arXiv subjects

Jiahao Yan

Publications and source records attributed to Jiahao Yan.

14 recordsLinked to original sources

arXiv2026

Analytic Theory of Phase Transitions in Optical Metamaterials

Optical metamaterials provide a versatile platform for engineering homogeneous electromagnetic media whose distinct phases are characterized by phase diagrams in constitutive-parameter space. However, existing studies of hyperbolicity, topological properties, and exceptional-point formation often rely on highly symmetric models or case-by-case numerical parameter scans, leaving a unified analytic framework that identifies phases and phase transitions directly from the constitutive tensors lacking. Here, we develop a general theory that yields exact analytic criteria for topological transitions, exceptional-point transitions, pinch-off Lifshitz transitions, and optical Lifshitz transitions in homogeneous media. Applying this framework to a tractable example of a gyroelectric medium with anisotropic chirality, we uncover exceptional rings and negative refraction induced by gyroelectric-chiral coupling. By enabling the exact determination of phase boundaries, our theory provides a predictive framework for discovering previously unexplored electromagnetic phases and offers new principles for the systematic design of optical metamaterials.

physics.optics↗

arXiv2026

EChO-Agent: Evidence Chain Orchestration Agent for Audio Reasoning

While LALMs show promise on audio question answering, they fail to focus on question-relevant segments of audio and provide a clear, checkable reasoning process when dealing with complex audio reasoning. Reinforcement learning and tool-augmented prompting can help models better relate questions to audio but lack a reliable way to understand, integrate, and self-verify audio segments. To address this gap, we present EChO-Agent, a modular agent framework that reformulates complex audio QA as a planning, tool execution, evidence integration, and answer verification workflow. Experiments on MMAR benchmark show EChO-Agent improves both accuracy and rubric scores over baseline and ablation studies show evidence integration is the key factor.

eess.AS↗

arXiv2026

AngstromPro: A software platform for STM data management, visualization and analysis

Modern scanning tunneling microscopy (STM) experiments increasingly generate multidimensional datasets that require coordinated data management, visualization, and analysis. We present AngstromPro, an open-source Python-based software platform designed to support these activities within a common interactive environment. Its application architecture organizes inapplication data through independent module instances, each combining a Workspace with a common structure and module-specific visualization and interaction. Scientific processing methods are registered independently of individual modules and operate on explicitly selected inputs through common execution services. A unified I/O subsystem maps heterogeneous ondisk formats to supported in-application data types. The plugin subsystem can add new modules, processing methods, I/O handlers, and in-application data types without modifying the core package. Representative workflows demonstrate the integration of established methods. In the registration example, the normalized cross-correlation maximum increases from 0.855 to 0.966. AngstromPro therefore provides an integrated software architecture for organizing, visualizing, processing, and extending STM data-analysis workflows.

cond-mat.supr-con↗

arXiv2025

MemMamba: Rethinking Memory Patterns in State Space Model

With the explosive growth of data, long-sequence modeling has become increasingly important in tasks such as natural language processing and bioinformatics. However, existing methods face inherent trade-offs between efficiency and memory. Recurrent neural networks suffer from gradient vanishing and explosion, making them hard to scale. Transformers can model global dependencies but are constrained by quadratic complexity. Recently, selective state-space models such as Mamba have demonstrated high efficiency with O(n) time and O(1) recurrent inference, yet their long-range memory decays exponentially. In this work, we conduct mathematical derivations and information-theoretic analysis to systematically uncover the memory decay mechanism of Mamba, answering a fundamental question: what is the nature of Mamba's long-range memory and how does it retain information? To quantify key information loss, we further introduce horizontal-vertical memory fidelity metrics that capture degradation both within and across layers. Inspired by how humans distill and retain salient information when reading long documents, we propose MemMamba, a novel architectural framework that integrates state summarization mechanism together with cross-layer and cross-token attention, which alleviates long-range forgetting while preserving linear complexity. MemMamba achieves significant improvements over existing Mamba variants and Transformers on long-sequence benchmarks such as PG19 and Passkey Retrieval, while delivering a 48% speedup in inference efficiency. Both theoretical analysis and empirical results demonstrate that MemMamba achieves a breakthrough in the complexity-memory trade-off, offering a new paradigm for ultra-long sequence modeling.

cs.LG↗

arXiv2025

ManuSearch: Democratizing Deep Search in Large Language Models with a Transparent and Open Multi-Agent Framework

Recent advances in web-augmented large language models (LLMs) have exhibited strong performance in complex reasoning tasks, yet these capabilities are mostly locked in proprietary systems with opaque architectures. In this work, we propose \textbf{ManuSearch}, a transparent and modular multi-agent framework designed to democratize deep search for LLMs. ManuSearch decomposes the search and reasoning process into three collaborative agents: (1) a solution planning agent that iteratively formulates sub-queries, (2) an Internet search agent that retrieves relevant documents via real-time web search, and (3) a structured webpage reading agent that extracts key evidence from raw web content. To rigorously evaluate deep reasoning abilities, we introduce \textbf{ORION}, a challenging benchmark focused on open-web reasoning over long-tail entities, covering both English and Chinese. Experimental results show that ManuSearch substantially outperforms prior open-source baselines and even surpasses leading closed-source systems. Our work paves the way for reproducible, extensible research in open deep search systems. We release the data and code in https://github.com/RUCAIBox/ManuSearch

cs.CL↗

arXiv2025

MetasurfaceViT: A generic AI model for metasurface inverse design

Metasurfaces, sub-wavelength artificial structures, can control light's amplitude, phase, and polar ization, enabling applications in efficient imaging, holograms, and sensing. Recent years, AI has witnessed remarkable progress and spurred scientific discovery. In metasurface design, optical inverse design has recently emerged as a revolutionary approach. It uses deep learning to create a nonlinear mapping between optical structures and functions, bypassing time-consuming traditional design and attaining higher accuracy. Yet, current deep-learning models for optical design face limitations. They often work only for fixed wavelengths and polarizations, and lack universality as input-output vector size changes may require retraining. There's also a lack of compatibility across different application scenarios. This paper introduces MetasurfaceViT, a revolutionary generic AI model. It leverages a large amount of data using Jones matrices and physics-informed data augmentation. By pre-training through masking wavelengths and polarization channels, it can reconstruct full-wavelength Jones matrices, which will be utilized by fine-tuning model to enable inverse design. Finally, a tandem workflow appended by a forward prediction network is introduced to evaluate performance. The versatility of MetasurfaceViT with high prediction accuracy will open a new paradigm for optical inverse design.

physics.optics↗

arXiv2023

Inertia of partial transpose of positive semidefinite matrices

We show that the partial transpose of $9\times 9$ positive semidefinite matrices do not have inertia (4,1,4) and (3,2,4). It solves an open problem in "LINEAR AND MULTILINEAR ALGEBRA, Changchun Feng et al, 2022". We apply our results to construct some inertia, as well as present the list of all possible inertia of partial transpose of $12\times 12$ positive semidefinite matrices.

quant-ph↗

arXiv2023

Design of multifunctional color routers with Kerker switching using generative adversarial networks

To achieve optoelectronic devices with high resolution and efficiency, there is a pressing need for optical structural units that possess an ultrasmall footprint yet exhibit strong controllability in both the frequency and spatial domains. For dielectric nanoparticles, the overlap of electric and magnetic dipole moments can scatter light completely forward or backward, which is called Kerker theory. This effect can expand to any multipoles and any directions, re-named as generalized Kerker effect, and realize controllable light manipulation at full space and full spectrum using well-designed dielectric structures. However, the complex situations of multipole couplings make it difficult to achieve structural design. Here, generative artificial intelligence (AI) is utilized to facilitate multi-objective-oriented structural design, wherein we leverage the concept of "combined spectra" that consider both spectra and direction ratios as labels. The proposed generative adversarial network (GAN) is named as DDGAN (double-discriminator GAN) which discriminates both images and spectral labels. Using trained networks, we achieve the simultaneous design for scattering color and directivities, RGB color routers, as well as narrowband light routers. Notably, all generated structures possess a footprint less than 600x600 nm indicating their potential applications in optoelectronic devices with ultrahigh resolution.

physics.optics↗

arXiv2022

Energy Dissipation and Asymmetric Excitation in Hybrid Waveguides for Routing and Coloring

The delivery of optical signals from an external light source to a nanoscale waveguide is highly important for the development of nanophotonic circuits. However, the efficient coupling of external light energy into nanophotonic components is difficult and still remains a challenge. Herein, we use an external silica nanofiber to light up an organic-inorganic hybrid nano-waveguide, namely a system composed of a polymer filament doped with MoS$_{2}$ quantum dots. Nanofiber-excited nano-waveguides in a crossed geometry are found to asymmetrically couple excitation signals along two opposite directions, with different energy dissipation resulting in different colors of the light emitted by MoS$_{2}$ quantum dots and collected from the waveguide terminals. Interestingly, rainbow-like light in the hybrid waveguide is achieved by three-in-one mixing of red, green, and blue components. This hetero-dimensional system of dots-in-waveguide represents a significant advance towards all-optical routing and full-color display in integrated nanophotonic devices.

physics.optics↗

arXiv2022

Observation of Full-Parameter Jones Matrix in Bilayer Metasurface

Metasurfaces, artificial 2D structures, have been widely used for the design of various functionalities in optics. Jones matrix, a 2*2 matrix with eight parameters, provides the most complete characterization of the metasurface structures in linear optics, and the number of free parameters (i.e., degrees of freedom, DOFs) in the Jones matrix determines the limit to what functionalities we can realize. Great efforts have been made to continuously expand the number of DOFs, and a maximal number of six has been achieved recently. However, the realization of 'holy grail' goal with eight DOFs (full free parameters) has been proven as a great challenge so far. Here, we show that by cascading two layer metasurfaces and utilizing the gradient descent optimization algorithm, a spatially varying Jones matrix with eight DOFs is constructed and verified numerically and experimentally in optical frequencies. Such ultimate control unlocks new opportunities to design optical functionalities that are unattainable with previously known methodologies and may find wide potential applications in optical fields.

physics.optics↗

arXiv2021

Atomically sharp interface enabled ultrahigh-speed, nonvolatile memory devices

Development of memory devices with ultimate performance has played a key role in innovation of modern electronics. As a mainstream technology nonvolatile memory devices have manifested high capacity and mechanical reliability, however current major bottlenecks include low extinction ratio and slow operational speed. Although substantial effort has been employed to improve their performance, a typical hundreds of micro- or even milli- second write time remains a few orders of magnitude longer than their volatile counterparts. We have demonstrated nonvolatile, floating-gate memory devices based on van der Waals heterostructures with atomically sharp interfaces between different functional elements, and achieved ultrahigh-speed programming/erasing operations verging on an ultimate theoretical limit of nanoseconds with extinction ratio up to 10^10. This extraordinary performance has allowed new device capabilities such as multi-bit storage, thus opening up unforeseen applications in the realm of modern nanoelectronics and offering future fabrication guidelines for device scale-up.

physics.app-ph↗

arXiv2021

A low-temperature scanning probe microscopy system with molecular beam epitaxy and optical access

A low-temperature ultra-high vacuum scanning probe microscopy (SPM) system with molecular beam epitaxy capability and optical access was conceived, built, and tested in our lab. The design of the whole system is discussed here, with special emphasis on some critical parts. We made an SPM scanner head with a modified Pan-type design, enclosed by a double-layer cold room under a bath type cryostat. The scanner head is very rigid, compatible with optical access paths, and can accommodate both scanning tunneling microscope (STM) tips and atomic force sensors. Two piezo-actuated focus-lens stages are mounted on the two sides of the cold room to couple light in and out. To demonstrate the system performance, we performed STM and scanning tunneling spectroscopy studies. The herringbone reconstruction and atomic structure of Au(111) surface were clearly resolved. The dI/dV spectra of an Au(111) surface were obtained at 5 K. In addition, a periodic 2D tellurium (Te) structure was grown on Au(111) surface using MBE.

physics.ins-det↗

arXiv2020

Seeking for low thermal conductivity atomic configurations in $\rm{Si_{0.5}Ge_{0.5}}$ alloys with Bayesian Optimization

The emergence of data-driven science has opened up new avenues for understanding the thermophysical properties of materials. For decades, alloys are known to possess very low thermal conductivity, but the extreme thermal conductivity can be achieved by alloying has never been identified. In this work, we combine the Bayesian optimization with a high throughput thermal conductivity calculation to search for the lowest thermal conductivity atomic configuration of $\rm{Si_{0.5}Ge_{0.5}}$ alloy. It is found layered structures are most beneficial for reducing the thermal conductivity among all atomic configurations, which is attributed to the strong branch-folding effect. Furthermore, the roles of interface roughness and layer thicknesses in producing the lowest thermal conductivity are investigated. Through another comprehensive search using Bayesian optimization, the layered structure with smooth interfaces and optimized layer thickness arrangement is identified as the optimal structure with the lowest thermal conductivity.

physics.app-ph↗

arXiv2017

Resonant Photon-Exciton Coupling in All-Semiconductor Heterostructures Composed of Silicon Nanosphere and Monolayer WS2

Tailoring and enhancing the interaction between light and matter is of great importance for both fundamental researches and future photonic and optoelectronic applications. Due to their high exciton oscillator strength and large exciton binding energy, two-dimensional atomic semiconducting transition metal dichalcogenides have recently emerged as an excellent platform for the strong photon-exciton interaction by integrating with optically resonant cavities. Here, we propose an all-semiconductor system composed of individual silicon nanospheres and monolayer WS2 and investigate the resonance coupling between these two constituents. By coating the silicon nanospheres with monolayer WS2, we demonstrate the strong resonance coupling between the magnetic dipole mode and A-exciton, evidenced by an anticrossing behavior in the scattering energy diagram with a Rabi splitting of 77 meV. Compared with the plasmonic analogues, the resonance coupling in all-semiconductor heterostructure is much stronger and less sensitive to the spacing between the silicon nanosphere core and WS2 shell. When the silicon nanospheres are placed onto the WS2 monolayer with a point contact, resonance coupling manifested by the quenching dips in the scattering spectra can also be observed at ambient conditions, which involves only a few excitons. Finally, resonance coupling in the all-semiconductor heterostructure can be active controlled by temperature scanning. Our findings suggest that this all-semiconductor heterostructure can be exploited for future on-chip nanophotonics associated with strong light-matter interactions.

physics.optics↗