arXiv subjects

Peng He

Publications and source records attributed to Peng He.

At least 55 records · Page 3Linked to original sources

arXiv2025

Floquet Amorphous Topological Orders in a One-dimensional Rydberg Glass

The topological orders in amorphous systems that lack crystalline symmetry have gained considerable attention recently. Here we propose the Floquet amorphous topological matter, among which the topological orders are explored in experimentally accessible one-dimensional array of randomly pointed Rydberg atoms with periodic driving. The topological properties are comprehensively characterized, considering both the single-particle and many-body perspectives. It is found that the periodic driving leads to rich topological phases of matter. At the single-particle level, we evaluate the real space winding numbers and polarization, revealing robust amorphous topological phases with 0-type and $π$-type edge modes. We show a structural disorder induced topological phase transition associated with localization transition in the nonequilibrium system. Remarkably, in the many-body case it is discovered that the amorphous topological order exists in the chain of hardcore bosons, captured by the topological entanglement entropy and the string order. Moreover, feasible experimental probe protocols are also elaborated.

cond-mat.mes-hall↗

arXiv2025

Enhancing LLM-Based Short Answer Grading with Retrieval-Augmented Generation

Short answer assessment is a vital component of science education, allowing evaluation of students' complex three-dimensional understanding. Large language models (LLMs) that possess human-like ability in linguistic tasks are increasingly popular in assisting human graders to reduce their workload. However, LLMs' limitations in domain knowledge restrict their understanding in task-specific requirements and hinder their ability to achieve satisfactory performance. Retrieval-augmented generation (RAG) emerges as a promising solution by enabling LLMs to access relevant domain-specific knowledge during assessment. In this work, we propose an adaptive RAG framework for automated grading that dynamically retrieves and incorporates domain-specific knowledge based on the question and student answer context. Our approach combines semantic search and curated educational sources to retrieve valuable reference materials. Experimental results in a science education dataset demonstrate that our system achieves an improvement in grading accuracy compared to baseline LLM approaches. The findings suggest that RAG-enhanced grading systems can serve as reliable support with efficient performance gains.

cs.CL↗

arXiv2025

Revisiting Communication Efficiency in Multi-Agent Reinforcement Learning from the Dimensional Analysis Perspective

In this work, we introduce a novel perspective, i.e., dimensional analysis, to address the challenge of communication efficiency in Multi-Agent Reinforcement Learning (MARL). Our findings reveal that simply optimizing the content and timing of communication at sending end is insufficient to fully resolve communication efficiency issues. Even after applying optimized and gated messages, dimensional redundancy and confounders still persist in the integrated message embeddings at receiving end, which negatively impact communication quality and decision-making. To address these challenges, we propose Dimensional Rational Multi-Agent Communication (DRMAC), designed to mitigate both dimensional redundancy and confounders in MARL. DRMAC incorporates a redundancy-reduction regularization term to encourage the decoupling of information across dimensions within the learned representations of integrated messages. Additionally, we introduce a dimensional mask that dynamically adjusts gradient weights during training to eliminate the influence of decision-irrelevant dimensions. We evaluate DRMAC across a diverse set of multi-agent tasks, demonstrating its superior performance over existing state-of-the-art methods in complex scenarios. Furthermore, the plug-and-play nature of DRMAC's key modules highlights its generalizable performance, serving as a valuable complement rather than a replacement for existing multi-agent communication strategies.

cs.MA↗

arXiv2025

EventMamba: Enhancing Spatio-Temporal Locality with State Space Models for Event-Based Video Reconstruction

Leveraging its robust linear global modeling capability, Mamba has notably excelled in computer vision. Despite its success, existing Mamba-based vision models have overlooked the nuances of event-driven tasks, especially in video reconstruction. Event-based video reconstruction (EBVR) demands spatial translation invariance and close attention to local event relationships in the spatio-temporal domain. Unfortunately, conventional Mamba algorithms apply static window partitions and standard reshape scanning methods, leading to significant losses in local connectivity. To overcome these limitations, we introduce EventMamba--a specialized model designed for EBVR tasks. EventMamba innovates by incorporating random window offset (RWO) in the spatial domain, moving away from the restrictive fixed partitioning. Additionally, it features a new consistent traversal serialization approach in the spatio-temporal domain, which maintains the proximity of adjacent events both spatially and temporally. These enhancements enable EventMamba to retain Mamba's robust modeling capabilities while significantly preserving the spatio-temporal locality of event data. Comprehensive testing on multiple datasets shows that EventMamba markedly enhances video reconstruction, drastically improving computation speed while delivering superior visual quality compared to Transformer-based methods.

cs.CV↗

arXiv2025

Redshift drift effect through the observation of HI 21cm signal with SKA

This study presents the findings of using the Square Kilometre Array (SKA) telescope to measure redshift drift via the HI 21cm signal, employing semi-annual observational interval within redshift around z $\sim$ 1 with main goal is to directly gauge the universe's expansion acceleration rate with millimeter-per-second (mm/s) precision. The SKA can detect over a billion HI 21cm emissions from individual galaxies to the redshift z $\sim$ 2 and thousands of absorption lines from Damped Lyman-alpha (DLA) systems against bright quasars to the redshift z $\sim$ 13, with the sensitivity limit of 100 mJy. By utilizing SKA's high spectral resolution settings (0.001, 0.002, 0.005, 0.01 Hz) to detect redshift drift, particularly focusing on the 0.001 and 0.002 Hz configuration, one aims to achieve the necessary mm/s in precision measurement by the 0.5-year observation period. The velocity drift rate, crucially determined by the two operational regimes within 0.01 to 0.21 mm/s and 0.031 to 0.17 mm/s, respectively, exceeds the theoretical accuracy limit of 1.28 mm/s. The analysis thoroughly restricts cosmological parameters related to dark energy using the Sandage-Loeb (SL) signal from the HI 21cm emission and absorption lines. It estimates $\rm H_0$ of about 70 km/s/Mpc, $\rm Ω_m$ near 0.3, with w close to -1, $\rm w_0$ around -1, and $\rm w_a$ approaching -0.1. These results strongly endorse the SL effect as an effective method for confirming cosmic acceleration and exploring the dark sector in real-time cosmology with the SKA.

astro-ph.CO↗

arXiv2024

M2I2: Learning Efficient Multi-Agent Communication via Masked State Modeling and Intention Inference

Communication is essential in coordinating the behaviors of multiple agents. However, existing methods primarily emphasize content, timing, and partners for information sharing, often neglecting the critical aspect of integrating shared information. This gap can significantly impact agents' ability to understand and respond to complex, uncertain interactions, thus affecting overall communication efficiency. To address this issue, we introduce M2I2, a novel framework designed to enhance the agents' capabilities to assimilate and utilize received information effectively. M2I2 equips agents with advanced capabilities for masked state modeling and joint-action prediction, enriching their perception of environmental uncertainties and facilitating the anticipation of teammates' intentions. This approach ensures that agents are furnished with both comprehensive and relevant information, bolstering more informed and synergistic behaviors. Moreover, we propose a Dimensional Rational Network, innovatively trained via a meta-learning paradigm, to identify the importance of dimensional pieces of information, evaluating their contributions to decision-making and auxiliary tasks. Then, we implement an importance-based heuristic for selective information masking and sharing. This strategy optimizes the efficiency of masked state modeling and the rationale behind information sharing. We evaluate M2I2 across diverse multi-agent tasks, the results demonstrate its superior performance, efficiency, and generalization capabilities, over existing state-of-the-art methods in various complex scenarios.

cs.MA↗

arXiv2024

Triangular norms on finite lattices

This article intends to characterize triangular norms on a finite lattice. We first give a method for generating a triangular norm on an atomistic lattice by the values of atoms. Then we prove that every triangular norm on a non-Boolean atomistic lattice is not left-continuous and $T_M$ is the uniquely left-continuous triangular norm on an atomistic Boolean lattice. Furthermore, we show that each atomistic Boolean lattice can be represented by a family of triangular norms on an atomistic lattice with the same number of atoms. Finally, we construct a triangular norm on a finite lattice by restricting a triangular norm on an extended atomistic lattice of the finite lattice to the finite lattice.

math.CO↗

arXiv2024

A Robust Multisource Remote Sensing Image Matching Method Utilizing Attention and Feature Enhancement Against Noise Interference

Image matching is a fundamental and critical task of multisource remote sensing image applications. However, remote sensing images are susceptible to various noises. Accordingly, how to effectively achieve accurate matching in noise images is a challenging problem. To solve this issue, we propose a robust multisource remote sensing image matching method utilizing attention and feature enhancement against noise interference. In the first stage, we combine deep convolution with the attention mechanism of transformer to perform dense feature extraction, constructing feature descriptors with higher discriminability and robustness. Subsequently, we employ a coarse-to-fine matching strategy to achieve dense matches. In the second stage, we introduce an outlier removal network based on a binary classification mechanism, which can establish effective and geometrically consistent correspondences between images; through weighting for each correspondence, inliers vs. outliers classification are performed, as well as removing outliers from dense matches. Ultimately, we can accomplish more efficient and accurate matches. To validate the performance of the proposed method, we conduct experiments using multisource remote sensing image datasets for comparison with other state-of-the-art methods under different scenarios, including noise-free, additive random noise, and periodic stripe noise. Comparative results indicate that the proposed method has a more well-balanced performance and robustness. The proposed method contributes a valuable reference for solving the difficult problem of noise image matching.

cs.CV↗

arXiv2024

Reconstruction of the dark energy scalar field potential by Gaussian process

Dark energy is believed to be responsible for the acceleration of the universe. In this paper, we reconstruct the dark energy scalar field potential $V(ϕ)$ using the Hubble parameter $H(z)$ through Gaussian Process analysis. Our goal is to investigate dark energy using various $H(z)$ datasets and priors. We find that the selection of prior and the $H(z)$ dataset significantly affects the reconstructed $V(ϕ)$. And we compare two models, Power Law and Free Field, to the reconstructed $V(ϕ)$ by computing the reduced chi-square. The results suggest that the models are generally in agreement with the reconstructed potential within a $3σ$ confidence interval, except in the case of Observational $H(z)$ data (OHD) with the Planck 18 (P18) prior. Additionally, we simulate $H(z)$ data to measure the effect of increasing the number of data points on the accuracy of reconstructed $V(ϕ)$. We find that doubling the number of $H(z)$ data points can improve the accuracy rate of reconstructed $V(ϕ)$ by 5$\%$ to 30$\%$.

astro-ph.CO↗

arXiv2024

InsigHTable: Insight-driven Hierarchical Table Visualization with Reinforcement Learning

Embedding visual representations within original hierarchical tables can mitigate additional cognitive load stemming from the division of users' attention. The created hierarchical table visualizations can help users understand and explore complex data with multi-level attributes. However, because of many options available for transforming hierarchical tables and selecting subsets for embedding, the design space of hierarchical table visualizations becomes vast, and the construction process turns out to be tedious, hindering users from constructing hierarchical table visualizations with many data insights efficiently. We propose InsigHTable, a mixed-initiative and insight-driven hierarchical table transformation and visualization system. We first define data insights within hierarchical tables, which consider the hierarchical structure in the table headers. Since hierarchical table visualization construction is a sequential decision-making process, InsigHTable integrates a deep reinforcement learning framework incorporating an auxiliary rewards mechanism. This mechanism addresses the challenge of sparse rewards in constructing hierarchical table visualizations. Within the deep reinforcement learning framework, the agent continuously optimizes its decision-making process to create hierarchical table visualizations to uncover more insights by collaborating with analysts. We demonstrate the usability and effectiveness of InsigHTable through two case studies and sets of experiments. The results validate the effectiveness of the deep reinforcement learning framework and show that InsigHTable can facilitate users to construct hierarchical table visualizations and understand underlying data insights.

cs.HC↗

arXiv2024

Realizing and detecting Stiefel-Whitney insulators in an optical Raman lattice

We propose a feasible scheme to realize a four-band Stiefel-Whitney insultor (SWI) with spin-orbit coupled ultracold atoms in an optical Raman lattice. Four selected spin states are coupled by carefully designed Raman lasers, to generate the desired spin-orbit interactions with spacetime inversion symmetry. We map out a phase diagram with respect to the experimental parameters, where a large topological phase region exists. We further present two distinct detection methods to resolve the non-abelian band topology, in both equilibrium and dynamical ways. The detection relies on the spin textures extracted from the time-of-flight imaging, showing the tomographic signatures in the ground states and long-time averaged patterns on certain submanifold via a bulk-surface duality. Our work paves a realistic way to explore novel real topology with quantum matters.

cond-mat.quant-gas↗

arXiv2023

Spectral2Spectral: Image-spectral Similarity Assisted Spectral CT Deep Reconstruction without Reference

Spectral computed tomography based on a photon-counting detector (PCD) attracts more and more attentions since it has the capability to provide more accurate identification and quantitative analysis for biomedical materials. The limited number of photons within narrow energy bins leads to imaging results of low signal-noise ratio. The existing supervised deep reconstruction networks for CT reconstruction are difficult to address these challenges because it is usually impossible to acquire noise-free clinical images with clear structures as references. In this paper, we propose an iterative deep reconstruction network to synergize unsupervised method and data priors into a unified framework, named as Spectral2Spectral. Our Spectral2Spectral employs an unsupervised deep training strategy to obtain high-quality images from noisy data in an end-to-end fashion. The structural similarity prior within image-spectral domain is refined as a regularization term to further constrain the network training. The weights of neural network are automatically updated to capture image features and structures within the iterative process. Three large-scale preclinical datasets experiments demonstrate that the Spectral2spectral reconstructs better image quality than other the state-of-the-art methods.

eess.IV↗

arXiv2023

Quantum circuit for measuring an operator's generalized expectation values and its applications to non-Hermitian winding numbers

We propose a general quantum circuit based on the swap test for measuring the quantity $\langle ψ_1 | A | ψ_2 \rangle$ of an arbitrary operator $A$ with respect to two quantum states $|ψ_{1,2}\rangle$. This quantity is frequently encountered in many fields of physics, and we dub it the generalized expectation as a two-state generalization of the conventional expectation. We apply the circuit, in the field of non-Hermitian physics, to the measurement of generalized expectations with respect to left and right eigenstates of a given non-Hermitian Hamiltonian. To efficiently prepare the left and right eigenstates as the input to the general circuit, we also develop a quantum circuit via effectively rotating the Hamiltonian pair $(H,-H^\dagger)$ in the complex plane. As applications, we demonstrate the validity of these circuits in the prototypical Su-Schrieffer-Heeger model with nonreciprocal hopping by measuring the Bloch and non-Bloch spin textures and the corresponding winding numbers under periodic and open boundary conditions (PBCs and OBCs), respectively. The numerical simulation shows that non-Hermitian spin textures building up these winding numbers can be well captured with high fidelity, and the distinct topological phase transitions between PBCs and OBCs are clearly characterized. We may expect that other non-Hermitian topological invariants composed of non-Hermitian spin textures, such as non-Hermitian Chern numbers, and even significant generalized expectations in other branches of physics would also be measured by our general circuit, providing a different perspective to study novel properties in non-Hermitian as well as other physics realized in qubit systems.

quant-ph↗

arXiv2023

Quantum quenches in a pseudo-Hermitian Chern insulator

We propose to uncover the topology of a pseudo-Hermitian Chern insulator by quantum quench dynamics. The Bloch Hamiltonian of the pseudo-Hermitian Chern insulator is defined in the basis of the q-deformed Pauli matrices, which are related to the representation of the deformed algebras. We show the bulk-surface duality of the pseudo-Hermitian phases, then further build a concrete relation between the static band topology and quench dynamics, in terms of the time-averaged spin textures. The results are also generalized into a fully nonequilibrium case where the postquench evolution is governed by a Floquet pseudo-Hermitian Hamiltonian. Furthermore, we propose a possible scheme to realize the seemingly challenging model in a bilayer lattice and detect the dynamics with a double-quench protocol.

cond-mat.mes-hall↗

arXiv2023

The pseudocomplementedness of modular lattices described by two 0-sublattices

In this article, we first characterize pseudocomplemented inductive modular lattices by using their two 0-sublattices. Then we use two 0-sublattices of a subgroup lattice to describe all locally cyclic abelian groups. In particular, we show that a locally cyclic abelian group can be characterized by its three subgroups.

math.GR↗

arXiv2022

New trends in quantum integrability: Recent experiments with ultracold atoms

Over the past two decades quantum engineering has made significant advances in our ability to create genuine quantum many-body systems using ultracold atoms. In particular, some prototypical exactly solvable Yang-Baxter systems have been successfully realized allowing us to confront elegant and sophisticated exact solutions of these systems with their experimental counterparts. The new experimental developments show a variety of fundamental one-dimensional (1D) phenomena, ranging from the generalized hydrodynamics to dynamical fermionization, Tomonaga-Luttinger liquids, collective excitations, fractional exclusion statistics, quantum holonomy, spin-charge separation, competing orders with high spin symmetry and quantum impurity problems. This article briefly reviews these developments and provides rigorous understanding of those observed phenomena based on the exact solutions while highlighting the uniqueness of 1D quantum physics. The precision of atomic physics realizations of integrable many-body problems continues to inspire significant developments in mathematics and physics while at the same time offering the prospect to contribute to future quantum technology.

cond-mat.quant-gas↗

arXiv2022

Network Pruning via Feature Shift Minimization

Channel pruning is widely used to reduce the complexity of deep network models. Recent pruning methods usually identify which parts of the network to discard by proposing a channel importance criterion. However, recent studies have shown that these criteria do not work well in all conditions. In this paper, we propose a novel Feature Shift Minimization (FSM) method to compress CNN models, which evaluates the feature shift by converging the information of both features and filters. Specifically, we first investigate the compression efficiency with some prevalent methods in different layer-depths and then propose the feature shift concept. Then, we introduce an approximation method to estimate the magnitude of the feature shift, since it is difficult to compute it directly. Besides, we present a distribution-optimization algorithm to compensate for the accuracy loss and improve the network compression efficiency. The proposed method yields state-of-the-art performance on various benchmark networks and datasets, verified by extensive experiments. Our codes are available at: https://github.com/lscgx/FSM.

cs.CV↗

arXiv2022

Dissecting Service Mesh Overheads

Service meshes play a central role in the modern application ecosystem by providing an easy and flexible way to connect different services that form a distributed application. However, because of the way they interpose on application traffic, they can substantially increase application latency and resource consumption. We develop a decompositional approach and a tool, called MeshInsight, to systematically characterize the overhead of service meshes and to help developers quantify overhead in deployment scenarios of interest. Using MeshInsight, we confirm that service meshes can have high overhead -- up to 185% higher latency and up to 92% more virtual CPU cores for our benchmark applications -- but the severity is intimately tied to how they are configured and the application workload. The primary contributors to overhead vary based on the configuration too. IPC (inter-process communication) and socket writes dominate when the service mesh operates as a TCP proxy, but protocol parsing dominates when it operates as an HTTP proxy. MeshInsight also enables us to study the end-to-end impact of optimizations to service meshes. We show that not all seemingly-promising optimizations lead to a notable overhead reduction in realistic settings.

cs.DC↗