arXiv subjects

Ang Li

Publications and source records attributed to Ang Li.

At least 217 records · Page 12Linked to original sources

arXiv2025

SPARO: Surface-code Pauli-based Architectural Resource Optimization for Fault-tolerant Quantum Computing

Surface codes represent a leading approach for quantum error correction (QEC), offering a path towards universal fault-tolerant quantum computing (FTQC). However, efficiently implementing algorithms, particularly using Pauli-based computation (PBC) with lattice surgery, necessitates careful resource optimization. Prior work often employs static layouts and simplified error models. These typically fail to capture the full costs and dynamic nature of active computation, leading to resource bottlenecks and suboptimal architectural designs. To address this, we introduce SPARO. SPARO features a comprehensive logical error model based on a large corpus of numerical simulations encompassing active Pauli-based computation (PBC) operations-including Pauli product measurements (PPMs), idling qubits, and patch rotations. Our numerical models are integrated within an end-to-end compilation pipeline. SPARO analyzes algorithm-specific bottlenecks arising from constraints such as limited routing areas or magic-state factory throughput. SPARO then dynamically allocates available hardware resources, balancing compute, routing, and magic-state distillation, to minimize space-time overhead and logical error rates for specific workloads. Our simulations demonstrate that SPARO effectively identifies critical resource trade-offs. When evaluated on benchmark circuits, SPARO identifies resource configurations achieving up to 51.11% logical error rate reductions for 433-qubit ADDER circuits when compared to state-of-the-art static layouts using an identical total resource budget. This dynamic approach enables effective co-optimization of PBC execution and surface-code architectures, significantly improving overall resource efficiency. SPARO will be open sourced.

quant-ph↗

arXiv2025

Multi-Target Rydberg Gates via Spatial Blockade Engineering

Multi-target gates offer the potential to reduce gate depth in syndrome extraction for quantum error correction. Although neutral-atom quantum computers have demonstrated native multi-qubit gates, existing approaches that avoid additional control or multiple atomic species have been limited to single-target gates. We propose single-control-multi-target CZ^{\otimes N}) gates on a single-species neutral-atom platform that require no extra control and have gate durations comparable to standard CZ gates. Our approach leverages tailored interatomic distances to create an asymmetric blockade between the control and target atoms. Using a GPU-accelerated pulse synthesis protocol, we design smooth control pulses for CZZ and CZZZ gates, achieving fidelities of up to 99.55% and 99.24%, respectively, even in the presence of simulated atom placement errors and Rydberg-state decay. This work presents a practical path to implementing multi-target gates in neutral-atom systems, significantly reducing the resource overhead for syndrome extraction.

quant-ph↗

arXiv2025

TrapSIMD: SIMD-Aware Compiler Optimization for 2D Trapped-Ion Quantum Machines

Modular trapped-ion (TI) architectures offer a scalable quantum computing (QC) platform, with native transport behaviors that closely resemble the Single Instruction Multiple Data (SIMD) paradigm. We present FluxTrap, a SIMD-aware compiler framework that establishes a hardware-software co-design interface for TI systems. FluxTrap introduces a novel abstraction that unifies SIMD-style instructions -- including segmented intra-trap shift SIMD (S3) and global junction transfer SIMD (JT-SIMD) operations -- with a SIMD-enriched architectural graph, capturing key features such as transport synchronization, gate-zone locality, and topological constraints. It applies two passes -- SIMD aggregation and scheduling -- to coordinate grouped ion transport and gate execution within architectural constraints. On NISQ benchmarks, FluxTrap reduces execution time by up to $3.82 \times$ and improves fidelity by several orders of magnitude. It also scales to fault-tolerant workloads under diverse hardware configurations, providing feedback for future TI hardware design.

quant-ph↗

arXiv2025

OneAdapt: Adaptive Compilation for Resource-Constrained Photonic One-Way Quantum Computing

Measurement-based quantum computing (MBQC), a.k.a. one-way quantum computing (1WQC), is a universal quantum computing model, which is particularly well-suited for photonic platforms. In this model, computation is driven by measurements on an entangled state, which serves as an intermediate representation (IR) between program and hardware. However, compilers on previous IRs lacks the adaptability to the resource constraint in photonic quantum computers. In this work, we propose a novel IR with new optimization passes. Based on this, it realizes a resource-adaptive compiler that minimizes the required hardware size and execution time while restricting the requirement for fusion devices within an adaptive limit. Moreover, our optimization can be integrated with Quantum Error Correction (QEC) to improve the efficiency of photonic fault-tolerant quantum computing (FTQC).

quant-ph↗

arXiv2025

Context Switching for Secure Multi-programming of Near-Term Quantum Computers

Multi-programming quantum computers improve device utilization and throughput. However, crosstalk from concurrent two-qubit CNOT gates poses security risks, compromising the fidelity and output of co-running victim programs. We design Zero Knowledge Tampering Attacks (ZKTAs), using which attackers can exploit crosstalk without knowledge of the hardware error profile. ZKTAs can alter victim program outputs in 40% of cases on commercial systems. We identify that ZKTAs succeed because the attacker's program consistently runs with the same victim program in a fixed context. To mitigate this, we propose QONTEXTS: a context-switching technique that defends against ZKTAs by running programs across multiple contexts, each handling only a subset of trials. QONTEXTS uses multi-programming with frequent context switching while identifying a unique set of programs for each context. This helps limit only a fraction of execution to ZKTAs. We enhance QONTEXTS with attack detection capabilities that compare the distributions from different contexts against each other to identify noisy contexts executed with ZKTAs. Our evaluations on real IBMQ systems show that QONTEXTS increases program resilience by three orders of magnitude and fidelity by 1.33$\times$ on average. Moreover, QONTEXTS improves throughput by 2$\times$, advancing security in multi-programmed environments.

cs.CR↗

arXiv2025

Towards Stepwise Domain Knowledge-Driven Reasoning Optimization and Reflection Improvement

Recently, stepwise supervision on Chain of Thoughts (CoTs) presents an enhancement on the logical reasoning tasks such as coding and math, with the help of Monte Carlo Tree Search (MCTS). However, its contribution to tasks requiring domain-specific expertise and knowledge remains unexplored. Motivated by the interest, we identify several potential challenges of vanilla MCTS within this context, and propose the framework of Stepwise Domain Knowledge-Driven Reasoning Optimization, employing the MCTS algorithm to develop step-level supervision for problems that require essential comprehension, reasoning, and specialized knowledge. Additionally, we also introduce the Preference Optimization towards Reflection Paths, which iteratively learns self-reflection on the reasoning thoughts from better perspectives. We have conducted extensive experiments to evaluate the advantage of the methodologies. Empirical results demonstrate the effectiveness on various legal-domain problems. We also report a diverse set of valuable findings, hoping to encourage the enthusiasm to the research of domain-specific LLMs and MCTS.

cs.AI↗

arXiv2025

ThermoStereoRT: Thermal Stereo Matching in Real Time via Knowledge Distillation and Attention-based Refinement

We introduce ThermoStereoRT, a real-time thermal stereo matching method designed for all-weather conditions that recovers disparity from two rectified thermal stereo images, envisioning applications such as night-time drone surveillance or under-bed cleaning robots. Leveraging a lightweight yet powerful backbone, ThermoStereoRT constructs a 3D cost volume from thermal images and employs multi-scale attention mechanisms to produce an initial disparity map. To refine this map, we design a novel channel and spatial attention module. Addressing the challenge of sparse ground truth data in thermal imagery, we utilize knowledge distillation to boost performance without increasing computational demands. Comprehensive evaluations on multiple datasets demonstrate that ThermoStereoRT delivers both real-time capacity and robust accuracy, making it a promising solution for real-world deployment in various challenging environments. Our code will be released on https://github.com/SJTU-ViSYS-team/ThermoStereoRT

cs.CV↗

arXiv2025

Exploring nuclear force with pulsar glitch observation

We connect nuclear forces to one of the most notable irregular behaviors observed in pulsars, already detected in approximately 6\% known pulsars, with increasingly accurate data expected from upcoming high-precision timing instruments on both ground and space. Built on Shang & Li (2021), we conduct a case study on the 2001 glitch of the Vela pulsar. For our purpose, we adopt the Relativistic Mean Field (RMF) model as the theoretical many-body framework to describe nuclear systems. We refit three representative RMF parameter sets (DD-ME2, PKDD, NL3), considering the uncertainties in nuclear matter saturation properties. Utilizing the resulting star structure, composition and nucleon properties in the medium obtained in a consistent manner, we calculate the pinning energy of superfluid vortex in the nuclear lattice in the inner crust. This leads to the evolution of associated pinning force that acts on the vortex, which can be confronted with observed glitch amplitude and short-time relaxation in the 2000 Vela glitch event, following the snowplow model of pulsar glitch. We discuss how the vortex configuration and pinning properties depend on the nuclear parameters, and find an interesting and dominant role of the nuclear symmetry energy slope on pinning strength.

nucl-th↗

arXiv2025

Role and Use of Race in AI/ML Models Related to Health

The role and use of race within health-related artificial intelligence and machine learning (AI/ML) models has sparked increasing attention and controversy. Despite the complexity and breadth of related issues, a robust and holistic framework to guide stakeholders in their examination and resolution remains lacking. This perspective provides a broad-based, systematic, and cross-cutting landscape analysis of race-related challenges, structured around the AI/ML lifecycle and framed through "points to consider" to support inquiry and decision-making.

cs.CY↗

arXiv2025

Agent S2: A Compositional Generalist-Specialist Framework for Computer Use Agents

Computer use agents automate digital tasks by directly interacting with graphical user interfaces (GUIs) on computers and mobile devices, offering significant potential to enhance human productivity by completing an open-ended space of user queries. However, current agents face significant challenges: imprecise grounding of GUI elements, difficulties with long-horizon task planning, and performance bottlenecks from relying on single generalist models for diverse cognitive tasks. To this end, we introduce Agent S2, a novel compositional framework that delegates cognitive responsibilities across various generalist and specialist models. We propose a novel Mixture-of-Grounding technique to achieve precise GUI localization and introduce Proactive Hierarchical Planning, dynamically refining action plans at multiple temporal scales in response to evolving observations. Evaluations demonstrate that Agent S2 establishes new state-of-the-art (SOTA) performance on three prominent computer use benchmarks. Specifically, Agent S2 achieves 18.9% and 32.7% relative improvements over leading baseline agents such as Claude Computer Use and UI-TARS on the OSWorld 15-step and 50-step evaluation. Moreover, Agent S2 generalizes effectively to other operating systems and applications, surpassing previous best methods by 52.8% on WindowsAgentArena and by 16.52% on AndroidWorld relatively. Code available at https://github.com/simular-ai/Agent-S.

cs.AI↗

arXiv2025

Carrollian propagator and amplitude in Rindler spacetime

We study the three-dimensional Carrollian field theory on the Rindler horizon which is dual to a bulk massless scalar field theory in the four-dimensional Rindler wedge. The Carrollian field theory could be mapped to a two-dimensional Euclidean field theory in the transverse plane by a Fourier transform. After defining the incoming and outgoing states at the future and past Rindler horizon, respectively, we construct the boundary-to-boundary and bulk-to-boundary propagators that are consistent with the bulk Green's function in the literature. We investigate the tree-level Carrollian amplitudes up to four points. The tree-level four-point Carrollian amplitude in $Φ^4$ theory has the same structure as the one-loop triangle Feynman integral in the Lee-Pomeransky representation with complex powers in the propagators and spacetime dimension. Moreover, the four-point Carrollian amplitude with a zero energy state inserted at infinity in $Φ^4$ theory is proportional to the three-point Carrollian amplitude in $Φ^3$ theory.

hep-th↗

arXiv2025

Why Representation Engineering Works: A Theoretical and Empirical Study in Vision-Language Models

Representation Engineering (RepE) has emerged as a powerful paradigm for enhancing AI transparency by focusing on high-level representations rather than individual neurons or circuits. It has proven effective in improving interpretability and control, showing that representations can emerge, propagate, and shape final model outputs in large language models (LLMs). However, in Vision-Language Models (VLMs), visual input can override factual linguistic knowledge, leading to hallucinated responses that contradict reality. To address this challenge, we make the first attempt to extend RepE to VLMs, analyzing how multimodal representations are preserved and transformed. Building on our findings and drawing inspiration from successful RepE applications, we develop a theoretical framework that explains the stability of neural activity across layers using the principal eigenvector, uncovering the underlying mechanism of RepE. We empirically validate these instrinsic properties, demonstrating their broad applicability and significance. By bridging theoretical insights with empirical validation, this work transforms RepE from a descriptive tool into a structured theoretical framework, opening new directions for improving AI robustness, fairness, and transparency.

cs.LG↗

arXiv2025

Unleashed from Constrained Optimization: Quantum Computing for Quantum Chemistry Employing Generator Coordinate Inspired Method

Hybrid quantum-classical approaches offer potential solutions to quantum chemistry problems, yet they often manifest as constrained optimization problems. Here, we explore the interconnection between constrained optimization and generalized eigenvalue problems through the Unitary Coupled Cluster (UCC) excitation generators. Inspired by the generator coordinate method, we employ these UCC excitation generators to construct non-orthogonal, overcomplete many-body bases, projecting the system Hamiltonian into an effective Hamiltonian, which bypasses issues such as barren plateaus that heuristic numerical minimizers often encountered in standard variational quantum eigensolver (VQE). Diverging from conventional quantum subspace expansion methods, we introduce an adaptive scheme that robustly constructs the many-body basis sets from a pool of the UCC excitation generators. This scheme supports the development of a hierarchical ADAPT quantum-classical strategy, enabling a balanced interplay between subspace expansion and ansatz optimization to address complex, strongly correlated quantum chemical systems cost-effectively, setting the stage for more advanced quantum simulations in chemistry.

quant-ph↗

arXiv2025

Speculate, then Collaborate: Fusing Knowledge of Language Models during Decoding

Large Language Models (LLMs) often excel in specific domains but fall short in others due to the limitations of their training. Thus, enabling LLMs to solve problems collaboratively by integrating their complementary knowledge promises to improve their performance across domains. To realize this potential, we introduce a novel Collaborative Speculative Decoding (CoSD) algorithm that enables efficient LLM knowledge fusion at test time without requiring additional model training. CoSD employs a draft model to generate initial sequences and an easy-to-learn rule or decision tree to decide when to invoke an assistant model to improve these drafts. CoSD not only enhances knowledge fusion but also improves inference efficiency, is transferable across domains and models, and offers greater explainability. Experimental results demonstrate that CoSD improves accuracy by up to 10\% across benchmarks compared to existing methods, providing a scalable and effective solution for LLM-based applications

cs.CL↗

arXiv2025

Prada: Black-Box LLM Adaptation with Private Data on Resource-Constrained Devices

In recent years, Large Language Models (LLMs) have demonstrated remarkable abilities in various natural language processing tasks. However, adapting these models to specialized domains using private datasets stored on resource-constrained edge devices, such as smartphones and personal computers, remains challenging due to significant privacy concerns and limited computational resources. Existing model adaptation methods either compromise data privacy by requiring data transmission or jeopardize model privacy by exposing proprietary LLM parameters. To address these challenges, we propose Prada, a novel privacy-preserving and efficient black-box LLM adaptation system using private on-device datasets. Prada employs a lightweight proxy model fine-tuned with Low-Rank Adaptation (LoRA) locally on user devices. During inference, Prada leverages the logits offset, i.e., difference in outputs between the base and adapted proxy models, to iteratively refine outputs from a remote black-box LLM. This offset-based adaptation approach preserves both data privacy and model privacy, as there is no need to share sensitive data or proprietary model parameters. Furthermore, we incorporate speculative decoding to further speed up the inference process of Prada, making the system practically deployable on bandwidth-constrained edge devices, enabling a more practical deployment of Prada. Extensive experiments on various downstream tasks demonstrate that Prada achieves performance comparable to centralized fine-tuning methods while significantly reducing computational overhead by up to 60% and communication costs by up to 80%.

cs.CR↗

arXiv2025

LLM Generated Persona is a Promise with a Catch

The use of large language models (LLMs) to simulate human behavior has gained significant attention, particularly through personas that approximate individual characteristics. Persona-based simulations hold promise for transforming disciplines that rely on population-level feedback, including social science, economic analysis, marketing research, and business operations. Traditional methods to collect realistic persona data face significant challenges. They are prohibitively expensive and logistically challenging due to privacy constraints, and often fail to capture multi-dimensional attributes, particularly subjective qualities. Consequently, synthetic persona generation with LLMs offers a scalable, cost-effective alternative. However, current approaches rely on ad hoc and heuristic generation techniques that do not guarantee methodological rigor or simulation precision, resulting in systematic biases in downstream tasks. Through extensive large-scale experiments including presidential election forecasts and general opinion surveys of the U.S. population, we reveal that these biases can lead to significant deviations from real-world outcomes. Our findings underscore the need to develop a rigorous science of persona generation and outline the methodological innovations, organizational and institutional support, and empirical foundations required to enhance the reliability and scalability of LLM-driven persona simulations. To support further research and development in this area, we have open-sourced approximately one million generated personas, available for public access and analysis at https://huggingface.co/datasets/Tianyi-Lab/Personas.

cs.CL↗

arXiv2025

Towards Efficient Mixture of Experts: A Holistic Study of Compression Techniques

Scaling large language models has driven remarkable advancements across various domains, yet the continual increase in model size presents significant challenges for real-world deployment. The Mixture of Experts (MoE) architecture offers a promising solution by dynamically selecting and activating only a subset of experts during inference, thus substantially reducing computational costs while preserving high performance. Despite these benefits, MoE introduces new inefficiencies, such as excessive parameters and communication overhead. In this work, we present a holistic study of compression techniques for Mixture of Experts to enhance both efficiency and scalability. While recent efforts have focused on Expert Trimming, which reduces the number of experts, these approaches still suffer from considerable communication and computational costs. To address this, we propose more aggressive strategies, such as Layer Drop, which removes entire MoE layers, and Block Drop, which eliminates transformer blocks. Surprisingly, these aggressive pruning techniques not only preserve model performance but also substantially improve computation and memory efficiency. Furthermore, beyond Expert Trimming, we also introduce Expert Slimming, which compresses individual experts to further boost performance and can be seamlessly integrated with Expert Trimming. Extensive experimental results demonstrate the effectiveness of our proposed methods-Layer Drop and Block Drop-along with the comprehensive recipe that integrates Expert Slimming and Expert Trimming, achieving a 6.05x speedup with 77.1% reduced memory usage while maintaining over 92% of performance on Mixtral-8x7B. Our code is released at https://github.com/CASE-Lab-UMD/Unified-MoE-Compression.

cs.LG↗

arXiv2025

Robust Full-Space Physical Layer Security for STAR-RIS-Aided Wireless Networks: Eavesdropper with Uncertain Location and Channel

A robust full-space physical layer security (PLS) transmission scheme is proposed in this paper considering the full-space wiretapping challenge of wireless networks supported by simultaneous transmitting and reflecting reconfigurable intelligent surface (STAR-RIS). Different from the existing schemes, the proposed PLS scheme takes account of the uncertainty on the eavesdropper's position within the 360$^\circ$ service area offered by the STAR-RIS. Specifically, the large system analytical method is utilized to derive the asymptotic expression of the average security rate achieved by the security user, considering that the base station (BS) only has the statistical information of the eavesdropper's channel state information (CSI) and the uncertainty of its location. To evaluate the effectiveness of the proposed PLS scheme, we first formulate an optimization problem aimed at maximizing the weighted sum rate of the security user and the public user. This optimization is conducted under the power allocation constraint, and some practical limitations for STAR-RIS implementation, through jointly designing the active and passive beamforming variables. A novel iterative algorithm based on the minimum mean-square error (MMSE) and cross-entropy optimization (CEO) methods is proposed to effectively address the established non-convex optimization problem with discrete variables. Simulation results indicate that the proposed robust PLS scheme can effectively mitigate the information leakage across the entire coverage area of the STAR-RIS-assisted system, leading to superior performance gain when compared to benchmark schemes encompassing traditional RIS-aided scheme.

cs.IT↗