arXiv subjects

Yi Li

Publications and source records attributed to Yi Li.

At least 217 records · Page 12Linked to original sources

arXiv2025

Low Spontaneous Brillouin Scattering in Anti-Resonant Hollow-Core Fibers in GHz Frequency Range

Brillouin light scattering (BLS) is a powerful experimental tool that can be used to get insights into the fundamental and applied properties of matter, like dispersions of quasiparticles in a solid, as well as their spatio-temporal dynamics. Many applications of light scattering favor the use of optical fibers in place of free-space optics. In this work, we compare the performance of anti-resonant hollow core fibers to that of conventional solid core fused silica fibers for BLS experiments in the GHz frequency range. Conventional fibers are barely suitable for low-noise measurements because of the spontaneous scattering of the photons on various phononic modes present in the core and cladding. In the case of the hollow-core fiber, we identify a range of discrete phononic modes and associate them with the various acoustic modes of the structure surrounding the hollow core using finite-element numerical simulations. The measured relative intensity of the spontaneous BLS signal from these modes is orders of magnitude smaller than that of a solid-core fiber, making anti-resonant hollow-core fibers one of the best solutions for the single-mode light guidance for BLS and potentially other low-noise photonic experiments.

cond-mat.mtrl-sci↗

arXiv2025

UniEval: Unified Holistic Evaluation for Unified Multimodal Understanding and Generation

The emergence of unified multimodal understanding and generation models is rapidly attracting attention because of their ability to enhance instruction-following capabilities while minimizing model redundancy. However, there is a lack of a unified evaluation framework for these models, which would enable an elegant, simplified, and overall evaluation. Current models conduct evaluations on multiple task-specific benchmarks, but there are significant limitations, such as the lack of overall results, errors from extra evaluation models, reliance on extensive labeled images, benchmarks that lack diversity, and metrics with limited capacity for instruction-following evaluation. To tackle these challenges, we introduce UniEval, the first evaluation framework designed for unified multimodal models without extra models, images, or annotations. This facilitates a simplified and unified evaluation process. The UniEval framework contains a holistic benchmark, UniBench (supports both unified and visual generation models), along with the corresponding UniScore metric. UniBench includes 81 fine-grained tags contributing to high diversity. Experimental results indicate that UniBench is more challenging than existing benchmarks, and UniScore aligns closely with human evaluations, surpassing current metrics. Moreover, we extensively evaluated SoTA unified and visual generation models, uncovering new insights into Univeral's unique values.

cs.CV↗

arXiv2025

$T\overline{T}$ Deformation: Introduction and Some Recent Advances

This review explores recent advances in the theory of $T\bar{T}$ deformation, an irrelevant yet solvable deformation of quantum field theories defined via the quadratic form of the energy-momentum tensor. It addresses classical and quantum aspects, highlighting significant developments across various fields, including field theory, holography, and string theory. Classically, $T\bar{T}$ deformation manifests through multiple geometric interpretations, notably random geometry, Jackiw-Teitelboim-like gravity, and uniform light-cone gauge frameworks. For quantum aspects, the deformation introduces notable features such as non-locality, UV-IR mixing, solvable renormalization structures, and intriguing modifications to correlation functions and entanglement properties. Furthermore, the paper examines the profound relationship between $T\bar{T}$ deformation and holography, particularly within the mixed boundary conditions/cutoff AdS holography proposal and holographic entanglement entropy. Connections to string theory through single-trace deformations and their holographic duals further reveal the deformed structure of the worldsheet. This review synthesizes recent developments and outlines potential directions for future research in the study of $T\bar{T}$-like deformation.

hep-th↗

arXiv2025

Transformer-Based Dual-Optical Attention Fusion Crowd Head Point Counting and Localization Network

In this paper, the dual-optical attention fusion crowd head point counting model (TAPNet) is proposed to address the problem of the difficulty of accurate counting in complex scenes such as crowd dense occlusion and low light in crowd counting tasks under UAV view. The model designs a dual-optical attention fusion module (DAFP) by introducing complementary information from infrared images to improve the accuracy and robustness of all-day crowd counting. In order to fully utilize different modal information and solve the problem of inaccurate localization caused by systematic misalignment between image pairs, this paper also proposes an adaptive two-optical feature decomposition fusion module (AFDF). In addition, we optimize the training strategy to improve the model robustness through spatial random offset data augmentation. Experiments on two challenging public datasets, DroneRGBT and GAIIC2, show that the proposed method outperforms existing techniques in terms of performance, especially in challenging dense low-light scenes. Code is available at https://github.com/zz-zik/TAPNet

cs.CV↗

arXiv2025

HAMSTER: Hierarchical Action Models For Open-World Robot Manipulation

Large foundation models have shown strong open-world generalization to complex problems in vision and language, but similar levels of generalization have yet to be achieved in robotics. One fundamental challenge is the lack of robotic data, which are typically obtained through expensive on-robot operation. A promising remedy is to leverage cheaper, off-domain data such as action-free videos, hand-drawn sketches or simulation data. In this work, we posit that hierarchical vision-language-action (VLA) models can be more effective in utilizing off-domain data than standard monolithic VLA models that directly finetune vision-language models (VLMs) to predict actions. In particular, we study a class of hierarchical VLA models, where the high-level VLM is finetuned to produce a coarse 2D path indicating the desired robot end-effector trajectory given an RGB image and a task description. The intermediate 2D path prediction is then served as guidance to the low-level, 3D-aware control policy capable of precise manipulation. Doing so alleviates the high-level VLM from fine-grained action prediction, while reducing the low-level policy's burden on complex task-level reasoning. We show that, with the hierarchical design, the high-level VLM can transfer across significant domain gaps between the off-domain finetuning data and real-robot testing scenarios, including differences on embodiments, dynamics, visual appearances and task semantics, etc. In the real-robot experiments, we observe an average of 20% improvement in success rate across seven different axes of generalization over OpenVLA, representing a 50% relative gain. Visual results, code, and dataset are provided at: https://hamster-robot.github.io/

cs.RO↗

arXiv2025

From Operator Product Expansion to Anomalous Dimensions

We propose a new method for computing the renormalization functions, which is based on the ideas of operator product expansion and large momentum expansion. In this method, the renormalization $Z$-factors are determined by the ultraviolet finiteness of Wilson coefficients in the dimensional regularization scheme. The ultraviolet divergence is extracted solely from two-point integrals at the large momentum limit. We develop this method in scalar field theories and establish a general framework for computing anomalous dimensions of fields, mass, couplings and composite operators. In particular, it is applied to the 6-dimensional cubic scalar theory and the 4-dimensional quartic scalar theory. We demonstrate this method by computing the anomalous dimension of the $ϕ^Q$ operator in cubic theory up to four loops for arbitrary $Q$, which is in agreement with the known result in the large $N$ limit. The idea of computing anomalous dimensions from the operator production expansion is general and can be extended beyond scalar theories. This is demonstrated through examples of the Gross-Neveu-Yukawa model with generic operators.

hep-th↗

arXiv2025

TS-Diff: Two-Stage Diffusion Model for Low-Light RAW Image Enhancement

This paper presents a novel Two-Stage Diffusion Model (TS-Diff) for enhancing extremely low-light RAW images. In the pre-training stage, TS-Diff synthesizes noisy images by constructing multiple virtual cameras based on a noise space. Camera Feature Integration (CFI) modules are then designed to enable the model to learn generalizable features across diverse virtual cameras. During the aligning stage, CFIs are averaged to create a target-specific CFI$^T$, which is fine-tuned using a small amount of real RAW data to adapt to the noise characteristics of specific cameras. A structural reparameterization technique further simplifies CFI$^T$ for efficient deployment. To address color shifts during the diffusion process, a color corrector is introduced to ensure color consistency by dynamically adjusting global color distributions. Additionally, a novel dataset, QID, is constructed, featuring quantifiable illumination levels and a wide dynamic range, providing a comprehensive benchmark for training and evaluation under extreme low-light conditions. Experimental results demonstrate that TS-Diff achieves state-of-the-art performance on multiple datasets, including QID, SID, and ELD, excelling in denoising, generalization, and color consistency across various cameras and illumination levels. These findings highlight the robustness and versatility of TS-Diff, making it a practical solution for low-light imaging applications. Source codes and models are available at https://github.com/CircccleK/TS-Diff

cs.CV↗

arXiv2025

Pump-induced magnon anticrossing due to three-magnon splitting and confluence

We present a plausible mechanism for achieving magnon-magnon level repulsion spectrum originating from the oscillation between the splitting and confluence in a three-magnon scattering process. When a magnetostatic mode on a YIG sphere is pumped by a microwave signal near the magnon resonance frequency with an increasing amplitude, the generated magnon condensate at half of the pumping frequency exerts a back-action to the original magnon mode. Such a strong nonlinear coupling manifests a striking feature of a 'bending effect' of the magnetostatic mode spectra, akin to the anti-crossing observed in a strongly coupled magnon-photon system.

cond-mat.mes-hall↗

arXiv2025

Experimental Proposal on Non-Abelian Aharonov-Bohm Caging Effect with a Single Trapped Ion

In the lattice system, when the synthetic flux reaches a $π$ phase along a closed loop under the synthetic gauge field, destructive interference occurs and gives rise to the localization phenomenon. This is known as the Aharonov-Bohm (AB) caging effect. It provides a powerful tool for the study of quantum transportation and dynamical effects. In the system where lattice sites possess internal structure and the underlying gauge field is non-Abelian, localization can also occur, forming the non-Abelian AB caging. Here, we propose an experimental scheme to synthesize non-Abelian gauge fields with a single trapped ion by coupling multiple internal levels and Fock states in its motion via laser fields. In contrast to the Abelian AB caging, we numerically observe that the non-Abelian AB caging occurs either when the interference matrix is nilpotent, or when the initial state is specifically set. Our experimental scheme broadens the study of localization phenomena and provides a novel tool for the study of non-Abelian physics.

quant-ph↗

arXiv2025

Parabolic frequency monotonicity for two nonlinear equations under Ricci flow

In this paper, we consider the parabolic frequency for positive solutions of two nonlinear parabolic equations under the Ricci flow on closed manifolds. We obtain the monotonicity of parabolic frequency for the solution of two nonlinear parabolic equations with bounded Ricci curvature, then we apply the parabolic frequency monotonicity to get some integral type Harnack inequalities and we use -K1 instead of the lower bound 0 of Ricci curvature from Theorem 4.3 in 16, where K1 is any positive constant.

math.DG↗

arXiv2025

Specification Mining for Smart Contracts with Trace Slicing and Predicate Abstraction

Smart contracts are computer programs running on blockchains to implement Decentralized Applications. The absence of contract specifications hinders routine tasks, such as contract understanding and testing. In this work, we propose a specification mining approach to infer contract specifications from past transaction histories. Our approach derives high-level behavioral automata of function invocations, accompanied by program invariants statistically inferred from the transaction histories. We implemented our approach as tool SMCON and evaluated it on eleven well-studied Azure benchmark smart contracts and six popular real-world DApp smart contracts. The experiments show that SMCON mines reasonably accurate specifications that can be used to enhance symbolic analysis of smart contracts achieving higher code coverage and up to 56% speedup, and facilitate DApp developers in maintaining high-quality documentation and test suites.

cs.SE↗

arXiv2025

Effective Index Construction Algorithm for Optimal $(k,η)$-cores Computation

Computing $(k,η)$-cores from uncertain graphs is a fundamental problem in uncertain graph analysis. UCF-Index is the state-of-the-art resolution to support $(k,η)$-core queries, allowing the $(k,η)$-core for any combination of $k$ and $η$ to be computed in an optimal time. However, this index constructed by current algorithm is usually incorrect. During decomposition, the key is to obtain the $k$-probabilities of its neighbors when the vertex with minimum $k$-probability is deleted. Current method uses recursive floating-point division to update it, which can lead to serious errors. We propose a correct and efficient index construction algorithm to address this issue. Firstly, we propose tight bounds on the $k$-probabilities of the vertices that need to be updated, and the accurate $k$-probabilities are recalculated in an on-demand manner. Secondly, vertices partitioning and progressive refinement strategy is devised to search the vertex with the minimum $k$-probability, thereby reducing initialization overhead for each $k$ and avoiding unnecessary recalculations. Finally, extensive experiments demonstrate the efficiency and scalability of our approach.

cs.DS↗

arXiv2025

Beating the break-even point with autonomous quantum error correction

Quantum error correction (QEC) is essential for practical quantum computing, as it protects fragile quantum information from errors by encoding it in high-dimensional Hilbert spaces. Conventional QEC protocols typically require repeated syndrome measurements, real-time feedback, and the use of multiple physical qubits for encoding. Such implementations pose significant technical complexities, particularly for trapped-ion systems, with high demands on precision and scalability. Here, we realize autonomous QEC with a logical qubit encoded in multiple internal spin states of a single trapped ion, surpassing the break-even point for qubit lifetime. Our approach leverages engineered spin-motion couplings to transfer error-induced entropy into motional modes, which are subsequently dissipated through sympathetic cooling with an ancilla ion, fully eliminating the need for measurement and feedback. By repetitively applying this autonomous QEC protocol under injected low-frequency noise, we extend the logical qubit lifetime to approximately 11.6 ms, substantially outperforming lifetime for both the physical qubit ($\simeq$0.9 ms) and the uncorrected logical qubit ($\simeq$0.8 ms), thereby beating the break-even point with autonomous protection of quantum information without measurement or post-selection. This work presents an efficient approach to fault-tolerant quantum computing that harnesses the intrinsic multi-level structure of trapped ions, providing a distinctive path toward scalable architectures and robust quantum memories with reduced overhead.

quant-ph↗

arXiv2025

MOSAIC: Magnonic Observations of Spin-dependent Axion-like InteraCtions

We introduce an array-scalable, magnon-based detector (MOSAIC) to search for the spin-dependent interactions of electron-coupled axion dark matter. These axions can excite single magnons in magnetic targets, such as the yttrium iron garnet (YIG) spheres used here, which are subsequently sensed by the detector. For MOSAIC, this sensing is implemented by coupling the magnons in the YIG spheres to magnetic-field-resilient single-electron charge-qubits, whose state is then interrogated with a quantum non-demolition measurement. Using standard superconducting fabrication techniques, MOSAIC can integrate many YIG sphere-qubit sensors, forming a large detector array. We outline the detector design and operation, and determine its sensitivity to axion dark matter. We find that a detector built with available technology will exceed the sensitivity of previous ferromagnetic haloscopes, and provides a platform where further improvements in performance would search for electron-coupled axion dark matter in unexplored parameter space.

hep-ph↗

arXiv2025

Edge-prompted spin waves probed via magneto-optical effect in the infrared

Recent advances in magnonics highlight the need for employing spin wave characteristics as new state variables, which is to be detected and mapped out with high precision in all-onchip, small scale devices. Spin wave modes that are prompted by the sample edges in planar structures are of particular fundamental importance to the future design and characterization of magnonic device functionalities. In this work, we demonstrate an infrared light-based, magneto-optical technique that concurrently functions both as a spectroscopic and an imaging tool. The detected precessional phase contrast can be directly used to construct the map of the spin wave wavefront in the continuous wave regime of spin-wave propagation and in the stationary state, without needing any optical reference path. We experimentally detect the edge spin wave modes of a Y3Fe5O12 film sample, and study the evolution and interference from edge dominated modes to their bulk counterparts. Distinct wavevector groups and propagation characteristics are observed for the edge and bulk spin wave modes. Further, the magnon dispersion relation in the infrared is obtained by Fourier transforming the wavefronts measured in real space, which is a huge advantage to the wavevector-resolved Brillouin light scattering (BLS) technique, where the angle of incidence has to be varied, and is also much less cumbersome than the phase-resolved BLS. Our results demonstrate that the infrared optical strobe light can serve as a versatile platform for magneto-optical probing of magnetization dynamics, with potential implications in investigating hybrid magnonic systems.

cond-mat.mes-hall↗

arXiv2025

Interventional Imbalanced Multi-Modal Representation Learning via $β$-Generalization Front-Door Criterion

Multi-modal methods establish comprehensive superiority over uni-modal methods. However, the imbalanced contributions of different modalities to task-dependent predictions constantly degrade the discriminative performance of canonical multi-modal methods. Based on the contribution to task-dependent predictions, modalities can be identified as predominant and auxiliary modalities. Benchmark methods raise a tractable solution: augmenting the auxiliary modality with a minor contribution during training. However, our empirical explorations challenge the fundamental idea behind such behavior, and we further conclude that benchmark approaches suffer from certain defects: insufficient theoretical interpretability and limited exploration capability of discriminative knowledge. To this end, we revisit multi-modal representation learning from a causal perspective and build the Structural Causal Model. Following the empirical explorations, we determine to capture the true causality between the discriminative knowledge of predominant modality and predictive label while considering the auxiliary modality. Thus, we introduce the $β$-generalization front-door criterion. Furthermore, we propose a novel network for sufficiently exploring multi-modal discriminative knowledge. Rigorous theoretical analyses and various empirical evaluations are provided to support the effectiveness of the innate mechanism behind our proposed method.

cs.LG↗

arXiv2025

On the Seven-loop Renormalization of Gross-Neveu Model

The presence of an infinite number of marginal four-fermion operators is a key characteristic of the two-dimensional Gross-Neveu model. In this study, we investigate the structure of UV divergences in this model, and by symmetry argument we found that the renormalizability only requires a subset of evanescent operators. We perform a 7-loop renormalization computation of beta function for the corresponding evanescent operator, and confirm its non-trivial contribution to UV divergences. By integrating infrared rearrangement, dimensional shifting, and large momentum expansion techniques, we systematically reduce the two-dimensional tensor integrals in the four-fermion correlation functions into four-dimensional scalar integrals. These scalar integrals are subsequently evaluated using the graphical function method, which marks the first application of the method to models with fermionic fields. Our result represents the first time that beta functions have been computed analytically beyond 5-loop in a model with spinning particles.

hep-th↗

arXiv2025

TickIt: Leveraging Large Language Models for Automated Ticket Escalation

In large-scale cloud service systems, support tickets serve as a critical mechanism for resolving customer issues and maintaining service quality. However, traditional manual ticket escalation processes encounter significant challenges, including inefficiency, inaccuracy, and difficulty in handling the high volume and complexity of tickets. While previous research has proposed various machine learning models for ticket classification, these approaches often overlook the practical demands of real-world escalations, such as dynamic ticket updates, topic-specific routing, and the analysis of ticket relationships. To bridge this gap, this paper introduces TickIt, an innovative online ticket escalation framework powered by Large Language Models. TickIt enables topic-aware, dynamic, and relationship-driven ticket escalations by continuously updating ticket states, assigning tickets to the most appropriate support teams, exploring ticket correlations, and leveraging category-guided supervised fine-tuning to continuously improve its performance. By deploying TickIt in ByteDance's cloud service platform Volcano Engine, we validate its efficacy and practicality, marking a significant advancement in the field of automated ticket escalation for large-scale cloud service systems.

cs.SE↗