arXiv subjects

Na Li

Publications and source records attributed to Na Li.

At least 19 recordsLinked to original sources

arXiv2026

Colossal reversible conductivity switching by room-temperature oxygen-vacancy ordering in Aurivillius oxide films

Oxygen vacancies are central to the functionality of oxides, yet they typically exist as randomly distributed point defects, limiting the ability to precisely manipulate their collective behavior. Here, we report the room-temperature formation of a long-range-ordered oxygen-vacancy superstructure in single-crystalline Aurivillius-phase Bi2WO6 thin films via a mild nitrogen-plasma treatment. This structural transformation unlocks a colossal, reversible modulation of electrical conductivity by more than nine orders of magnitude, accompanied by a striking optical transition from transparent to black. Atomic-resolution imaging and spectroscopy reveal that the vacancies selectively order within the perovskite-like tungsten oxide layers, forming a coherent defect lattice that is absent in the pristine film. Oxygen-plasma treatment removes the vacancy superstructure and restores the initial state, whereas subsequent nitrogen-plasma treatment reconstructs it, enabling repeatable room-temperature switching between distinct structural, electronic and optical states. The phenomenon is also observed in another Aurivillius member, Bi2MoO6, suggesting its generality across the Aurivillius family. These findings establish a new paradigm for atomic-scale defect engineering - using gentle plasma chemistry to construct ordered defect lattices, opening avenues for reversible property modulation in complex oxides.

cond-mat.mtrl-sci↗

arXiv2026

ConvergeFlow: Language Flow with Provable Convergence to Token Embeddings

Recent advances in continuous diffusion and flow-based language models (LMs) have achieved performance competitive with discrete LMs. However, existing continuous frameworks still rely on decoders supervised with cross entropy (CE) because the flow trajectories are not guaranteed to terminate at valid token embeddings. Motivated by this limitation, we introduce \textbf{ConvergeFlow}, an embedding-space flow-based LM, which constrains the data predictor to the convex hull of token embeddings and trains it solely with the mean squared error objective induced by flow matching. Under suitable regularity conditions, we prove that the resulting flow converges to valid token embeddings despite errors in the data predictor, enabling direct token prediction without a CE-supervised decoder. We further develop three sampling mechanisms for controlling the trade-off between the generative perplexity and entropy. Experiments on OpenWebText demonstrate that ConvergeFlow achieves performance competitive with existing continuous and discrete diffusion LMs. These findings demonstrate the potential of the flow-based paradigm for language modeling. Our code is available at https://github.com/Na-Li66/ConvergeFlow.

cs.CL↗

arXiv2026

Scaling Muon for Diffusion Transformers

The matrix-aware optimizer Muon improves large model training by balancing updates across singular directions, yet its scaling behavior and end-to-end efficiency on large Diffusion Transformers (DiTs) remain unclear. We first establish Muon's scaling behavior on DiTs from 1.3B to 15B parameters, showing that its optimization and generative quality advantages over AdamW persist across model scales. However, at scale, the 5-step Newton--Schulz iteration (NS5) performed at every optimization step, together with full-momentum materialization, introduces substantial computation and communication overhead that can offset Muon's step-efficiency advantage. We introduce \emph{Periodic Row-wise Muon}, which performs a full NS5 spectral update once every $K$ steps and applies a low compute and communication cost row-wise constrained update based on the current momentum at the remaining steps. We further co-design a distributed implementation that operates directly on sharded momentum during non-refresh steps and accelerates spectral refreshes through bucketed all-gather and communication--computation overlap. Across all scales, Muon improves the best observed generative quality over AdamW by 12.9--19.1\%. Compared with vanilla Muon, Periodic Row-wise Muon remains within 0.5\% in best generative quality on the 1.3B--4B models and improves it by 4.5\% at 9B. It reduces optimizer time by 46.9--54.3\%, end-to-end step time by 15.7--24.3\%, and logical communication volume by 66.7\%, while reaching its respective best generative quality with 33.7--64.8\% less active training time. These results show that Periodic Row-wise Muon preserves Muon's generative quality advantage while translating it into end-to-end training efficiency for large DiTs.

cs.LG↗

arXiv2026

High fidelity control of superconducting qubits with optical transmitted signal

Superconducting circuits exhibit remarkable potential for constructing large-scale quantum simulation and computation systems, featuring numerous qubits, extended coherence time, and precise control. Nevertheless, the growing number of signal cables poses a challenge in dilution refrigerators due to space and heat load constraints. To overcome this issue, we experimentally implemented an optically-assisted transmission line as an alternative to coaxial cables. By modulating microwave signals on laser intensities at room temperature and regenerating the signals at a cryogenic plate within the dilution refrigerator, we demonstrated full control of superconducting qubits using photocurrent. We demonstrate and benchmark both single-qubit and two-qubit gates on frequency tunable transmon qubits, achieving fidelities of 99.915% $\pm$ 0.005% and 99.676% $\pm$ 0.041%, respectively, which have reached the requirement of the surface code.

quant-ph↗

arXiv2026

Indefinite Stochastic Linear-Quadratic Optimal Control Problem for a Markov Regime-Switching Model

This paper investigates an indefinite stochastic linear-quadratic (SLQ) control problem with parameters subject to Markov regime-switching. Based on the well-posedness of the SLQ problem, we introduce a relaxed compensator that extends SLQ control problems from the positive definite case to the indefinite case. We analyze the corresponding stochastic Hamiltonian system for both unconstrained and constrained control cases under the indefinite framework and derive the corresponding optimal open-loop controls. We further investigate the associated Riccati equations for both unconstrained and constrained control cases and derive the closed-loop feedback forms of optimal controls. We illustrate the theoretical results with an equity-bond asset allocation problem under un-constrained and non-negative control constraints. Numerical simulations validate the effectiveness of the theoretical framework and demonstrate its practical value in solving complex stochastic control problems with Markov regime-switching.

math.OC↗

arXiv2026

Learning Physical Interaction: A Survey of Tactile- and Force-aware Robot Learning

Physically grounded robot intelligence requires robots to perceive, reason about, and regulate their interactions with the physical world. This capability is particularly critical in contact-sensitive manipulation, where successful task execution depends not only on visual perception and motion generation, but also on force regulation and adaptive control. In this context, recent robot learning methods have made substantial progress by integrating force, tactile, vision, language, and proprioceptive sensing into learned manipulation policies. In parallel, many systems adopt multi-phase architectures that combine high-level policies, action-refinement modules, and low-level controllers to bridge semantic task understanding with reactive physical execution. Despite these advances, existing surveys have not explicitly reviewed force- and tactile-aware robot learning from a unified perspective that jointly captures multimodal sensing and multi-phase system design. This survey addresses this gap by proposing TF-ART, a Tactile/Force-Aware Robot learning Taxonomy for multimodal and multi-phase frameworks, which maps individual methods into a unified hierarchical structure. The framework characterizes how recent works organize observation modalities, encode and fuse heterogeneous sensory inputs, generate and refine actions across multiple phases, and connect learned policies to reactive robot-end control. Building on this methodological view, we further examine the task settings and infrastructure requirements of physical interaction, thereby integrating both algorithmic and practical perspectives on force- and tactile-aware robot learning.

cs.RO↗

arXiv2026

Model-Free Q-Learning for Infinite-Horizon Stochastic Linear Quadratic Problems with Regime Switching

This paper addresses infinite-horizon continuous-time stochastic linear quadratic optimal control problems with regime switching. We propose a paradigm shift from model-based design by adopting an adaptive dynamic programming approach, specifically developing on-policy and off-policy Q-learning algorithms that learn the optimal controller solely from online state trajectory data. The theoretical core of our work consists of a complete proof of the equivalence between the on- and off-policy architectures, alongside a rigorous analysis establishing the stability of the closed-loop system and the convergence of the algorithms to the optimal solution. For computational tractability, we implement these algorithms using vectorization and Kronecker product algebra. The theoretical results are corroborated by numerical case studies that clearly demonstrate the operational effectiveness and practical feasibility of the proposed model-free control strategy.

math.OC↗

arXiv2026

Provable diffusion-based posterior sampling for linear inverse problems via DDIM

Diffusion-based methods have achieved remarkable empirical success in solving inverse problems. However, many existing posterior samplers either lack rigorous theoretical guarantees or incur substantial computational overhead. We propose a simple and efficient algorithm, called \pddim, for solving linear inverse problems with diffusion priors via a DDIM-type sampler. Our method requires only lightweight, coordinate-wise modifications to the standard DDIM update, while explicitly incorporating the measurement model. The key idea is to perform posterior sampling separately along each singular direction of the measurement operator: for each direction, the sampler follows the learned diffusion prior when the observation signal-to-noise ratio (SNR) is below the corresponding diffusion SNR, and switches to a calibrated measurement-based predictor otherwise. We prove that the proposed sampler converges to the Bayesian posterior conditioned on the measurements. Empirical results show that the proposed sampler performs favorably against existing diffusion-based posterior samplers across a range of image restoration tasks, achieving the best performance on the majority of evaluation metrics considered. Overall, our results convert posterior sampling for noisy linear inverse problems to simple coordinate-wise DDIM updates, yielding an efficient, easy-to-implement algorithm with provable posterior consistency.

cs.LG↗

arXiv2026

When T2I Synthetic Data Backfires: Amplified Privacy Risks in Real-Synthetic Mix Training

To overcome data scarcity and privacy constraints in data collection, it has become standard practice across academia and industry to augment real training data with text-to-image (T2I)-generated synthetic data, a paradigm we term Real-Synthetic Mix-Training (RSMT). While substituting synthetic data for sensitive real samples is widely regarded as a means to mitigate privacy exposure of the substituted data, the risk to the remaining real samples that actively participate in training has remained largely unexamined. This work reveals, for the first time, that RSMT can substantially amplify privacy leakage of these real training samples. We establish a theoretical framework, RSMT Memorization Amplification, proving that incorporating synthetic data displaces real samples toward peripheral regions of the mixed feature space, in turn forcing the model to memorize them more aggressively. Guided by this foundation, we propose RSMixLeak to systematically assess this risk through membership inference attacks (MIAs). RSMixLeak comprises two variants depending on the adversary's capability. The non-adversarial variant audits a benign RSMT pipeline with an honest T2I provider, establishing a lower bound on the leakage induced by the intrinsic gap between real and T2I-generated data. The adversarial variant considers an adversary who controls the T2I model or contributes crafted data to the T2I provider, and deliberately enlarges this distributional gap on a target class via either high-level semantic attribute binding or imperceptible pixel-level coating, further amplifying leakage on real training data while improving downstream model utility. Motivated by these findings, we further propose a lightweight leakage propensity indicator computable from real data alone that reliably identifies high-risk datasets unsuitable for entering RSMT, as a self-assessable mitigation.

cs.CR↗

arXiv2026

A Single Diffusion-Policy Controller for Multi-Task Block Pushing with Zero-Shot Sim-to-Real Transfer

Diffusion policies have shown promising empirical performance in representing and learning complex maneuvers for robots using behavior cloning (BC). In this paper, we explore training diffusion policies from scratch using reinforcement learning (RL) for multi-task robotic manipulation. Specifically, we aim to train a single diffusion policy for block-pushing tasks with multiple shapes. The proposed framework features a simple policy loss function, which is a reweighted evidence lower bound used in BC-based diffusion policy training and can seamlessly serve as the policy learning module in RL algorithms. To address the exploration challenges arising from the absence of demonstrations, we incorporate reverse curriculum generation and objective-centric representations. Combined with the expressiveness of diffusion policies, our design supports learning of multi-task block-pushing policies in our sparse-reward simulation setting. We further evaluate whether the trained diffusion policy transfers in zero-shot to real-world tasks under varying environmental conditions including goal positions, block shapes, block weights and surface friction, providing evidence that this pipeline can transfer to our real-world block-pushing setup under the tested variations.

cs.RO↗

arXiv2026

EHR-MPC: Inference-Time Control for Sepsis Treatment with Generative Patient Digital Twins

Sepsis is a leading cause of mortality, yet optimal treatment policies remain contested. Existing reinforcement learning (RL) approaches learn fixed strategies for sepsis treatment, limiting adaptability to changing clinical objectives during inference. We propose EHRMPC, a framework that decouples learning patient dynamics from optimizing treatment by training a patient digital twin in the form of a generative electronic health record (EHR) model. The digital twin predicts clinical trajectories under interventions and enables model predictive control (MPC) to optimize treatments via inference-time planning over simulations. We evaluate EHR-MPC on a multicenter ICU sepsis cohort spanning 8 hospitals in the Mass General Brigham health system using both off-policy importance sampling and on-policy simulation-based evaluation. Relative to RL baselines, EHR-MPC achieves comparable off-policy performance and improved simulation performance. Unlike RL, this work frames sepsis treatment optimization as inference-time control over learned patient dynamics, establishing a general framework for decision making with generative clinical models.

stat.ML↗

arXiv2026

From Multiplicity to Vulnerability: Privacy Amplification Risk from One-Dataset-Multiple-Model Exposure

To efficiently exploit a valuable data source (e.g., facial or medical images), it is frequently harnessed to fulfill multiple learning objectives (e.g., facial recognition, age estimation, and race classification). Each trained model is then deployed as an independent API service for corresponding inference. However, the privacy risk introduced by this one-dataset-multiple-model (ODMM) paradigm is completely overlooked by the community. For the first time, this work reveals that the ODMM setting substantially amplifies privacy leakage. We establish a theoretical framework that proves that privacy leakage accumulates as more ODMM models are exposed, a phenomenon we term ODMM privacy composition. Guided by this theoretical foundation, we propose PRIME (Privacy Amplification RIsk from One-Dataset-Multiple-Model Exposure) to systematically assess this risk and quantify the resulting leakage using membership inference attacks (MIAs). Under black-box access to ODMM models, we design an aggregation mechanism that collectively captures carefully identified privacy signals leaked by individual ODMM models, and construct an attack meta-classifier over the aggregated meta-information to infer the membership status of a given sample jointly. Our results provide strong evidence that dataset reuse across ODMM models strikingly jeopardizes privacy, which is consistently evident across five privacy-sensitive image and textual benchmark datasets and diverse model architectures (from ResNet and ViT to Qwen3-1.7B), spanning three domains: facial analysis, medical imaging, and textual attribution analysis. While mitigations such as differential privacy can reduce the effectiveness of PRIME with trade-offs, our attack still consistently outperforms single-task MIAs.

cs.CR↗

arXiv2026

Inverse Optimal Control for Linear Quadratic Problem with Poisson Jumps: Model-Free Inverse Reinforcement Learning Approaches

This paper addresses the inverse optimal control (IOC) problem for stochastic linear systems subject to both Brownian motion and Poisson jumps, using an inverse reinforcement learning (IRL) framework. Given a target feedback gain from an expert, the objective is to identify an equivalent cost functional-specifically, the set of all cost weights-that yields this same gain. To solve this problem when system dynamics are unknown, we propose two model-free, off-policy IRL algorithms that operate entirely from data, circumventing the need to solve the generalized algebraic Riccati equation or compute the cost weights analytically. The first is an inverse Q-learning algorithm that constructs data-driven equations from expert demonstrations to compute the Q-function matrix, with equivalent cost weights updated algebraically and without requiring additional trajectory data. The second is a model-free off-policy inverse policy iteration algorithm that leverages data collected under an initial stabilizing policy, offering a complementary approach suited to different data availability scenarios. Crucially, by decoupling the data-collection behavior policies from the policies being iteratively updated, both algorithms can learn equivalent cost weights from sufficiently excited trajectories without identifying the system dynamics or jump intensity. Numerical simulations validate the effectiveness of the proposed methods.

math.OC↗

arXiv2026

Can neurons speak? Semantic narration of vision at single-cell resolution

Identifying what individual neurons encode in higher-order visual cortex is an open problem. Responses resist intuitive parameterization, and the deep-network embeddings used in their place are black boxes. Here, we introduce NEURRATOR, a framework that decodes spiking activity into free-form natural-language narration of the viewed scene at single-neuron resolution. A learned encoder maps spike trains from arbitrary subsets of simultaneously-recorded neurons into the patch-embedding space of a frozen CLIP, from which a multimodal language model and sparse autoencoder generates and validates a description with no language-side training. Applied to Neuropixel recordings of mouse visual cortex during natural-movie viewing, NEURRATOR narrates from thousands of neurons, singular cortical regions, local populations, or from a molecularly-defined cell-types. We use this property to (i) quantify how decoding fidelity scales with population size and cortical region, and (ii) "neurrate", in plain language, what individual neurons and genetically-tagged inhibitory cell-types contribute to visual representation. This recasts cell identity from a classification target into a functional probe of the visual system, providing a new unit of biological insights in neural systems.

q-bio.NC↗

arXiv2026

Learning to Adapt: Representation-Based Reinforcement Learning for Multi-Task Skill Transfer

Reinforcement learning has achieved remarkable success in learning complex control policies, yet its applicability remains limited due to sample inefficiency and poor generalization across tasks. In this work, we propose RepMT-SAC, a framework for multi-task RL that enables efficient knowledge sharing and robust transfer to new tasks. RepMT-SAC uses spectral MDP decomposition to capture transferable dynamics, structuring the value function into a task-agnostic core with a minimal task-specific adjustment. This design allows for strong zero-shot performance on in-distribution tasks and rapid few-shot adaptation to out-of-distribution tasks. We evaluate RepMT-SAC on quadcopter trajectory-following tasks across in-distribution and out-of-distribution contexts, demonstrating that it outperforms baselines by up to 30%.

cs.RO↗

arXiv2026

Study of $\Lambda^0_b \to \Lambda D$ decays with the rescattering mechanism

We employ the final-state rescattering mechanism to systematically investigate the non-leptonic decays $\Lambda_b^0 \to \Lambda D$ (where $D = D^0$, $\overline{D}^0$, $D_+$, $D_-$), which can help improve the experimental precision of the $CP$-violating phase angle $\gamma$. The framework integrates short-distance factorizable amplitudes with long-distance non-factorizable contributions arising from hadronic triangle rescattering diagrams. We find that the final-state rescattering contributions are essential, not only because they provide the strong phases necessary for $CP$ violation, but also because they enhance the $\overline{D}^0 \Lambda$ branching fraction by two orders of magnitude. Numerically, we calculate the branching ratios of the channels with different partial waves, as well as the corresponding $CP$-violating observables. In particular, the direct $CP$ asymmetries $a_{CP}^{\mathrm{dir}}(D_+) = 0.28^{+0.07}_{-0.11}$ and $a_{CP}^{\mathrm{dir}}(D_-) = -0.14^{+0.06}_{-0.04}$ are found to be very significant. These results provide a reliable theoretical benchmark and can be tested against future measurements from the LHCb experiment.

hep-ph↗

arXiv2026

SAID: Accelerating Diffusion-Based Language Models via Scaffold-Aware Iterative Decoding

Diffusion large language models (DLLMs) enable non-autoregressive generation by iteratively denoising corrupted token sequences with bidirectional context. Despite their ability to update multiple positions in parallel, inference remains costly due to the many denoising steps required for high-quality generation. We propose SAID, a Scaffold-Aware Iterative Decoding framework that accelerates DLLMs by reallocating computation across tokens. SAID first spends denoising computation on scaffold tokens to establish the coarse semantic structure, and then completes predictable detail tokens with fewer steps. We further adapt SAID to block-wise diffusion decoding and introduce Confidence-Hierarchical Layered Generation (CHLG), which assigns additional steps only to low-confidence tokens. Experiments on LLaDA-8B and LLaDA 1.5 across math, coding, and knowledge benchmarks show that SAID significantly accelerates DLLM inference with a maximum speedup of 9.1x while maintaining competitive performance. Our code is publicly available: https://github.com/TH-AI-Lab-PKU/SAID.

cs.CL↗

arXiv2026

VLESA: Vision-Language Embodied Safety Agent for Human Activity Monitoring

As AI systems increasingly assist humans in physical tasks, ensuring safety becomes paramount -- physical actions carry immediate and irreversible consequences that digital errors do not. We introduce the Vision-Language Embodied Safety Agent (VLESA), a framework that monitors human activities from egocentric video and triggers real-time safety interventions when dangerous actions are predicted. VLESA addresses intent-dependent safety where identical actions can be safe or dangerous depending on context. A dataset pairing egocentric frames with goal-conditioned safety annotations is introduced, enabling a goal-conditioned safety Q-filter trained via GRPO that evaluates actions with respect to inferred intent without retraining. On top of that, an intent-action prediction agent is proposed to jointly infer goals and predict future actions from video. On the ASIMOV-2.0 benchmark, VLESA achieves higher intervention accuracy at the exact ground-truth frame compared to baselines, while the GRPO-trained Q-filter improves action safety by over 41 percentage points through goal-conditioned constrained decoding. Code is available at https://github.com/HanjiangHu/VLESA.

cs.CV↗