation that challenges the conventional paradigm of building specialized models for individual tasks and embodiments. In this report, we ask not only how capable GPT-6-Astra is, but what its capabilities imply for the field. First, we probe the limits of standalone GPT-6-Astra. We conduct a large-scale evaluation across 12 representative benchmarks spanning tabletop manipulation, humanoids, mobile platforms, and dexterous hands. Under a unified protocol that allows in-context learning (ICL) from training-set demonstrations and substantially increases the number of execution steps, GPT-6-Astra achieves unprecedented performance (Fig. 1). Future research may focus on improving 1) the generality of ICL 2) cost and execution time, 3) physical-state verification, 4) recovery from irreversible failures, and 5) long-horizon preservation of progress. Second, we study the agentic harness for embodied control. We identify that close-loop coordination between a high-level "System 2" reasoner (i.e., GPT-6-Astra) and a low-level "System 1" controller (e.g., π0.5) is costly but effective. Third, we examine GPT-6-Astra on real-world deployment. Our experiments on 8 robot configurations reveal substantial gaps: 1) robot demonstrations required for ICL are expensive to collect, and 2) it is implausible to trade longer execution for higher success rates with physical constraints. Fourth, we envision the next generation of embodied benchmarks. We develop a prototype that 1) derives tasks from real-world needs, 2) provides no in-domain demonstrations, and 3) judges agents on a world clock that keeps running while they deliberate. Taken together, our results suggest that GPT-6-Astra has largely crossed the threshold of embodied competence, but not yet that of practical embodied intelligence.

cs.RO↗

arXiv2026

CoTS: Competitive Token Selection for Latent Reasoning in Large Language Models

Recent latent reasoning methods move beyond discrete Chain-of-Thought (CoT) by replacing sampled tokens with continuous soft embeddings or hidden states, enabling richer intermediate representations and preserving alternative reasoning paths. However, the need for latent reasoning varies across decoding steps: confident predictions favor discrete commitment, whereas ambiguous steps benefit from preserving multiple plausible alternatives. Existing approaches typically handle this variation through either global soft activation or hard switching based on uncertainty thresholds, treating latent intervention as an all-or-nothing decision rather than continuously adapting its strength. To address this limitation, we propose Competitive Token Selection (CoTS), a training-free framework that reframes latent intervention from binary mode selection into continuous allocation at the token level. CoTS uses the probability competition between the two most likely candidates to dynamically balance discrete commitment and soft exploration at each reasoning step. The second candidate represents the strongest alternative to the model's preferred continuation, so the separation between the two directly reflects the stability of the current discrete choice. When one candidate clearly dominates, CoTS keeps the representation close to the sampled discrete token to preserve a stable reasoning trajectory; as the competition strengthens, it increases the contribution of the soft representation to retain plausible alternatives. This competition-based allocation requires neither manually tuned switching thresholds nor mixing coefficients, and introduces no additional forward passes. Experiments across six reasoning benchmarks demonstrate that CoTS consistently outperforms standard CoT and state-of-the-art training-free latent reasoning methods. Our code is available at https://github.com/OkinoYu/CoTSReasoning.

cs.CL↗

arXiv2026

Scalable tests of quantum contextuality from stabilizer-testing nonlocal games

Soon after the dawn of quantum error correction, DiVincenzo and Peres observed that stabilizer codewords could give rise to simple proofs of quantumness via contextuality. This discovery can be recast in the language of nonlocal games: every $n$-qubit stabilizer state defines a specific "stabilizer-testing" $n$-player nonlocal game, which quantum players can win with probability one. If quantum players can moreover outperform all possible classical players, then the state is contextual. However, the classical values of stabilizer-testing games are largely unknown for scalable examples beyond the $n$-qubit GHZ state. We introduce several new methods for upper-bounding the classical values of these games. We first prove a general coding-theory bound for all stabilizer-testing games: if the classical value $p_{\mathrm{cl}}^* < 1$, then $p_{\mathrm{cl}}^* \leq 7/8$, i.e., there is no classical strategy that can perform as well as the optimal quantum strategy even in an asymptotic sense. We then show how to tighten this bound for the most common scalable examples, namely GHZ, toric-code and cyclic cluster states. In particular, we establish an asymptotically tight upper bound for cyclic cluster states using transfer-matrix methods. This leads to the striking conclusion that measuring an exponentially small fidelity to the cyclic cluster state will suffice to witness its contextuality.

quant-ph↗

arXiv2026

Crystallography, Group Cohomology, and Lieb-Schultz-Mattis Constraints

We present a computational study of the mod-2 cohomology of three-dimensional (3D) space groups, with an eye toward their applications in Lieb--Schultz--Mattis constraints. We prove finite-generation results for the cohomology of crystallographic groups and give ring presentations for $H^*(G,\mathbb Z_2)$ for \emph{all} 230 3D space groups, together with explicit inhomogeneous representatives for the degree-$\leq 3$ cocycles used in the lattice applications. The all-degree interpretation of the ring presentations is organized through finite LHS-spectral-sequence and Hilbert-series verification checks. We then associate distinguished classes in $H^3(G,\mathbb{Z}_2)$ to irreducible Wyckoff positions and use these classes as cohomological representatives of Lieb--Schultz--Mattis anomaly candidates when the on-site projective representations are classified by powers of $\mathbb{Z}_2$. Finally, we apply the resulting anomaly data to $\mathrm U(1)$ quantum spin liquids on the 3D pyrochlore lattice and compare the symmetry-fractionalization constraints with projective-symmetry-group calculations.

cond-mat.str-el↗

arXiv2026

Kitaev-Heisenberg model on the square-hexagon-dodecagon lattice

We study the spin-$1/2$ Kitaev-Heisenberg model on the square-hexagon-dodecagon lattice with a symmetry-preserving tri-coloring of the Kitaev exchanges. We focus on two cases: (i) the pure Kitaev model with anisotropic exchanges and (ii) the Kitaev-Heisenberg model with isotropic Kitaev exchange. In the absence of Heisenberg interactions, we use the standard mapping of the Kitaev model onto a quadratic Majorana-fermion problem in a $\mathbb{Z}_2$ gauge field. We find a gapped energy spectrum throughout the positive-coupling phase diagram, except at a single point where the twofold-degenerate energy bands form a Dirac cone. This unique gapped region corresponds to a toric code phase. However, depending on the couplings, vortex excitations on square, hexagonal, and dodecagonal plaquettes realize three distinct relative assignments to the $e$ and $m$ anyons. We determine these relative assignments from the physical fermion parity of the vortex sectors and locate their boundaries by tracking zero crossings of vortex-bound Majorana levels. For the isotropic Kitaev-Heisenberg model, we combine graph-based projected entangled-pair-state calculations, exact diagonalizations, and linear spin-wave theory to study the entire phase diagram. Apart from the toric code topological phases that are robust around the Kitaev limits, we obtain four collinear magnetically ordered phases that can be characterized by their local and relative orderings of hexagonal plaquettes. We also identify a special point at which the ground state is a product state.

cond-mat.str-el↗

arXiv2026

Xiaomi-TabLDM: A Tabular Foundation Model Technical Report

We introduce Xiaomi-TabLDM, a tabular large data foundation model for classification and regression via in-context learning, which delivers superior prediction accuracy without requiring task-specific fine-tuning. Pretrained exclusively on synthetic data generated from structural causal models (SCMs), our model enables more flexible context utilization and more efficient capacity scaling. i) A new performance standard. Strong regression performance across benchmarks: Xiaomi-TabLDM ranks 1st on OpenML-CTR23 and 2nd on regression across TALENT, TabArena, and BCCO, demonstrating consistently strong regression performance across four complementary benchmark suites. Favorable performance--efficiency trade-off: Xiaomi-TabLDM combines strong predictive performance with substantially lower computational cost. For example, on TabArena regression, it achieves the second-highest Elo while using 82% less training time and 68% less prediction time than the top-ranked TabFM. ii) Large-scale synthetic pretraining. Xiaomi-TabLDM expands the coverage and diversity of synthetic tabular data used for pretraining. We also adopt a three-stage training strategy together with dual-stream feature grouping, lightweight Attention Residual, and sparse Mixture-of-Experts, enabling Xiaomi-TabLDM to learn richer feature interactions and expert specialization across diverse tabular tasks. iii) Test-time scaling. Xiaomi-TabLDM further extends tabular prediction through test-time compute scaling, where allocating additional computation at inference time consistently improves predictive performance over the base model.

cs.AI↗

arXiv2026

ACE-Ego-Hand: Repurposing Video Diffusion Models for Occlusion-Robust Egocentric 3D Hand Motion Recovery

Egocentric video offers scalable manipulation data for embodied AI, yet recovering metric 3D hand trajectories remains challenging due to severe object occlusion and frequent out-of-sight gaps. Existing single-frame and windowed temporal regressors fail when a hand shortly leaves the frame, while recent video diffusion models (VDMs) rely on heavy, stochastic multi-step sampling as pixel-space renderers. We instead repurpose VDM into a deterministic geometry encoder. A single forward pass over the clean latent exposes scene content beyond current observations, including occluded and out-of-sight hands. We introduce ACE-Ego-Hand, an offline clip-level framework that extracts features via a Deterministic Clean-Latent Encoder and decodes them with a Bidirectional Spatiotemporal Decoder. ACE-Ego-Hand recovers continuous bimanual trajectories with metric placement and no external detector, while a Ray-Based Camera Solver supports a second configuration that requires no test-time camera intrinsics. Across five egocentric benchmarks, ACE-Ego-Hand sets a new state of the art, cutting MPJPE-p by 30% on occlusion-heavy ARCTIC and 40% on HOT3D. These gains reach 46%-61% once out-of-sight hands are included in the evaluation, offering a scalable path from everyday human video to robot manipulation data.

cs.CV↗

arXiv2026

Preserve More Details: Mitigating Content Drift in Real-World Image Super-Resolution

Real-world image super-resolution (Real-ISR) aims to reconstruct high-quality (HQ) images from low-quality (LQ) inputs subject to diverse real-world degradations. Recent advances have leveraged the LQ inputs and natural image priors learned by Stable Diffusion models to achieve impressive results. However, existing methods often overlook insufficient clarity of LQ inputs inevitably induce content drift in the generated HQ images. This manifests primarily as visual detail degradation and textual semantic shift, severely compromising both fidelity and perceptual quality. To address this challenge, we propose FSP-Diff, a novel one-step diffusion model featuring a dual-pathway architecture. This architecture comprises a Detail-Conditioned Pathway for injecting structured details to recover fine structures, and a Detail-Modulated Semantic Pathway that refines semantic guidance using structured details to mitigate semantic deviations. Extensive experiments on standard Real-ISR benchmarks demonstrate that FSP-Diff surpasses existing one-step diffusion methods in both quantitative and qualitative metrics.

cs.CV↗

arXiv2026

Beyond Uniform Restoration: Empowering All-in-One Restoration with Pixel-Level Multimodal Guidance

All-in-one image restoration is a unified low-level vision task that aims to effectively recover high-quality images from inputs degraded by various types and levels of corruption using a single model. Recent works have achieved remarkable progress by learning degradation-adaptive prompts or network architectures. However, these methods typically apply a uniform restoration strategy across the entire image, neglecting the fact that different regions may suffer from distinct degradation types and varying degrees of severity. In contrast, we propose to perform restoration at the pixel level, thereby enabling more fine-grained and precise control over the restoration process. Specifically, we present MGN-AIR, a novel pixel-level restoration framework for all-in-one image restoration. Our approach first learns to estimate a pixel-level visual prompt. Then, it leverages both textual and visual prompts to provide global and local degradation cues, guiding the model on where to look and how to restore at each pixel. We conduct extensive experiments on multiple all-in-one image restoration benchmarks, covering a wide range of tasks including denoising, deraining, deblurring, dehazing, desnowing, and low-light enhancement. Experimental results demonstrate that our proposed method consistently and significantly outperforms existing approaches.

cs.CV↗

arXiv2026

Groc-PO: Grounded Context Preference Optimization for Truthful Multimodal LLMs

Despite the rapid progress of Multimodal Large Language Models (MLLMs), they still suffer from untruthfulness issues, such as visual hallucinations, content fabrication, and unfaithful reasoning, which substantially undermine their faithfulness and practical utility. Alignment methods based on human preference, such as Direct Preference Optimization (DPO), have been widely adopted to address these issues. However, multimodal reasoning errors often propagate across stages, and final-answer errors can often be traced to mistakes in early grounding stages, yet standard DPO typically applies preference optimization at the final-answer level. This credit-assignment challenge means that supervision for early grounding stages is indirect rather than stage-specific, making it difficult to suppress error propagation arising from grounding drift and context inconsistency. To address this, we propose Grounded Context Preference Optimization (Groc-PO), a grounded preference optimization framework for MLLMs. We further construct the Grounded Context Preference Dataset (GCPD), organizing multi-stage preference samples around three stages of Object Grounding, Contextual Grounding, and Grounded Reasoning, to capture the formation, integration, and utilization of grounded context. By introducing more explicit preference supervision over multiple grounded stages, Groc-PO strengthens context-dependent reasoning and mitigates cross-stage error propagation. Extensive experiments show that, compared with standard DPO and other strong baselines, Groc-PO achieves improved performance in hallucination mitigation, faithful reasoning, and overall reliability, supporting the value of more explicit grounded supervision for trustworthy multimodal reasoning.

cs.CV↗

arXiv2026

Classification and correlation signatures of chiral spin liquids on the pyrochlore lattice

We present a systematic classification and variational study of chiral quantum spin liquids on the pyrochlore lattice based on fermionic parton constructions. Focusing on chiral $\mathrm{U(1)}$ and $\mathbb{Z}_2$ spin-liquid Ansätze, we characterize their symmetry properties, flux structures, and low-energy spinon spectra within a projective symmetry group framework, and incorporate gauge fluctuations through Gutzwiller-projected wave functions studied by variational Monte Carlo. From the equal-time spin structure factor, we develop correlation-based diagnostics that distinguish gauge-dominated Coulomb phases from states with substantial matter-field and short-range contributions. Distinct chiral flux sectors, though close in energy, exhibit markedly different degrees of emergent $\mathrm{U(1)}$ gauge-field dominance, reflected in the geometry and contrast of pinch-point singularities. Although these states are not competitive ground states of the nearest-neighbor Heisenberg model, they define a physically meaningful family of proximate chiral phases relevant to extended pyrochlore Hamiltonians.

cond-mat.str-el↗

arXiv2026

ACE-Brain-0.5: A Unified Embodied Foundational Model for Physical Agentic AI

Embodied AI is moving from isolated perception or action modules toward physical agents that understand, plan under goals, act through robot bodies, monitor progress, and improve from experience. Existing systems address this loop only in parts: end-to-end policies generate actions but often lack spatial reasoning, planning, and execution assessment, while robot-agent systems orchestrate tools or specialists but do not learn a shared representation. This fragmentation limits general Physical Agentic AI. We present ACE-Brain-0.5, a unified embodied foundation model that organizes robot intelligence into five coupled functions: spatial perception, decision making, embodied interaction, self-monitoring, and self-improvement. Built on ACE-Brain-0, which established spatial intelligence as a shared scaffold across robot platforms, ACE-Brain-0.5 extends an understanding-centric model into a closed-loop foundation model. A single 8B backbone instantiates the first four functions: grounding objects and affordances, reasoning over 3D and egocentric spatial relations, decomposing instructions into subgoals, generating navigation and manipulation actions, and estimating progress for verification and recovery. To unify these capabilities without cross-task interference, we introduce SSR+, which extends Scaffold-Specialize-Reconcile with a Reactivate stage after task-vector merging. The fifth function, self-improvement, is realized by a companion framework that updates external execution state, including task schemas, spatial memory, and failure-recovery cases, from rollouts. Across fifteen benchmarks, ACE-Brain-0.5 improves over ACE-Brain-0 on 14 of 18 spatial perception and grounding benchmarks, achieves competitive navigation and manipulation performance, and provides strong progress estimation in ID and OOD settings. Together, these results mark an early step toward general Physical Agentic AI.

cs.RO↗

arXiv2026

ACE-Ego-0: Unifying Egocentric Human and Robotic Data for VLA Pretraining

Vision-Language-Action (VLA) models benefit from large-scale and diverse embodied data, yet scaling robot trajectory collection is costly and labor-intensive. Recent advances show that large-scale egocentric human videos provide complementary real-world supervision in pretraining. However, joint training on human and robot data remains challenging due to divergences in action spaces, embodiment structures, temporal dynamics, and supervision quality. We introduce ACE-EGO-0, a unified VLA pretraining framework jointly leveraging heterogeneous data sources. To extract large-scale pretraining supervision from egocentric human videos, we build a scalable egocentric video-to-action pipeline that converts raw human videos into robot-format pseudo-action trajectories. To make these labels comparable with robot demonstrations, ACE-EGO-0 uses a unified action representation based on camera-space actions, morphology conditioning, and time-aligned action chunking. To robustly leverage noisy pseudo-action supervision from egocentric human videos, we formulate a reliability-aware training objective with a human auxiliary loss that concentrates supervision on reliable signals. We instantiate ACE-EGO-0 on 4.53K hours of robot and simulation data, together with 1.48K hours of pseudo-action-labeled egocentric human data. Experiments show that incorporating large-scale human supervision under reliability-aware weighting consistently improves both unified joint pretraining and supervised fine-tuning. ACE-EGO-0 achieves state-of-the-art performance on RoboCasa GR1 TableTop and RoboTwin 2.0, while demonstrating strong transfer to real-world bimanual manipulation.

cs.RO↗

arXiv2026

Beyond Ideal Instruction: A Comprehensive Framework for Evaluating LLMs in Realistic Interactions

Despite great advances in tool-use capabilities of large language models (LLMs), existing evaluation benchmarks struggle to fully align with real-world scenarios. Such benchmarks mostly rely on simulated idealized user assumptions and lacks experience-oriented evaluation. These limitations fail to account for the ambiguity, uncooperative behaviors, and shifting intentions characteristic of real-world users. To fill this gap, we propose RUT-Bench, a dedicated benchmark designed to assess LLMs under diverse Real-world User Tool calling scenarios. RUT-Bench supports high-fidelity simulations covering both ideal rational patterns and heterogeneous non-ideal behaviors across single-turn and multi-turn dialogues. We conduct comprehensive evaluations on 19 widely adopted open-source and proprietary LLMs using our benchmark. Experimental results reveal that no tested LLMs achieve an overall success rate above 40%, and nearly all of them experience noticeable performance drops when facing more complicated non-ideal user inputs. Our code and data is available at https://github.com/Miaow-Lab/RUT-Bench.

cs.CL↗

arXiv2026

STT-Arena: A More Realistic Environment for Tool-Using with Spatio-Temporal Dynamics

Large language models (LLMs) deployed in real-world agentic applications must be capable of replanning and adapting when mid-task disruptions invalidate their prior decisions. Existing dynamic benchmarks primarily measure whether LLMs can detect temporal changes in a timely manner, leaving the complementary challenge of adaptive replanning under spatio-temporal dynamics largely unexplored. We introduce STT-Arena (Spatio-Temporal Tool-Use Arena), a benchmark of 227 high-quality interactive tasks spanning nine spatio-temporal conflict types and four solvability levels. Each task is grounded in a realistic, executable environment equipped with injected spatio-temporal triggers that can abruptly invalidate an ongoing plan, forcing the model to detect the state shift and construct a revised execution strategy. Extensive evaluation of frontier LLMs reveals that even the SOTA proprietary models, including Claude-4.6-Opus, achieves less than 40\% overall accuracies, highlighting the fundamental difficulty of spatio-temporal dynamic reasoning. Systematic analysis of failure trajectories uncovers three recurring error modes of existing models: Stale-State Execution, Misdiagnosis of Dynamic Triggers, and Missing Post-Adaptation Verification. Guided by these findings, we propose an iterative trajectory refinement technique that eliminates these failure patterns from training data, and combine it with online RL to produce STT-Agent-4B which outperforms frontier LLMs on STT-Arena.

cs.CL↗