arXiv subjects

Tao Liu

Publications and source records attributed to Tao Liu.

At least 37 records · Page 2Linked to original sources

arXiv2026

Composite-State Localization Beyond the External Landscape in Non-Hermitian Quasicrystals

A composite excitation need not inherit the localization behavior of its constituents. We show that an interacting non-Hermitian quasiperiodic ladder realizes a controllable and reversible localization inversion between composite and unbound excitations, where internal configuration, rather than only the external potential, becomes a control parameter for localization. Opposite complex potentials on the two legs cancel at first order for a same-rung pair but act directly on separated particles, allowing extended composite states to persist while the unpaired sector becomes localized. A strong-coupling theory identifies the composite state as an emergent weakly modulated non-Hermitian quasicrystal generated by virtual unpaired configurations. Breaking the potential antisymmetry restores a direct modulation of the composite band and reverses the localization hierarchy. Engineering configuration-space pathways further stabilizes an extended composite band embedded within a localized continuum, the inverse of the conventional bound-state-in-the-continuum scenario. Our results establish internal configuration as a reversible control parameter for localization.

cond-mat.dis-nn↗

arXiv2026

Simple-OPD: Demystifying Warm-up for On-policy Distillation

On-policy distillation (OPD) trains a student on its own rollouts with token-level supervision from teacher models, but its effectiveness can depend strongly on the warm-up stage before OPD. In this paper, we demystify warm-up for OPD from both data and training perspectives. For data, we find that effective warm-up relies on teacher-compatible chain-of-thought supervision, and that even incorrect teacher rollouts can provide comparable benefits to correct ones. This suggests that warm-up primarily transfers a teacher-compatible thinking pattern rather than merely correct answers. For training, we show that low-rank adaptation (LoRA) with a near-saturation training duration better balances in-domain adaptation and out-of-distribution generalization than full-parameter SFT. Based on these findings, we propose Simple-OPD, a plug-and-play initialization method that warms up the student on teacher-generated CoT with LoRA before OPD. Experiments across diverse settings demonstrate the effectiveness and robustness of Simple-OPD.

cs.CL↗

arXiv2026

DenialRAG: Single-Document RAG Poisoning via Embedded Parametric Denial

Retrieval-augmented generation (RAG) systems are vulnerable to corpus poisoning: an attacker who inserts a crafted document into the retrieval corpus can steer the underlying large language model (LLM) toward an attacker-chosen wrong answer. Prior single-document attacks typically avoid explicitly naming and refuting the correct answer inside the poisoned passage. In this paper, we examine a complementary design and propose \emph{DenialRAG}, a single-document poisoning attack that explicitly names the correct answer, denies it, and presents an attacker-controlled explanation for favoring the wrong answer. By placing both the correct answer and the corresponding poisoned answer inside the same retrieved passage, DenialRAG embeds the conflict directly into the context seen by the generator. We evaluate DenialRAG against four published single-document poisoning attacks across three open-domain question-answering datasets, eight target LLMs from four vendors, and five inference-time defenses. The results show that attack effectiveness is strongly model-dependent: DenialRAG achieves the highest attack success rate (ASR) on all three Mistral-7B datasets and remains effective on several other target LLMs, while other attacks dominate in some model regimes. Defense results show meaningful ASR reductions but non-uniform protection, with each defense leaving residual ASR in some settings. Component-level and cross-model analyses further identify the embedded denial as the most influential tested component and show that different poisoning mechanisms lose effectiveness at different rates across model groups. Together, these results show that RAG poisoning risk cannot be fully characterized by a single attack family or a single target model.

cs.CR↗

arXiv2026

AttnLink: Turning Attention into Schema Links for Text-to-SQL

Schema linking is a critical component of Text-to-SQL systems, but existing approaches often trade off contextual modeling capacity, score-based controllability, and inference efficiency. We introduce AttnLink, an attention-based framework that converts LLMs' internal attention into continuous relevance scores for schema items. AttnLink extracts the attention from the generation-start position to candidate schema spans, enabling all candidates to be ranked in a single prefill pass without autoregressive decoding. We develop two variants: AttnLink-U, which directly probes pretrained attention without parameter updates, and AttnLink-S, which aligns the attention distribution with gold schema items through direct supervision. To improve coverage of multiple relevant schema items, AttnLink-S combines a set-mass objective with an adaptive probability-floor regularizer. The resulting scores support post-hoc precision-recall control through temperature scaling and cumulative-mass selection. Experiments on Spider, BIRD, and Spider2-SQLite show that AttnLink-S achieves mAP scores of 99.22%, 95.95%, and 83.29%, respectively, with millisecond-scale schema-linking latency. It also yields the best or tied-best execution accuracy for downstream SQL generation in seven of nine generator-dataset settings.

cs.CL↗

arXiv2026

The Second LoViF 2026 Challenge on Real-World All-in-One Image Restoration: Methods and Results

This paper presents a review of the second LoViF Challenge on Real-World All-in-One Image Restoration. The challenge aims to advance unified image restoration under diverse real-world degradation conditions, including blur, low-light, haze, rain, and snow. It provides a common benchmark for evaluating the restoration accuracy, robustness, and generalization capability of models across multiple degradation categories within a unified framework. The competition attracted 158 registered participants, and 20 teams were included in the final ranking after their submitted results were successfully reproduced and verified. This report provides a comprehensive analysis of the submitted solutions and corresponding results, highlighting recent advances in real-world all-in-one image restoration. The summarized methods and empirical findings reveal effective design strategies and establish an updated benchmark for future research in real-world low-level vision.

cs.CV↗

arXiv2026

Non-Monotonic Exciton Transport Efficiency in Multi-layer Ring-shaped Quantum Networks inspired by Photosystem I

Photosystem I (PSI) is one of the most efficient natural photosynthetic complexes and serves as a structural prototype for understanding excitation energy transfer (EET) and designing artificial light-harvesting systems. Here, we construct a multilayer elliptical-conical quantum network inspired by PSI to investigate how antenna architecture affects EET. Excitation dynamics are simulated using the quantum jump approach combined with coherent modified Redfield theory (QJA-CMRT). The transfer efficiency exhibits a pronounced nonmonotonic dependence on pigment density, reaching a maximum of 0.86 at x=6 in the five-layer network. The optimal configuration contains 91 pigments, close to the number of core antenna pigments in natural PSI. Our combined density-of-states and kinetic analyses reveal that this nonmonotonic dependence originates from a trade-off: at low pigment densities, enhanced band overlap promotes energy transfer, whereas at high densities, excessive exciton delocalization compromises funnel directionality and reduces efficiency; the optimal configuration is consequently achieved at x=6. These findings provide insights into PSI antenna organization and offer guidelines for designing efficient artificial light-harvesting systems.

cond-mat.soft↗

arXiv2026

Arbitrary Control of Non-Hermitian Skin Modes via Disorder and An Electric Field

The non-Hermitian skin effect (NHSE), characterized by the accumulation of a macroscopic number of bulk states at system boundaries, is a hallmark of non-Hermitian physics. However, in higher dimensions, achieving deterministic control over where skin modes accumulate remains a major challenge. Here, we propose a versatile route to program the skin-mode localization site in two-dimensional non-Hermitian lattices by combining disorder with a static electric field. While the electric field alone suppresses the NHSE in a clean system, the introduction of disorder induces transverse wave-packet transport perpendicular to the field. In nonreciprocal lattices, when the nonreciprocal hopping is misaligned with the electric field, the hopping component perpendicular to the field guides wave-packet propagation and produces boundary localization. By tuning the relative orientation between the electric field and the nonreciprocal hopping direction, the boundary localization position can be continuously and arbitrarily controlled. We further demonstrate distinct geometry-dependent manipulation of skin modes in reciprocal lattices, where controllable boundary localization emerges solely from the lattice geometry. Our results establish a robust and broadly applicable route to engineer boundary accumulation and directed transport along prescribed directions in two-dimensional non-Hermitian systems, enabling reconfigurable wave routing in classical platforms and programmable transport functionalities in quantum settings.

cond-mat.dis-nn↗

arXiv2026

Global Precision Limits in Critical Quantum Metrology: From Cramér-Rao to Ziv-Zakai

Critical quantum metrology with equilibrium states predicts quantum-enhanced sensitivity only in the vicinity of criticality, where large prior information about the parameter is required. By employing quantum Ziv-Zakai bounds, we derive a limit on the mean-square error in critical quantum metrology. For second-order quantum phase transitions, we show that the precision predicted by the Cramér-Rao bound offers no substantial improvement over the prior standard deviation. Thus, the critical quantum sensor's precision can only achieve a constant gain compared to the prior standard deviation, even without performing any measurement. We elucidate the fundamental limitation on the achievable precision in critical quantum metrology in the context of local sensing, even without considering state-preparation costs or noise. Thus, the super-Heisenberg-limited sensitivity at criticality arises from precise prior knowledge rather than a genuine gain due to criticality. Our work provides a practical framework for assessing critical quantum metrology and a routine for studying quantum sensing with many-body systems.

quant-ph↗

arXiv2026

Remembering George Smoot

George Smoot (1945-2025) was a major figure in observational cosmology, particularly through his role in the precision measurement of the CMB dipole and the discovery of the primordial CMB temperature anisotropy as part of the COBE team. This collection of 33 essays from collaborators, colleagues, students, and friends offers varied insights into the work, career, and life of George Smoot.

astro-ph.CO↗

arXiv2026

Sensing-Limited Control of Noiseless Linear Systems Under Nonlinear Observations

This paper investigates the fundamental information-theoretic limits for the control and sensing of noiseless linear dynamical systems subject to a broad class of nonlinear observations. We analyze the interactions between the control and sensing components by characterizing the minimum information flow required for stability. Specifically, we derive necessary conditions for mean-square observability and stabilizability, demonstrating that the average directed information rate from the state to the observations must exceed the intrinsic expansion rate of the unstable dynamics. Furthermore, to address the challenges posed by non-Gaussian distributions inherent to nonlinear observation channels, we establish sufficient conditions by imposing regularity assumptions, specifically log-concavity, on the system's probabilistic components. We show that under these conditions, the divergence of differential entropy implies the convergence of the estimation error, thereby closing the gap between information-theoretic bounds and estimation performance. By establishing these results, we unveil the fundamental performance limits imposed by the sensing layer, extending classical data-rate constraints to the more challenging regime of nonlinear observation models.

eess.SY↗

arXiv2026

MatchAttention: Embedding Explicit Matching Constraints into Attention for Efficient Stereo Matching

Standard attention mechanisms are not well suited to stereo matching. Global attention scales quadratically and provides no explicit matching constraint, while local attention is efficient but loses long-range correspondences. We propose MatchAttention, an attention mechanism that embeds an explicit matching constraint into attention by treating the relative position between a query and its matched key as a learnable component of attention sampling. Centering a small contiguous sampling window on this learnable relative position enforces the matching constraint and supports long-range correspondence at strictly linear attention complexity. A differentiable contiguous attention sampling (CAS) operator enables sub-pixel accuracy, and cascaded MatchAttention blocks iteratively refine the relative positions through residual connections. We instantiate MatchAttention as a hierarchical coarse-to-fine stereo network with two variants. MatchAttentionXL targets accuracy and MatchAttentionRT targets real-time edge inference. MatchAttentionXL achieves state-of-the-art accuracy on Middlebury V3 and top results across KITTI 2012/2015 and ETH3D. MatchAttentionRT runs at 9.3 ms on RTX 4060 Ti and 79.1 ms on Jetson Orin NX 16 GB at 1024 x 512, making it the first stereo model to deliver real-time edge inference without sacrificing zero-shot generalization. The code is available at https://github.com/TingmanYan/MatchAttention.

cs.CV↗

arXiv2026

EyeMVP: OCT-Informed Fundus Representation Learning via Paired CFP--OCT Pretraining

Color fundus photography (CFP) is the mainstay of large-scale retinal screening, but its diagnostic capacity is limited by the lack of depth-resolved structure, which optical coherence tomography (OCT) provides yet is less accessible at population scale. We present EyeMVP, a cross-modal retinal foundation model that uses paired CFP--OCT pretraining to learn OCT-informed CFP representations while requiring only CFP at inference. Pretrained on 674,893 same-eye same-day CFP--OCT triples from 112,642 patients across eight hospitals, EyeMVP uses cross-modal masked reconstruction to enrich CFP features with OCT-associated supervision, and combines source-constrained cross-attention with CFP-derived structural masks to accommodate the non-aligned geometry of en-face CFP and cross-sectional OCT. Across 15 dataset-level settings spanning classification and segmentation, under both full-data and few-shot regimes, EyeMVP performs on par with or better than representative retinal foundation models, with consistent gains on macular and optic-nerve tasks; it attains AUROCs of 0.923 for macular edema and 0.867 for myopic macular schisis, two conditions poorly resolved in CFP. In an exploratory reader study, EyeMVP surpasses junior and intermediate ophthalmologists but not seniors on macular edema, while exceeding all groups on myopic macular schisis. These results indicate that cross-modal reconstruction can enrich CFP representations with OCT-associated supervision, offering a practical route to stronger CFP-based screening.

cs.CV↗

arXiv2026

Tensor Train Decomposition-based 3D Implicit Full Waveform Inversion with Multi-scale Structural Similarity

Three-dimensional full waveform inversion (3DFWI) is a powerful technique for reconstructing high-resolution subsurface velocity models. However, its application is often limited by high memory requirements, computational costs, and sensitivity to cycle skipping. To overcome these challenges, we propose a novel tensor train (TT) decomposition-based 3D implicit full waveform inversion framework (TT-3DIFWI) combined with a multi-scale structural similarity (M-SSIM) objective function. In this framework, the 3D velocity model is represented by TT decomposition as a product of a series of low-rank core tensors. Then, three axis-specific implicit neural network representations (INR) based on one-dimensional vector coordinates as input are constructed to predict these core tensors, rather than directly predicting the velocity model. This INR reparameterization method based on TT decomposition can significantly reduce the memory consumption of INR training while maintaining the accuracy and resolution of the 3D velocity model reconstruction. Meanwhile, the low-rank structure of TT decomposition also ensures the structural consistency of the reconstruction velocity, thereby improving the accuracy and continuity of the inversion result. Furthermore, the M-SSIM objective function can compare the multi-scale structural differences between predicted and observed data, and utilize the ultra-low frequency features to reduce cycle skipping. Numerical experiments on synthetic and challenging land datasets demonstrate that TT-3DIFWI with M-SSIM achieves accurate and continuous velocity reconstruction, even with poor initial models or missing low-frequency data.

physics.geo-ph↗

arXiv2026

Nonlinear cascaded quantum network with giant emitters

Chiral quantum optics is central to developing scalable quantum networks, yet existing approaches rely predominantly on linear single-photon regimes. It remains unclear how to generate directional multiphotons. Here we show that giant emitters coupled to nonlinear quantum optical baths enable tunable directional correlated photons, revealing a mechanism for multiphoton directional emission. We demonstrate that the propagation phases of correlated photons, together with the coupling phases of giant emitters, can generate destructive interference in one direction while enhancing emission in the opposite direction, making directionality fully tunable. Building on this mechanism, we introduce a nonlinear cascaded quantum network paradigm mediated by correlated flying qubits, providing a configurable building block enabling distinct many-body applications beyond linear unidirectional setups. These results reveal a rich landscape for engineering multiphoton propagation and correlations through interference in giant emitter-nonlinear bath architectures, offering pathways for quantum networks and strongly correlated light-matter platforms.

quant-ph↗

arXiv2026

Theoretical and experimental studies of energy modulation to demodulation in seeded free-electron lasers

Laser manipulation plays a critical role in precisely tailoring relativistic electron beams through energy modulation, enabling the generation of coherent, intense, and ultrashort radiation in accelerator-based light sources such as synchrotron radiation facilities and free-electron lasers (FELs). However, laser-induced energy modulation inevitably degrades electron beam quality by increasing the energy spread, thereby limiting high-repetition-rate operation. Here, we investigate energy modulation and demodulation in a seeded FEL using two modulators separated by a tunable phase shifter. Analytical analysis and three-dimensional simulations show that a $π$ phase delay can nearly reverse the laser-beam interaction and substantially suppress the residual modulation. Diagnostics based on coherent undulator radiation and time-resolved measurements are established to characterize weak residual modulation, and a dedicated demodulation undulator is designed for controlled studies. Preliminary experiments performed at the Shanghai soft X-ray FEL facility using the existing seeding beamline demonstrate laser-induced energy-modulation suppression. Together with the analytical and numerical studies, these results establish a practical framework for investigating the transition from energy modulation to demodulation in seeded FELs, with potential applications in high-repetition-rate, fully coherent X-ray sources with improved preservation of electron beam quality.

physics.acc-ph↗

arXiv2026

V2V-Bench: A Comprehensive Benchmark for Video-to-Video Generation Evaluation

Video-to-video (V2V) generation is difficult to evaluate because outputs must both follow editing instructions and preserve frame-level correspondence with the source video, which existing T2V and I2V metrics do not capture. We introduce V2V-Bench, a 11-dimension benchmark organized into five categories: temporal alignment, structural fidelity, transformation quality, video quality, and semantic alignment. V2V-Bench pairs diverse source videos with challenging editing tasks and evaluates two commercial models, Grok Imagine and Gemini Veo3, and one open-source model, Open Sora 2. Results show complementary model strengths: Grok performs better on editing fidelity, while Veo3 achieves stronger visual quality. On six V2V-specific dimensions, V2V-Bench reaches a Spearman correlation of 0.905 with human judgments.

cs.CV↗

arXiv2026

LongSpace: Exploring Long-Horizon Spatial Memory from Perception to Recall in Video

Multimodal Large Language Models (MLLMs) have advanced image and video understanding and can increasingly handle longer visual inputs. Long-horizon tasks such as autonomous driving and robotic navigation require more than recognizing the current view, as models must remember and retrieve previously observed spatial layouts, routes, viewpoint changes, and object states. To evaluate this capability, we introduce LongSpace-Bench, a room-tour video benchmark for long-horizon spatial memory, covering scene perception, spatial relations, and spatial memory. In this work, we further propose LongSpace, a memory framework for long-video spatial reasoning. LongSpace models long videos as sequential chunks, incorporates 3D structural cues into early decoder layers, and constructs layer-aware memory for question-guided retrieval. Experiments on multiple spatial reasoning benchmarks show that LongSpace improves long-video spatial understanding, further demonstrating explicit spatial memory as a key capability for long-horizon video MLLMs.

cs.CV↗

arXiv2026

Elmes*: Automated Construction of Fine-Grained Evaluation Rubrics for Large Language Models in Long-Tail Educational Scenarios

Evaluating large language models (LLMs) for education requires measuring how models teach, not only what they know. Existing benchmarks emphasize domain-general correctness or depend on manually designed rubrics that scale poorly to long-tail pedagogical scenarios. We introduce Elmes*, an end-to-end framework for constructing, refining, and applying fine-grained scenario-specific rubrics. Elmes* combines a declarative multi-agent engine for teacher--student--judge interactions with SceneGen, a self-evolving module that co-optimizes evaluation criteria and test data from expert-defined pedagogical dimensions. Using Elmes*, we build Edu-330, covering 330 scenarios across 11 subjects, 3 grade bands, and 10 task types, with over 1{,}000 second-level indicators. Experiments on Edu-330 and four expert-authored gold-standard scenarios show that educational capability is multidimensional: top-tier LLMs differ mainly in creativity and values integration, knowledge-strong models may fail at Socratic scaffolding, and the education-specialized InnoSpark achieves the best human-evaluated average score. LLM judges preserve human-comparable rankings with much lower scoring variance, but exhibit judge-specific biases such as self-preference. Ablations show that expert-scored few-shot anchoring improves human--LLM alignment, while reasoning enforcement and greedy decoding are model-dependent. Elmes* thus provides scalable diagnostic infrastructure for pedagogically grounded LLM evaluation.

cs.LG↗