Showing 1–50 of 53 results for author: Zhai, D

Spatial Annealing Smoothing for Efficient Few-shot Neural Rendering

Authors: Yuru Xiao, Xianming Liu, Deming Zhai, Kui Jiang, Junjun Jiang, Xiangyang Ji

Abstract: Neural Radiance Fields (NeRF) with hybrid representations have shown impressive capabilities in reconstructing scenes for view synthesis, delivering high efficiency. Nonetheless, their performance significantly drops with sparse view inputs, due to the issue of overfitting. While various regularization strategies have been devised to address these challenges, they often depend on inefficient assum… ▽ More Neural Radiance Fields (NeRF) with hybrid representations have shown impressive capabilities in reconstructing scenes for view synthesis, delivering high efficiency. Nonetheless, their performance significantly drops with sparse view inputs, due to the issue of overfitting. While various regularization strategies have been devised to address these challenges, they often depend on inefficient assumptions or are not compatible with hybrid models. There is a clear need for a method that maintains efficiency and improves resilience to sparse views within a hybrid framework. In this paper, we introduce an accurate and efficient few-shot neural rendering method named Spatial Annealing smoothing regularized NeRF (SANeRF), which is specifically designed for a pre-filtering-driven hybrid representation architecture. We implement an exponential reduction of the sample space size from an initially large value. This methodology is crucial for stabilizing the early stages of the training phase and significantly contributes to the enhancement of the subsequent process of detail refinement. Our extensive experiments reveal that, by adding merely one line of code, SANeRF delivers superior rendering quality and much faster reconstruction speed compared to current few-shot NeRF methods. Notably, SANeRF outperforms FreeNeRF by 0.3 dB in PSNR on the Blender dataset, while achieving 700x faster reconstruction speed. △ Less

Submitted 11 June, 2024; originally announced June 2024.

Mesh Denoising Transformer

Authors: Wenbo Zhao, Xianming Liu, Deming Zhai, Junjun Jiang, Xiangyang Ji

Abstract: Mesh denoising, aimed at removing noise from input meshes while preserving their feature structures, is a practical yet challenging task. Despite the remarkable progress in learning-based mesh denoising methodologies in recent years, their network designs often encounter two principal drawbacks: a dependence on single-modal geometric representations, which fall short in capturing the multifaceted… ▽ More Mesh denoising, aimed at removing noise from input meshes while preserving their feature structures, is a practical yet challenging task. Despite the remarkable progress in learning-based mesh denoising methodologies in recent years, their network designs often encounter two principal drawbacks: a dependence on single-modal geometric representations, which fall short in capturing the multifaceted attributes of meshes, and a lack of effective global feature aggregation, hindering their ability to fully understand the mesh's comprehensive structure. To tackle these issues, we propose SurfaceFormer, a pioneering Transformer-based mesh denoising framework. Our first contribution is the development of a new representation known as Local Surface Descriptor, which is crafted by establishing polar systems on each mesh face, followed by sampling points from adjacent surfaces using geodesics. The normals of these points are organized into 2D patches, mimicking images to capture local geometric intricacies, whereas the poles and vertex coordinates are consolidated into a point cloud to embody spatial information. This advancement surmounts the hurdles posed by the irregular and non-Euclidean characteristics of mesh data, facilitating a smooth integration with Transformer architecture. Next, we propose a dual-stream structure consisting of a Geometric Encoder branch and a Spatial Encoder branch, which jointly encode local geometry details and spatial information to fully explore multimodal information for mesh denoising. A subsequent Denoising Transformer module receives the multimodal information and achieves efficient global feature aggregation through self-attention operators. Our experimental evaluations demonstrate that this novel approach outperforms existing state-of-the-art methods in both objective and subjective assessments, marking a significant leap forward in mesh denoising. △ Less

Submitted 10 May, 2024; originally announced May 2024.

Fractional quantum anomalous Hall effect in a semimetal

Authors: Wenqi Yang, Dawei Zhai, Feng-Ren Fan, Wang Yao

Abstract: In the search of fractional quantum anomalous Hall (FQAH) effects, the conventional wisdom is to start from a flat Chern band isolated from the rest of the Hilbert space by bandgap, so that many-body interaction can be projected to a landscape that mimics a Landau level. Here we report the finding of FQAH in a 2D semimetal. Described by a 2π-flux dice lattice, the model features a flat band in tou… ▽ More In the search of fractional quantum anomalous Hall (FQAH) effects, the conventional wisdom is to start from a flat Chern band isolated from the rest of the Hilbert space by bandgap, so that many-body interaction can be projected to a landscape that mimics a Landau level. Here we report the finding of FQAH in a 2D semimetal. Described by a 2π-flux dice lattice, the model features a flat band in touching with a dispersive lower band, where the band touching is symmetry-protected from being gapped by electron interaction. At 1/3 and 2/3 filling of the gapless flat band, FQAH phases are found using density matrix renormalisation group calculations taking into accounts all bands. Symmetry breaking to gap the band touching can turn the semimetal into a Chern insulator while kee** the Chern band nearly flat, but counter-intuitively this suppresses the FQAH, as the gap opening introduces strong inhomogeneity to the quantum geometry. We show an optical scheme to realize the 2π-flux dice lattice for cold atoms. Our finding uncovers a new arena for the exploration of fractional quantum Hall physics in addition to the Landau levels and Chern insulators. △ Less

Submitted 2 May, 2024; originally announced May 2024.

Egret: Reinforcement Mechanism for Sequential Computation Offloading in Edge Computing

Authors: Haosong Peng, Yufeng Zhan, DiHua Zhai, Xiaopu Zhang, Yuanqing Xia

Abstract: As an emerging computing paradigm, edge computing offers computing resources closer to the data sources, hel** to improve the service quality of many real-time applications. A crucial problem is designing a rational pricing mechanism to maximize the revenue of the edge computing service provider (ECSP). However, prior works have considerable limitations: clients are static and are required to di… ▽ More As an emerging computing paradigm, edge computing offers computing resources closer to the data sources, hel** to improve the service quality of many real-time applications. A crucial problem is designing a rational pricing mechanism to maximize the revenue of the edge computing service provider (ECSP). However, prior works have considerable limitations: clients are static and are required to disclose their preferences, which is impractical in reality. However, previous works assume user privacy information to be known or consider the number of users in edge scenarios to be static. To address this issue, we propose a novel sequential computation offloading mechanism, where the ECSP posts prices of computing resources with different configurations to clients in turn. Clients independently choose which computing resources to purchase and how to offload based on their prices. Then Egret, a deep reinforcement learning-based approach that achieves maximum revenue, is proposed. Egret determines the optimal price and visiting orders online without considering clients' preferences. Experimental results show that the revenue of ECSP in Egret is only 1.29\% lower than Oracle and 23.43\% better than the state-of-the-art when the client arrives dynamically. △ Less

Submitted 29 April, 2024; v1 submitted 14 April, 2024; originally announced April 2024.

Comments: Submitted to IEEE TSC

SGCNeRF: Few-Shot Neural Rendering via Sparse Geometric Consistency Guidance

Authors: Yuru Xiao, Xianming Liu, Deming Zhai, Kui Jiang, Junjun Jiang, Xiangyang Ji

Abstract: Neural Radiance Field (NeRF) technology has made significant strides in creating novel viewpoints. However, its effectiveness is hampered when working with sparsely available views, often leading to performance dips due to overfitting. FreeNeRF attempts to overcome this limitation by integrating implicit geometry regularization, which incrementally improves both geometry and textures. Nonetheless,… ▽ More Neural Radiance Field (NeRF) technology has made significant strides in creating novel viewpoints. However, its effectiveness is hampered when working with sparsely available views, often leading to performance dips due to overfitting. FreeNeRF attempts to overcome this limitation by integrating implicit geometry regularization, which incrementally improves both geometry and textures. Nonetheless, an initial low positional encoding bandwidth results in the exclusion of high-frequency elements. The quest for a holistic approach that simultaneously addresses overfitting and the preservation of high-frequency details remains ongoing. This study introduces a novel feature matching based sparse geometry regularization module. This module excels in pinpointing high-frequency keypoints, thereby safeguarding the integrity of fine details. Through progressive refinement of geometry and textures across NeRF iterations, we unveil an effective few-shot neural rendering architecture, designated as SGCNeRF, for enhanced novel view synthesis. Our experiments demonstrate that SGCNeRF not only achieves superior geometry-consistent outcomes but also surpasses FreeNeRF, with improvements of 0.7 dB and 0.6 dB in PSNR on the LLFF and DTU datasets, respectively. △ Less

Submitted 17 June, 2024; v1 submitted 1 April, 2024; originally announced April 2024.

Layer Coherence Origin of Intrinsic Planar Hall Effect in 2D Limit

Authors: Huiyuan Zheng, Dawei Zhai, Cong Xiao, Wang Yao

Abstract: The intrinsic planar Hall effect has attracted intensive interest inspired by recent experiments. Existing theories of this effect require three dimensional orbital motion, or strong spin-orbit coupling of certain forms, which do not exist in van der Waals thin films. Here, we uncover a new origin of the planar Hall effect - as an intrinsic property of layer coherent electrons - that allows its pr… ▽ More The intrinsic planar Hall effect has attracted intensive interest inspired by recent experiments. Existing theories of this effect require three dimensional orbital motion, or strong spin-orbit coupling of certain forms, which do not exist in van der Waals thin films. Here, we uncover a new origin of the planar Hall effect - as an intrinsic property of layer coherent electrons - that allows its presence even in bilayer and trilayer atomically thin limit. As examples, we show that the effect can be triggered by strain and interlayer sliding respectively in twisted bilayer graphene and trilayer transition metal dichalcogenides, where the effect features rich tunability and even stronger magnitude than those induced by topological nodal structures in bulk materials. The layer mechanism also provides a new route towards quantized Hall response upon a topological phase transition induced by in-plane magnetic field. These results unveil the unexplored potential of quantum layertronics and moiré flat band for planar Hall transport. △ Less

Submitted 12 March, 2024; v1 submitted 26 February, 2024; originally announced February 2024.

Comments: 6 pages, 5 figures

Supersymmetry dictated topology in periodic gauge fields and realization in strained and twisted 2D materials

Authors: Dawei Zhai, Zuzhang Lin, Wang Yao

Abstract: Supersymmetry (SUSY) of Hamiltonian dictates double degeneracy between a pair of superpartners (SPs) transformed by supercharge, except at zero energy where modes remain unpaired in many cases. Here we explore a SUSY of complete isospectrum between SPs -- with paired zero modes -- realized by 2D electrons in zero-flux periodic gauge fields, which can describe twisted or periodically strained 2D ma… ▽ More Supersymmetry (SUSY) of Hamiltonian dictates double degeneracy between a pair of superpartners (SPs) transformed by supercharge, except at zero energy where modes remain unpaired in many cases. Here we explore a SUSY of complete isospectrum between SPs -- with paired zero modes -- realized by 2D electrons in zero-flux periodic gauge fields, which can describe twisted or periodically strained 2D materials. We find their low-energy sector containing zero (or threshold) modes must be topologically non-trivial, by proving that Chern numbers of the two SPs have a finite difference dictated by the number of zero modes and energy dispersion in their vicinity. In $30^\circ$ twisted bilayer (double bilayer) transition metal dichalcogenides subject to periodic strain, we find one SP is topologically trivial in its lowest miniband, while the twin SP of identical dispersion has a Chern number of $1$ ($2$), in stark contrast to time-reversal partners that have to be simultaneously trivial or nontrivial. For systems whose physical Hamiltonian corresponds to the square root of a SUSY Hamiltonian, such as twisted or strained bilayer graphene, we reveal that topological properties of the two SUSY SPs are transferred respectively to the conduction and valence bands, including the contrasted topology in the low-energy sector and identical topology in the high-energy sector. This offers a unified perspective for understanding topological properties in many flat-band systems described by such square-root models. Both types of SUSY systems provide unique opportunities for exploring correlated and topological phases of matter. △ Less

Submitted 8 May, 2024; v1 submitted 5 February, 2024; originally announced February 2024.

Comments: Examples and discussions on correlated phenomena are added

Dynamical Chiral Nernst Effect in Twisted Van der Waals Few Layers

Authors: Juncheng Li, Dawei Zhai, Cong Xiao, Wang Yao

Abstract: The Nernst effect is a fundamental thermoelectric conversion phenomenon that was deemed to be possible only in systems with magnetic field or magnetization. In this work, we propose a novel dynamical chiral Nernst effect that can appear in two-dimensional van der Waals materials with chiral structural symmetry in the absence of any magnetic degree of freedom. This unconventional effect is triggere… ▽ More The Nernst effect is a fundamental thermoelectric conversion phenomenon that was deemed to be possible only in systems with magnetic field or magnetization. In this work, we propose a novel dynamical chiral Nernst effect that can appear in two-dimensional van der Waals materials with chiral structural symmetry in the absence of any magnetic degree of freedom. This unconventional effect is triggered by time variation of an out-of-plane electric field, and has an intrinsic quantum geometric origin linked to not only the intralayer center-of-mass motion but also the interlayer coherence of electronic states. We demonstrate the effect in twisted homobilayer and homotrilayer transition metal dichalcogenides, where the strong twisted interlayer coupling leads to sizable intrinsic Nernst conductivities well within the experimental capacity. This work suggests a new route for electric control of thermoelectric conversion. △ Less

Submitted 24 January, 2024; originally announced January 2024.

Journal ref: Quantum Front 3, 11 (2024)

QoS-Aware 3D Coverage Deployment of UAVs for Internet of Vehicles in Intelligent Transportation

Authors: engfei Du, Tingyue Xiao, Haotong Cao, Daosen Zhai

Abstract: It is a challenging problem to characterize the air-to-ground (A2G) channel and identify the best deployment location for 3D UAVs with the QoS awareness. To address this problem, we propose a QoS-aware UAV 3D coverage deployment algorithm, which simulates the three-dimensional urban road scenario, considers the UAV communication resource capacity and vehicle communication QoS requirements comprehe… ▽ More It is a challenging problem to characterize the air-to-ground (A2G) channel and identify the best deployment location for 3D UAVs with the QoS awareness. To address this problem, we propose a QoS-aware UAV 3D coverage deployment algorithm, which simulates the three-dimensional urban road scenario, considers the UAV communication resource capacity and vehicle communication QoS requirements comprehensively, and then obtains the optimal UAV deployment position by improving the genetic algorithm. Specifically, the K-means clustering algorithm is used to cluster the vehicles, and the center locations of these clusters serve as the initial UAV positions to generate the initial population. Subsequently, we employ the K-means initialized grey wolf optimization (KIGWO) algorithm to achieve the UAV location with an optimal fitness value by performing an optimal search within the grey wolf population. To enhance the algorithm's diversity and global search capability, we randomly substitute this optimal location with one of the individual locations from the initial population. The fitness value is determined by the total number of vehicles covered by UAVs in the system, while the allocation scheme's feasibility is evaluated based on the corresponding QoS requirements. Competitive selection operations are conducted to retain individuals with higher fitness values, while crossover and mutation operations are employed to maintain the diversity of solutions. Finally, the individual with the highest fitness, which represents the UAV deployment position that covers the maximum number of vehicles in the entire system, is selected as the optimal solution. Extensive experimental results demonstrate that the proposed algorithm can effectively enhance the reliability and vehicle communication QoS. △ Less

Submitted 17 January, 2024; originally announced January 2024.

Layer Hall counterflow as a model probe of magic-angle twisted bilayer graphene

Authors: Jihang Zhu, Dawei Zhai, Cong Xiao, Wang Yao

Abstract: The recent constructions of flat moiré minibands in specifically twisted multilayer graphene and twisted transition metal dichalcogenides (TMDs) have facilitated the observation of strong correlations with a convenient tunability. These correlations in flat bands result in the band dispersion heavily influenced by carrier densities, leading to filling-dependent quasiparticle band renormalizations.… ▽ More The recent constructions of flat moiré minibands in specifically twisted multilayer graphene and twisted transition metal dichalcogenides (TMDs) have facilitated the observation of strong correlations with a convenient tunability. These correlations in flat bands result in the band dispersion heavily influenced by carrier densities, leading to filling-dependent quasiparticle band renormalizations. Particularly, in magic-angle twisted bilayer graphene (MATBG), the band structure--including the quasiparticle energy and wavefunction--is crucial in understanding the correlated properties. Previous theoretical studies have demonstrated the presence of a time-reversal-even charge Hall counterflow in response to a direct current (DC) electric field in twisted bilayers as chiral structures. In this study, we show that such layer Hall counterflow can serve as a sensitive probe for MATBG model parameters, which are currently ambiguous as a result of unavoidable structural relaxation and twist-angle disorder. We present the layer Hall counterflow and the associated in-plane magnetization for three different MATBG continuum models, based on which many-body interacting models have been widely applied to study strong correlations in MATBG. At the single-particle level, our findings indicate notable differences in layer-projected Hall conductivity, both in magnitude and sign, between different MATBG continuum models. Furthermore, our self-consistent Hartree calculations, performed on each of these single-particle continuum models, reveal renormalized layer-projected Hall conductivity by the self-consistent Hartree field. △ Less

Submitted 18 December, 2023; originally announced December 2023.

Journal ref: Phys. Rev. B 109, 155114 (2024)

On the Dynamics Under the Unhinged Loss and Beyond

Authors: Xiong Zhou, Xianming Liu, Hanzhang Wang, Deming Zhai, Junjun Jiang, Xiangyang Ji

Abstract: Recent works have studied implicit biases in deep learning, especially the behavior of last-layer features and classifier weights. However, they usually need to simplify the intermediate dynamics under gradient flow or gradient descent due to the intractability of loss functions and model architectures. In this paper, we introduce the unhinged loss, a concise loss function, that offers more mathem… ▽ More Recent works have studied implicit biases in deep learning, especially the behavior of last-layer features and classifier weights. However, they usually need to simplify the intermediate dynamics under gradient flow or gradient descent due to the intractability of loss functions and model architectures. In this paper, we introduce the unhinged loss, a concise loss function, that offers more mathematical opportunities to analyze the closed-form dynamics while requiring as few simplifications or assumptions as possible. The unhinged loss allows for considering more practical techniques, such as time-vary learning rates and feature normalization. Based on the layer-peeled model that views last-layer features as free optimization variables, we conduct a thorough analysis in the unconstrained, regularized, and spherical constrained cases, as well as the case where the neural tangent kernel remains invariant. To bridge the performance of the unhinged loss to that of Cross-Entropy (CE), we investigate the scenario of fixing classifier weights with a specific structure, (e.g., a simplex equiangular tight frame). Our analysis shows that these dynamics converge exponentially fast to a solution depending on the initialization of features and classifier weights. These theoretical results not only offer valuable insights, including explicit feature regularization and rescaled learning rates for enhancing practical training with the unhinged loss, but also extend their applicability to other loss functions. Finally, we empirically demonstrate these theoretical results and insights through extensive experiments. △ Less

Submitted 12 December, 2023; originally announced December 2023.

Comments: 62 pages, 19 figures

Interlayer electric multipoles induced by in-plane field from quantum geometric origins

Authors: Huiyuan Zheng, Dawei Zhai, Cong Xiao, Wang Yao

Abstract: We show that interlayer charge transfer in 2D materials can be driven by an in-plane electric field, giving rise to electrical multipole generation in linear and second order of in-plane field. The linear and nonlinear effects have quantum geometric origins in the Berry curvature and quantum metric respectively, defined in extended parameter spaces characteristic of layered materials. We elucidate… ▽ More We show that interlayer charge transfer in 2D materials can be driven by an in-plane electric field, giving rise to electrical multipole generation in linear and second order of in-plane field. The linear and nonlinear effects have quantum geometric origins in the Berry curvature and quantum metric respectively, defined in extended parameter spaces characteristic of layered materials. We elucidate their symmetry characters, and demonstrate sizable dipole and quadrupole polarizations respectively in twisted bilayers and trilayers of transition metal dichalcogenides. Furthermore, we show that the effect is strongly enhanced during the topological phase transition tuned by interlayer translation. The effects point to a new electric control on layer quantum degree of freedom. △ Less

Submitted 20 November, 2023; originally announced November 2023.

Comments: 13 pages, 4 figures

Backdoor Attacks Against Incremental Learners: An Empirical Evaluation Study

Authors: Yiqi Zhong, Xianming Liu, Deming Zhai, Junjun Jiang, Xiangyang Ji

Abstract: Large amounts of incremental learning algorithms have been proposed to alleviate the catastrophic forgetting issue arises while dealing with sequential data on a time series. However, the adversarial robustness of incremental learners has not been widely verified, leaving potential security risks. Specifically, for poisoning-based backdoor attacks, we argue that the nature of streaming data in IL… ▽ More Large amounts of incremental learning algorithms have been proposed to alleviate the catastrophic forgetting issue arises while dealing with sequential data on a time series. However, the adversarial robustness of incremental learners has not been widely verified, leaving potential security risks. Specifically, for poisoning-based backdoor attacks, we argue that the nature of streaming data in IL provides great convenience to the adversary by creating the possibility of distributed and cross-task attacks -- an adversary can affect \textbf{any unknown} previous or subsequent task by data poisoning \textbf{at any time or time series} with extremely small amount of backdoor samples injected (e.g., $0.1\%$ based on our observations). To attract the attention of the research community, in this paper, we empirically reveal the high vulnerability of 11 typical incremental learners against poisoning-based backdoor attack on 3 learning scenarios, especially the cross-task generalization effect of backdoor knowledge, while the poison ratios range from $5\%$ to as low as $0.1\%$. Finally, the defense mechanism based on activation clustering is found to be effective in detecting our trigger pattern to mitigate potential security risks. △ Less

Submitted 28 May, 2023; originally announced May 2023.

Super-Resolving Face Image by Facial Parsing Information

Authors: Chenyang Wang, Junjun Jiang, Zhiwei Zhong, Deming Zhai, Xianming Liu

Abstract: Face super-resolution is a technology that transforms a low-resolution face image into the corresponding high-resolution one. In this paper, we build a novel parsing map guided face super-resolution network which extracts the face prior (i.e., parsing map) directly from low-resolution face image for the following utilization. To exploit the extracted prior fully, a parsing map attention fusion blo… ▽ More Face super-resolution is a technology that transforms a low-resolution face image into the corresponding high-resolution one. In this paper, we build a novel parsing map guided face super-resolution network which extracts the face prior (i.e., parsing map) directly from low-resolution face image for the following utilization. To exploit the extracted prior fully, a parsing map attention fusion block is carefully designed, which can not only effectively explore the information of parsing map, but also combines powerful attention mechanism. Moreover, in light of that high-resolution features contain more precise spatial information while low-resolution features provide strong contextual information, we hope to maintain and utilize these complementary information. To achieve this goal, we develop a multi-scale refine block to maintain spatial and contextual information and take advantage of multi-scale features to refine the feature representations. Experimental results demonstrate that our method outperforms the state-of-the-arts in terms of quantitative metrics and visual quality. The source codes will be available at https://github.com/wcy-cs/FishFSRNet. △ Less

Submitted 6 April, 2023; originally announced April 2023.

Comments: TBIOM 2023

Crossed Nonlinear Dynamical Hall Effect in Twisted Bilayers

Authors: Cong Chen, Dawei Zhai, Cong Xiao, Wang Yao

Abstract: We propose a novel nonlinear dynamical Hall effect characteristic of layered materials with chiral symmetry, which is driven by the joint action of in-plane and time variation of out-of-plane ac fields $\boldsymbol{j}_{\text{H}}\sim\boldsymbol{\dot{E}_{\perp}}\times\boldsymbol{E}_{\parallel}$. A new band geometric quantity -- interlayer Berry connection polarizability, which probes a mixed quantum… ▽ More We propose a novel nonlinear dynamical Hall effect characteristic of layered materials with chiral symmetry, which is driven by the joint action of in-plane and time variation of out-of-plane ac fields $\boldsymbol{j}_{\text{H}}\sim\boldsymbol{\dot{E}_{\perp}}\times\boldsymbol{E}_{\parallel}$. A new band geometric quantity -- interlayer Berry connection polarizability, which probes a mixed quantum metric characteristic of layer hybridized electrons by twisted interlayer coupling, underlies this effect. When the two orthogonal fields have common frequency, their phase difference controls the on/off, direction and magnitude of the rectified Hall current. We show sizable effects in twisted homobilayer transition metal dichalcogenides and twisted bilayer graphene over broad range of twist angles. Our work opens the door to discovering mixed quantum metric responses unique to van der Waals stacking and concomitant applications under the nonlinear spotlight. △ Less

Submitted 4 February, 2024; v1 submitted 17 March, 2023; originally announced March 2023.

Comments: Add discussions of quantum metric contribution to the effect. To appear in Phys. Rev. Res

Journal ref: Phys. Rev. Research 6, L012059 (2024)

Single photon emitters with polarization and orbital angular momentum locking in monolayer semiconductors

Authors: Di Zhang, Sha Deng, Dawei Zhai, Wang Yao, Qizhong Zhu

Abstract: Excitons in monolayer transition metal dichalcogenide are endowed with intrinsic valley-orbit coupling between their center-of-mass motion and valley pseudospin. When trapped in a confinement potential, e.g., generated by strain field, we find that intralayer excitons are valley and orbital angular momentum (OAM) entangled. By tuning trap profile and external magnetic field, one can engi… ▽ More Excitons in monolayer transition metal dichalcogenide are endowed with intrinsic valley-orbit coupling between their center-of-mass motion and valley pseudospin. When trapped in a confinement potential, e.g., generated by strain field, we find that intralayer excitons are valley and orbital angular momentum (OAM) entangled. By tuning trap profile and external magnetic field, one can engineer the exciton states at ground state, and realize a series of valley-OAM entangled states. We further show that the OAM of excitons can be transferred to emitted photons, and these novel exciton states can naturally serve as polarization-OAM locked single photon emitters, which under certain circumstance become polarization-OAM entangled, highly tunable by strain trap and magnetic field. Our proposal demonstrates a novel scheme to generate polarization-OAM locked/entangled photons at nanoscale with high degree of integrability and tunability, pointing to exciting opportunities for quantum information applications. △ Less

Submitted 28 January, 2023; originally announced January 2023.

Comments: 10 pages, 5 figures

Journal ref: Nano Lett. 23, 3851 (2023)

Topological flat bands in strained graphene: substrate engineering and optical control

Authors: Md Tareq Mahmud, Dawei Zhai, Nancy Sandler

Abstract: The discovery of correlated phases in twisted moiré superlattices accelerated the search for low-dimensional materials with exotic properties. A promising approach uses engineered substrates to strain the material. However, designing substrates for tailored properties is hindered by the incomplete understanding of the relationship between substrate's shapes and electronic properties of the deposit… ▽ More The discovery of correlated phases in twisted moiré superlattices accelerated the search for low-dimensional materials with exotic properties. A promising approach uses engineered substrates to strain the material. However, designing substrates for tailored properties is hindered by the incomplete understanding of the relationship between substrate's shapes and electronic properties of the deposited materials. By analyzing effective models of graphene under periodic deformations with generic crystalline profiles, we identify strong $C_{2z}$ symmetry breaking as the critical substrate geometric feature for emerging energy gaps and quasi-flat bands. We find continuous strain profiles producing connected pseudo-magnetic field landscapes are important for band topology. We show that the resultant electronic and topological properties from a substrate can be controlled with circularly polarized light, which also offers unique signatures for identifying the band topology imprinted by strain. Our results can guide experiments on strain engineering for exploring interesting transport and topological phenomena. △ Less

Submitted 8 August, 2023; v1 submitted 18 January, 2023; originally announced January 2023.

Comments: Title changed, second part replaced by new contents. Supporting information will be freely available from the website of Nano Letters

Journal ref: Nano Lett. 23, 7725 (2023)

Local inversion-symmetry breaking in a bismuthate high-$T_c$ superconductor

Authors: S. Griffitt, M. Spaić, J. Joe, Z. Anderson, D. Zhai, M. J. Krogstad, R. Osborn, D. Pelc, M. Greven

Abstract: The doped perovskite BaBiO$_3$ exhibits a maximum superconducting transition temperature ($T_c$) of 34 K and was the first high-$T_c$ oxide to be discovered, yet pivotal questions regarding the nature of both the metallic and superconducting states remain unresolved. Although it is generally thought that superconductivity in the bismuthates is of the conventional s-wave type, the pairing mechanism… ▽ More The doped perovskite BaBiO$_3$ exhibits a maximum superconducting transition temperature ($T_c$) of 34 K and was the first high-$T_c$ oxide to be discovered, yet pivotal questions regarding the nature of both the metallic and superconducting states remain unresolved. Although it is generally thought that superconductivity in the bismuthates is of the conventional s-wave type, the pairing mechanism is still debated, with strong electron-phonon coupling and bismuth valence or bond disproportionation possibly playing a role. Here we use diffuse x-ray scattering and Monte Carlo modeling to study the local structure of Ba$_{1-x}$K$_x$BiO$_3$ across its insulator-metal boundary. We find no evidence for either long- or short-range disproportionation, which resolves a major conundrum, as disproportionation and the related polaronic effects are likely not relevant for the metallic and superconducting states. Instead, we uncover nanoscale structural correlations that break inversion symmetry, which has far-reaching implications for the electronic physics, including the pairing mechanism. This unexpected finding furthermore establishes that the bismuthates belong to the broader classes of materials with hidden spin-orbit coupling and a tendency towards inversion-breaking displacements. △ Less

Submitted 3 September, 2022; originally announced September 2022.

Comments: 23 pages, 3 main figures, extended data figures included

Shorted Micro-Waveguide Array for High Optical Transparency and Superior Electromagnetic Shielding in Ultra-Wideband Frequency Spectrum

Authors: Yuanlong Liang, Xianjun Huang, Jisheng Pan, Wencong Liu, Kui Wen, Duocai Zhai, Peng Shang, Peiguo Liu

Abstract: While functional materials with both light transmitting and electromagnetic shielding are highly desirable, only very few of them meet the stringent electromagnetic interference (EMI) shielding criteria for optoelectronic systems. Here, a design strategy of shorted micro-waveguides (SMWs) array to decouple the light transmission and EMI shielding is proposed and experimentally demonstrated. The ar… ▽ More While functional materials with both light transmitting and electromagnetic shielding are highly desirable, only very few of them meet the stringent electromagnetic interference (EMI) shielding criteria for optoelectronic systems. Here, a design strategy of shorted micro-waveguides (SMWs) array to decouple the light transmission and EMI shielding is proposed and experimentally demonstrated. The array of SMWs, consisting of cut-off metallic micro-waveguides and shorting indium tin oxide (ITO) continuous conductive film, exhibits high optical transmittance of 90.4% and superior EMI shielding effectiveness of 62.2 dB on average over ultra-wide frequency spectrum (0.2-1.3 GHz & 1.7-18 GHz). Compared to previously reported works, an improvement of 19 dB in average shielding effectiveness has been achieved under the same level of light transmission, and the shielding frequency spectrum has been significantly expanded. The working principle has been explained in depth and factors influencing the performance have been investigated for design optimization. △ Less

Submitted 23 August, 2022; originally announced August 2022.

Time-Reversal Even Charge Hall Effect from Twisted Interface Coupling

Authors: Dawei Zhai, Cong Chen, Cong Xiao, Wang Yao

Abstract: Under time-reversal symmetry, a linear charge Hall response is usually deemed to be forbidden by the Onsager relation. In this work, we discover a scenario for realizing a time-reversal even linear charge Hall effect in a non-isolated two-dimensional crystal allowed by time reversal symmetry. The restriction by Onsager relation is lifted by interfacial coupling with an adjacent layer, where the ov… ▽ More Under time-reversal symmetry, a linear charge Hall response is usually deemed to be forbidden by the Onsager relation. In this work, we discover a scenario for realizing a time-reversal even linear charge Hall effect in a non-isolated two-dimensional crystal allowed by time reversal symmetry. The restriction by Onsager relation is lifted by interfacial coupling with an adjacent layer, where the overall chiral symmetry requirement is fulfilled by a twisted stacking. We reveal the underlying band geometric quantity as the momentum-space vorticity of layer current. The effect is demonstrated in twisted bilayer graphene and twisted homobilayer transition metal dichalcogenides with a wide range of twist angles, which exhibit giant Hall ratios under experimentally practical conditions, with gate voltage controlled on-off switch. This work reveals intriguing Hall physics in chiral structures, and opens up a research direction of layertronics that exploits the quantum nature of layer degree of freedom to uncover exciting effects. △ Less

Submitted 8 March, 2023; v1 submitted 29 July, 2022; originally announced July 2022.

Comments: Supplementary Information included. To appear in Nature Communications

Journal ref: Nat. Commun. 14, 1961 (2023)

Prototype-Anchored Learning for Learning with Imperfect Annotations

Authors: Xiong Zhou, Xianming Liu, Deming Zhai, Junjun Jiang, Xin Gao, Xiangyang Ji

Abstract: The success of deep neural networks greatly relies on the availability of large amounts of high-quality annotated data, which however are difficult or expensive to obtain. The resulting labels may be class imbalanced, noisy or human biased. It is challenging to learn unbiased classification models from imperfectly annotated datasets, on which we usually suffer from overfitting or underfitting. In… ▽ More The success of deep neural networks greatly relies on the availability of large amounts of high-quality annotated data, which however are difficult or expensive to obtain. The resulting labels may be class imbalanced, noisy or human biased. It is challenging to learn unbiased classification models from imperfectly annotated datasets, on which we usually suffer from overfitting or underfitting. In this work, we thoroughly investigate the popular softmax loss and margin-based loss, and offer a feasible approach to tighten the generalization error bound by maximizing the minimal sample margin. We further derive the optimality condition for this purpose, which indicates how the class prototypes should be anchored. Motivated by theoretical analysis, we propose a simple yet effective method, namely prototype-anchored learning (PAL), which can be easily incorporated into various learning-based classification schemes to handle imperfect annotation. We verify the effectiveness of PAL on class-imbalanced learning and noise-tolerant learning by extensive experiments on synthetic and real-world datasets. △ Less

Submitted 23 June, 2022; originally announced June 2022.

Comments: ICML 2022

Learning Towards the Largest Margins

Authors: Xiong Zhou, Xianming Liu, Deming Zhai, Junjun Jiang, Xin Gao, Xiangyang Ji

Abstract: One of the main challenges for feature representation in deep learning-based classification is the design of appropriate loss functions that exhibit strong discriminative power. The classical softmax loss does not explicitly encourage discriminative learning of features. A popular direction of research is to incorporate margins in well-established losses in order to enforce extra intra-class compa… ▽ More One of the main challenges for feature representation in deep learning-based classification is the design of appropriate loss functions that exhibit strong discriminative power. The classical softmax loss does not explicitly encourage discriminative learning of features. A popular direction of research is to incorporate margins in well-established losses in order to enforce extra intra-class compactness and inter-class separability, which, however, were developed through heuristic means, as opposed to rigorous mathematical principles. In this work, we attempt to address this limitation by formulating the principled optimization objective as learning towards the largest margins. Specifically, we firstly define the class margin as the measure of inter-class separability, and the sample margin as the measure of intra-class compactness. Accordingly, to encourage discriminative representation of features, the loss function should promote the largest possible margins for both classes and samples. Furthermore, we derive a generalized margin softmax loss to draw general conclusions for the existing margin-based losses. Not only does this principled framework offer new perspectives to understand and interpret existing margin-based losses, but it also provides new insights that can guide the design of new tools, including sample margin regularization and largest margin softmax loss for the class-balanced case, and zero-centroid regularization for the class-imbalanced case. Experimental results demonstrate the effectiveness of our strategy on a variety of tasks, including visual classification, imbalanced classification, person re-identification, and face verification. △ Less

Submitted 23 June, 2022; originally announced June 2022.

Comments: ICLR 2022

Constructing Trajectory and Predicting Estimated Time of Arrival for Long Distance Travelling Vessels: A Probability Density-based Scanning Approach

Authors: Deqing Zhai, Xiuju Fu, Xiao Feng Yin, Haiyan Xu, Wanbing Zhang, Ning Li

Abstract: In this study, a probability density-based approach for constructing trajectories is proposed and validated through an typical use-case application: Estimated Time of Arrival (ETA) prediction given origin-destination pairs. The ETA prediction is based on physics and mathematical laws given by the extracted information of probability density-based trajectories constructed. The overall ETA predictio… ▽ More In this study, a probability density-based approach for constructing trajectories is proposed and validated through an typical use-case application: Estimated Time of Arrival (ETA) prediction given origin-destination pairs. The ETA prediction is based on physics and mathematical laws given by the extracted information of probability density-based trajectories constructed. The overall ETA prediction errors are about 0.106 days (i.e. 2.544 hours) on average with 0.549 days (i.e. 13.176 hours) standard deviation, and the proposed approach has an accuracy of 92.08% with 0.959 R-Squared value for overall trajectories between Singapore and Australia ports selected. △ Less

Submitted 13 May, 2022; originally announced May 2022.

Predicting and Optimizing for Energy Efficient ACMV Systems: Computational Intelligence Approaches

Authors: Deqing Zhai, Yeng Chai Soh

Abstract: In this study, a novel application of neural networks that predict thermal comfort states of occupants is proposed with accuracy over 95%, and two optimization algorithms are proposed and evaluated under two real cases (general offices and lecture theatres/conference rooms scenarios) in Singapore. The two optimization algorithms are Bayesian Gaussian process optimization (BGPO) and augmented firef… ▽ More In this study, a novel application of neural networks that predict thermal comfort states of occupants is proposed with accuracy over 95%, and two optimization algorithms are proposed and evaluated under two real cases (general offices and lecture theatres/conference rooms scenarios) in Singapore. The two optimization algorithms are Bayesian Gaussian process optimization (BGPO) and augmented firefly algorithm (AFA). Based on our earlier studies, the models of energy consumption were developed and well-trained through neural networks. This study focuses on using novel active approaches to evaluate thermal comfort of occupants and so as to solves a multiple-objective problem that aims to balance energy-efficiency of centralized air-conditioning systems and thermal comfort of occupants. The study results show that both BGPO and AFA are feasible to resolve this no prior knowledge-based optimization problem effectively. However, the optimal solutions of AFA are more consistent than those of BGPO at given sample sizes. The best energy saving rates (ESR) of BGPO and AFA are around -21% and -10% respectively at energy-efficient user preference for both Case 1 and Case 2. As a result, an potential benefit of S$1219.1 can be achieved annually for this experimental laboratory level in Singapore. △ Less

Submitted 19 April, 2022; originally announced May 2022.

Predicting Berth Stay for Tanker Terminals: A Systematic and Dynamic Approach

Authors: Deqing Zhai, Xiuju Fu, Xiao Feng Yin, Haiyan Xu, Wanbing Zhang

Abstract: Given the trend of digitization and increasing number of maritime transport, prediction of vessel berth stay has been triggered for requirements of operation research and scheduling optimization problem in the era of maritime big data, which takes a significant part in port efficiency and maritime logistics enhancement. This study proposes a systematic and dynamic approach of predicting berth stay… ▽ More Given the trend of digitization and increasing number of maritime transport, prediction of vessel berth stay has been triggered for requirements of operation research and scheduling optimization problem in the era of maritime big data, which takes a significant part in port efficiency and maritime logistics enhancement. This study proposes a systematic and dynamic approach of predicting berth stay for tanker terminals. The approach covers three innovative aspects: 1) Data source employed is multi-faceted, including cargo operation data from tanker terminals, time-series data from automatic identification system (AIS), etc. 2) The process of berth stay is decomposed into multiple blocks according to data analysis and information extraction innovatively, and practical operation scenarios are also developed accordingly. 3) The predictive models of berth stay are developed on the basis of prior data analysis and information extraction under two methods, including regression and decomposed distribution. The models are evaluated under four dynamic scenarios with certain designated cargoes among two different terminals. The evaluation results show that the proposed approach can predict berth stay with the accuracy up to 98.81% validated by historical baselines, and also demonstrate the proposed approach has dynamic capability of predicting berth stay among the scenarios. The model may be potentially applied for short-term pilot-booking or scheduling optimizations within a reasonable time frame for advancement of port intelligence and logistics efficiency. △ Less

Submitted 18 May, 2022; v1 submitted 8 April, 2022; originally announced April 2022.

Optimizing Coordinative Schedules for Tanker Terminals: An Intelligent Large Spatial-Temporal Data-Driven Approach -- Part 2

Authors: Deqing Zhai, Xiuju Fu, Xiao Feng Yin, Haiyan Xu, Wanbing Zhang, Ning Li

Abstract: In this study, a novel coordinative scheduling optimization approach is proposed to enhance port efficiency by reducing weighted average turnaround time. The proposed approach is developed as a heuristic algorithm applied and investigated through different observation windows with weekly rolling horizon paradigm method. The experimental results show that the proposed approach is effective and prom… ▽ More In this study, a novel coordinative scheduling optimization approach is proposed to enhance port efficiency by reducing weighted average turnaround time. The proposed approach is developed as a heuristic algorithm applied and investigated through different observation windows with weekly rolling horizon paradigm method. The experimental results show that the proposed approach is effective and promising on mitigating the turnaround time of vessels. The results demonstrate that largest potential savings of turnaround time (weighted average) are around 17 hours (28%) reduction on baseline of 1-week observation, 45 hours (37%) reduction on baseline of 2-week observation and 70 hours (40%) reduction on baseline of 3-week observation. Even though the experimental results are based on historical datasets, the results potentially present significant benefits if real-time applications were applied under a quadratic computational complexity. △ Less

Submitted 8 April, 2022; originally announced April 2022.

Optimizing Coordinative Schedules for Tanker Terminals: An Intelligent Large Spatial-Temporal Data-Driven Approach -- Part 1

Authors: Deqing Zhai, Xiuju Fu, Xiao Feng Yin, Haiyan Xu, Wanbing Zhang, Ning Li

Abstract: In this study, a novel coordinative scheduling optimization approach is proposed to enhance port efficiency by reducing average wait time and turnaround time. The proposed approach consists of enhanced particle swarm optimization (ePSO) as kernel and augmented firefly algorithm (AFA) as global optimal search. Two paradigm methods of the proposed approach are investigated, which are batch method an… ▽ More In this study, a novel coordinative scheduling optimization approach is proposed to enhance port efficiency by reducing average wait time and turnaround time. The proposed approach consists of enhanced particle swarm optimization (ePSO) as kernel and augmented firefly algorithm (AFA) as global optimal search. Two paradigm methods of the proposed approach are investigated, which are batch method and rolling horizon method. The experimental results show that both paradigm methods of proposed approach can effectively enhance port efficiency. The average wait time could be significantly reduced by 86.0% - 95.5%, and the average turnaround time could eventually save 38.2% - 42.4% with respect to historical benchmarks. Moreover, the paradigm method of rolling horizon could reduce to 20 mins on running time over 3-month datasets, rather than 4 hrs on batch method at corresponding maximum performance. △ Less

Submitted 8 April, 2022; originally announced April 2022.

Anomalous magneto-optical response and chiral interface of dipolar excitons at twisted valleys

Authors: Huiyuan Zheng, Dawei Zhai, Wang Yao

Abstract: An anomalous magneto-optical spectrum is discovered for dipolar valley excitons in twisted double layer transition metal dichalcogenides (TMD), where in-plane magnetic field induces a sizable multiplet splitting of exciton states inside the light cone. Chiral dispersions of the split branches make possible efficient optical injection of unidirectional exciton current. We also find an analog effect… ▽ More An anomalous magneto-optical spectrum is discovered for dipolar valley excitons in twisted double layer transition metal dichalcogenides (TMD), where in-plane magnetic field induces a sizable multiplet splitting of exciton states inside the light cone. Chiral dispersions of the split branches make possible efficient optical injection of unidirectional exciton current. We also find an analog effect with a modest heterostrain replacing the magnetic field for introducing large splitting and chiral dispersions in the light cone. Angular orientation of photo-injected exciton flow can be controlled by strain, with left-right unidirectionality selected by circular polarisation. △ Less

Submitted 21 July, 2022; v1 submitted 9 March, 2022; originally announced March 2022.

Comments: 18 pages, 5 figures

Journal ref: Nano Lett. (2022) 22, 13, 5466-5472

Ultrafast control of moiré pseudo-electromagnetic field in homobilayer semiconductors

Authors: Dawei Zhai, Wang Yao

Abstract: In long-wavelength moiré patterns of homobilayer semiconductors, the layer pseudospin of electrons is subject to a sizable Zeeman field that is spatially modulated from the interlayer coupling in moiré. By interference of this spatial modulation with a homogeneous but dynamically tunable component from out-of-plane electric field, we show that the spatial-temporal profile of the overall Zeeman fie… ▽ More In long-wavelength moiré patterns of homobilayer semiconductors, the layer pseudospin of electrons is subject to a sizable Zeeman field that is spatially modulated from the interlayer coupling in moiré. By interference of this spatial modulation with a homogeneous but dynamically tunable component from out-of-plane electric field, we show that the spatial-temporal profile of the overall Zeeman field therefore features a topological texture that can be controlled in an ultrafast timescale by a terahertz field or an interlayer bias. Such dynamical modulation leads to the emergence of an in-plane electric field for low energy carriers, which is related to their real space Berry curvature -- the moiré magnetic field -- through the Faraday's law of induction. These emergent electromagnetic fields, having opposite signs at the time reversal pair of valleys, can be exploited to manipulate valley and spin in the moiré landscape under the control by a bias pulse or a terahertz irradiation. △ Less

Submitted 9 March, 2022; originally announced March 2022.

Comments: To appear in Natural Sciences

Journal ref: Natural Sciences 2, e20210101 (2022)

Shadows can be Dangerous: Stealthy and Effective Physical-world Adversarial Attack by Natural Phenomenon

Authors: Yiqi Zhong, Xianming Liu, Deming Zhai, Junjun Jiang, Xiangyang Ji

Abstract: Estimating the risk level of adversarial examples is essential for safely deploying machine learning models in the real world. One popular approach for physical-world attacks is to adopt the "sticker-pasting" strategy, which however suffers from some limitations, including difficulties in access to the target or printing by valid colors. A new type of non-invasive attacks emerged recently, which a… ▽ More Estimating the risk level of adversarial examples is essential for safely deploying machine learning models in the real world. One popular approach for physical-world attacks is to adopt the "sticker-pasting" strategy, which however suffers from some limitations, including difficulties in access to the target or printing by valid colors. A new type of non-invasive attacks emerged recently, which attempt to cast perturbation onto the target by optics based tools, such as laser beam and projector. However, the added optical patterns are artificial but not natural. Thus, they are still conspicuous and attention-grabbed, and can be easily noticed by humans. In this paper, we study a new type of optical adversarial examples, in which the perturbations are generated by a very common natural phenomenon, shadow, to achieve naturalistic and stealthy physical-world adversarial attack under the black-box setting. We extensively evaluate the effectiveness of this new attack on both simulated and real-world environments. Experimental results on traffic sign recognition demonstrate that our algorithm can generate adversarial examples effectively, reaching 98.23% and 90.47% success rates on LISA and GTSRB test sets respectively, while continuously misleading a moving camera over 95% of the time in real-world scenarios. We also offer discussions about the limitations and the defense mechanism of this attack. △ Less

Submitted 22 March, 2022; v1 submitted 7 March, 2022; originally announced March 2022.

Comments: This paper has been accepted by CVPR2022. Code: https://github.com/hncszyq/ShadowAttack

RFLBAT: A Robust Federated Learning Algorithm against Backdoor Attack

Authors: Yongkang Wang, Dihua Zhai, Yufeng Zhan, Yuanqing Xia

Abstract: Federated learning (FL) is a distributed machine learning paradigm where enormous scattered clients (e.g. mobile devices or IoT devices) collaboratively train a model under the orchestration of a central server (e.g. service provider), while kee** the training data decentralized. Unfortunately, FL is susceptible to a variety of attacks, including backdoor attack, which is made substantially wors… ▽ More Federated learning (FL) is a distributed machine learning paradigm where enormous scattered clients (e.g. mobile devices or IoT devices) collaboratively train a model under the orchestration of a central server (e.g. service provider), while kee** the training data decentralized. Unfortunately, FL is susceptible to a variety of attacks, including backdoor attack, which is made substantially worse in the presence of malicious attackers. Most of algorithms usually assume that the malicious at tackers no more than benign clients or the data distribution is independent identically distribution (IID). However, no one knows the number of malicious attackers and the data distribution is usually non identically distribution (Non-IID). In this paper, we propose RFLBAT which utilizes principal component analysis (PCA) technique and Kmeans clustering algorithm to defend against backdoor attack. Our algorithm RFLBAT does not bound the number of backdoored attackers and the data distribution, and requires no auxiliary information outside of the learning process. We conduct extensive experiments including a variety of backdoor attack types. Experimental results demonstrate that RFLBAT outperforms the existing state-of-the-art algorithms and is able to resist various backdoor attack scenarios including different number of attackers (DNA), different Non-IID scenarios (DNS), different number of clients (DNC) and distributed backdoor attack (DBA). △ Less

Submitted 10 January, 2022; originally announced January 2022.

DMS-GCN: Dynamic Mutiscale Spatiotemporal Graph Convolutional Networks for Human Motion Prediction

Authors: Zigeng Yan, Di-Hua Zhai, Yuanqing Xia

Abstract: Human motion prediction is an important and challenging task in many computer vision application domains. Recent work concentrates on utilizing the timing processing ability of recurrent neural networks (RNNs) to achieve smooth and reliable results in short-term prediction. However, as evidenced by previous work, RNNs suffer from errors accumulation, leading to unreliable results. In this paper, w… ▽ More Human motion prediction is an important and challenging task in many computer vision application domains. Recent work concentrates on utilizing the timing processing ability of recurrent neural networks (RNNs) to achieve smooth and reliable results in short-term prediction. However, as evidenced by previous work, RNNs suffer from errors accumulation, leading to unreliable results. In this paper, we propose a simple feed-forward deep neural network for motion prediction, which takes into account temporal smoothness and spatial dependencies between human body joints. We design a Multi-scale Spatio-temporal graph convolutional networks (GCNs) to implicitly establish the Spatio-temporal dependence in the process of human movement, where different scales fused dynamically during training. The entire model is suitable for all actions and follows a framework of encoder-decoder. The encoder consists of temporal GCNs to capture motion features between frames and semi-autonomous learned spatial GCNs to extract spatial structure among joint trajectories. The decoder uses temporal convolution networks (TCNs) to maintain its extensive ability. Extensive experiments show that our approach outperforms SOTA methods on the datasets of Human3.6M and CMU Mocap while only requiring much lesser parameters. Code will be available at https://github.com/yzg9353/DMSGCN. △ Less

Submitted 20 December, 2021; originally announced December 2021.

Twist versus heterostrain control of optical properties of moiré exciton minibands

Authors: Huiyuan Zheng, Dawei Zhai, Wang Yao

Abstract: We investigate the optical properties of interlayer excitons in heterobilayer transition metal dichalcogenides where moiré pattern is introduced by heterostrain, in comparison with that introduced by twisting (and/or lattice mismatch). Besides being a cause of the moiré texture, strain also effectively introduces a constant gauge potential on either electron or hole, which shifts the dispersion of… ▽ More We investigate the optical properties of interlayer excitons in heterobilayer transition metal dichalcogenides where moiré pattern is introduced by heterostrain, in comparison with that introduced by twisting (and/or lattice mismatch). Besides being a cause of the moiré texture, strain also effectively introduces a constant gauge potential on either electron or hole, which shifts the dispersion of kinetic energy with respect to the excitonic crystal momenta in the moiré superlattices.This leads to distinct exciton mini-band dispersions and light coupling properties from the twisting induced moiré, even if the excitonic moiré superlattice potentials have the similar real-space profile for the two cases. For strain that breaks the three-fold rotational symmetry at the atomic scale,the exciton wave packets trapped at the superlattice potential minima have elliptically polarized valley optical selection rules, in contrast to the circularly polarized ones in the twisting moiré. We investigate the evolution of the excitonic mini-bands and the optical dipoles of the bright states inside the light cones with the decrease of the moiré periodicity, upon which the excitonic wavefunctions evolve from localized wave packets to the extended Bloch states. Furthermore, moiré exciton properties under the interplay of twisting and heterostrain are also discussed. △ Less

Submitted 4 October, 2021; originally announced October 2021.

Comments: 17 pages, 13 figures

Learning with Noisy Labels via Sparse Regularization

Authors: Xiong Zhou, Xianming Liu, Chenyang Wang, Deming Zhai, Junjun Jiang, Xiangyang Ji

Abstract: Learning with noisy labels is an important and challenging task for training accurate deep neural networks. Some commonly-used loss functions, such as Cross Entropy (CE), suffer from severe overfitting to noisy labels. Robust loss functions that satisfy the symmetric condition were tailored to remedy this problem, which however encounter the underfitting effect. In this paper, we theoretically pro… ▽ More Learning with noisy labels is an important and challenging task for training accurate deep neural networks. Some commonly-used loss functions, such as Cross Entropy (CE), suffer from severe overfitting to noisy labels. Robust loss functions that satisfy the symmetric condition were tailored to remedy this problem, which however encounter the underfitting effect. In this paper, we theoretically prove that \textbf{any loss can be made robust to noisy labels} by restricting the network output to the set of permutations over a fixed vector. When the fixed vector is one-hot, we only need to constrain the output to be one-hot, which however produces zero gradients almost everywhere and thus makes gradient-based optimization difficult. In this work, we introduce the sparse regularization strategy to approximate the one-hot constraint, which is composed of network output sharpening operation that enforces the output distribution of a network to be sharp and the $\ell_p$-norm ($p\le 1$) regularization that promotes the network output to be sparse. This simple approach guarantees the robustness of arbitrary loss functions while not hindering the fitting ability. Experimental results demonstrate that our method can significantly improve the performance of commonly-used loss functions in the presence of noisy labels and class imbalance, and outperform the state-of-the-art methods. The code is available at https://github.com/hitcszx/lnl_sr. △ Less

Submitted 31 July, 2021; originally announced August 2021.

Comments: ICCV 2021

Layer pseudospin dynamics and genuine non-Abelian Berry phase in inhomogeneously strained moiré pattern

Authors: Dawei Zhai, Wang Yao

Abstract: Periodicity of long wavelength moiré patterns is very often destroyed by the inhomogeneous strain introduced in fabrications of van der Waals layered structures. We present a framework to describe massive Dirac fermions in such distorted moiré pattern of transition metal dichalcogenides homobilayers, accounting for the dynamics of layer pseudospin. In decoupled bilayers, we show two causes of in-p… ▽ More Periodicity of long wavelength moiré patterns is very often destroyed by the inhomogeneous strain introduced in fabrications of van der Waals layered structures. We present a framework to describe massive Dirac fermions in such distorted moiré pattern of transition metal dichalcogenides homobilayers, accounting for the dynamics of layer pseudospin. In decoupled bilayers, we show two causes of in-plane layer pseudospin precession: By the coupling of layer antisymmetric strain to valley magnetic moment; and by the Aharonov-Bohm effect in the SU(2) gauge potential for the case of R-type bilayer under antisymmetric strain and H-type under symmetric strain. With interlayer coupling in the moiré, its interplay with strain manifests as a non-Abelian gauge field. We show a genuine non-Abelian Aharonov-Bohm effect in such field, where the evolution operators for different loops are non-commutative. This provides an exciting platform to explore non-Abelian gauge field effects on electron, with remarkable tunability of the field by strain and interlayer bias. △ Less

Submitted 13 December, 2020; originally announced December 2020.

Comments: To appear in Phys. Rev. Lett

Journal ref: Phys. Rev. Lett. 125, 266404 (2020)

Performance Analysis for the CMSA/CA Protocol in UAV-based IoT network

Authors: Xianzhen Guo, Bin Li, Daosen Zhai, Ruonan Zhang

Abstract: UAV-based base station (UBS) has played an important role in the air-ground integration network due to its high flexibility and nice air-ground wireless channels. Especially in Internet of Things (IoT) services, UBS can provide an efficient way for data collection from the IoT devices. However, due to the continuous mobility of UBS, the communication durations of devices in different locations wit… ▽ More UAV-based base station (UBS) has played an important role in the air-ground integration network due to its high flexibility and nice air-ground wireless channels. Especially in Internet of Things (IoT) services, UBS can provide an efficient way for data collection from the IoT devices. However, due to the continuous mobility of UBS, the communication durations of devices in different locations with the UBS are not only time-limited, but also vary from each other. Therefore, it is a challenging task to analyze the throughput performance of the UAV-based IoT network. Accordingly, in this paper, we consider an air-ground network in which UAV flies straightly to collect information from the IoT devices based on CSMA/CA protocol. An analytical model analyzing the performance of this protocol in the network is proposed. In detail, we set up the system model for the network, and propose a new concept called quitting probability. Then, a modified Markov chain model integrating the quitting probability is introduced to describe the transmission state transition process and an accurately theoretical analysis of saturation throughput is given. In addition, the effects of the network parameters are discussed in the simulation section. △ Less

Submitted 10 December, 2020; v1 submitted 14 November, 2020; originally announced November 2020.

Comments: arXiv admin note: substantial text overlap with arXiv:2010.11623

Fully Unsupervised Person Re-identification viaSelective Contrastive Learning

Authors: Bo Pang, Deming Zhai, Junjun Jiang, Xianming Liu

Abstract: Person re-identification (ReID) aims at searching the same identity person among images captured by various cameras. Unsupervised person ReID attracts a lot of attention recently, due to it works without intensive manual annotation and thus shows great potential of adapting to new conditions. Representation learning plays a critical role in unsupervised person ReID. In this work, we propose a nove… ▽ More Person re-identification (ReID) aims at searching the same identity person among images captured by various cameras. Unsupervised person ReID attracts a lot of attention recently, due to it works without intensive manual annotation and thus shows great potential of adapting to new conditions. Representation learning plays a critical role in unsupervised person ReID. In this work, we propose a novel selective contrastive learning framework for unsupervised feature learning. Specifically, different from traditional contrastive learning strategies, we propose to use multiple positives and adaptively sampled negatives for defining the contrastive loss, enabling to learn a feature embedding model with stronger identity discriminative representation. Moreover, we propose to jointly leverage global and local features to construct three dynamic dictionaries, among which the global and local memory banks are used for pairwise similarity computation and the mixture memory bank are used for contrastive loss definition. Experimental results demonstrate the superiority of our method in unsupervised person ReID compared with the state-of-the-arts. △ Less

Submitted 3 March, 2021; v1 submitted 15 October, 2020; originally announced October 2020.

The TianQin project: current progress on science and technology

Authors: Jianwei Mei, Yan-Zheng Bai, Jiahui Bao, Enrico Barausse, Lin Cai, Enrico Canuto, Bin Cao, Wei-Ming Chen, Yu Chen, Yan-Wei Ding, Hui-Zong Duan, Huimin Fan, Wen-Fan Feng, Honglin Fu, Qing Gao, TianQuan Gao, Yungui Gong, Xingyu Gou, Chao-Zheng Gu, De-Feng Gu, Zi-Qi He, Martin Hendry, Wei Hong, Xin-Chun Hu, Yi-Ming Hu , et al. (82 additional authors not shown)

Abstract: TianQin is a planned space-based gravitational wave (GW) observatory consisting of three earth orbiting satellites with an orbital radius of about $10^5~{\rm km}$. The satellites will form a equilateral triangle constellation the plane of which is nearly perpendicular to the ecliptic plane. TianQin aims to detect GWs between $10^{-4}~{\rm Hz}$ and $1~{\rm Hz}$ that can be generated by a wide varie… ▽ More TianQin is a planned space-based gravitational wave (GW) observatory consisting of three earth orbiting satellites with an orbital radius of about $10^5~{\rm km}$. The satellites will form a equilateral triangle constellation the plane of which is nearly perpendicular to the ecliptic plane. TianQin aims to detect GWs between $10^{-4}~{\rm Hz}$ and $1~{\rm Hz}$ that can be generated by a wide variety of important astrophysical and cosmological sources, including the inspiral of Galactic ultra-compact binaries, the inspiral of stellar-mass black hole binaries, extreme mass ratio inspirals, the merger of massive black hole binaries, and possibly the energetic processes in the very early universe or exotic sources such as cosmic strings. In order to start science operations around 2035, a roadmap called the 0123 plan is being used to bring the key technologies of TianQin to maturity, supported by the construction of a series of research facilities on the ground. Two major projects of the 0123 plan are being carried out. In this process, the team has created a new generation $17~{\rm cm}$ single-body hollow corner-cube retro-reflector which has been launched with the QueQiao satellite on 21 May 2018; a new laser ranging station equipped with a $1.2~{\rm m}$ telescope has been constructed and the station has successfully ranged to all the five retro-reflectors on the Moon; and the TianQin-1 experimental satellite has been launched on 20 December 2019 and the first round result shows that the satellite has exceeded all of its mission requirements. △ Less

Submitted 24 August, 2020; originally announced August 2020.

Comments: 16 pages, 1 figure, invited article to a special section in PTEP

Single Image Deraining via Scale-space Invariant Attention Neural Network

Authors: Bo Pang, Deming Zhai, Junjun Jiang, Xianming Liu

Abstract: Image enhancement from degradation of rainy artifacts plays a critical role in outdoor visual computing systems. In this paper, we tackle the notion of scale that deals with visual changes in appearance of rain steaks with respect to the camera. Specifically, we revisit multi-scale representation by scale-space theory, and propose to represent the multi-scale correlation in convolutional feature d… ▽ More Image enhancement from degradation of rainy artifacts plays a critical role in outdoor visual computing systems. In this paper, we tackle the notion of scale that deals with visual changes in appearance of rain steaks with respect to the camera. Specifically, we revisit multi-scale representation by scale-space theory, and propose to represent the multi-scale correlation in convolutional feature domain, which is more compact and robust than that in pixel domain. Moreover, to improve the modeling ability of the network, we do not treat the extracted multi-scale features equally, but design a novel scale-space invariant attention mechanism to help the network focus on parts of the features. In this way, we summarize the most activated presence of feature maps as the salient features. Extensive experiments results on synthetic and real rainy scenes demonstrate the superior performance of our scheme over the state-of-the-arts. △ Less

Submitted 10 June, 2020; v1 submitted 9 June, 2020; originally announced June 2020.

Theory of tunable flux lattices in the homobilayer moiré of twisted and uniformly strained transition metal dichalcogenides

Authors: Dawei Zhai, Wang Yao

Abstract: The spatial texture of internal degree of freedom of electrons has profound effects on the properties of materials. Such texture in real space can manifest as an emergent magnetic field (or Berry curvature), which is expected to induce interesting valley/spin-related transport phenomena. Moiré pattern, which emerges as a spatial variation at the interface of 2D atomic crystals, provides a natural… ▽ More The spatial texture of internal degree of freedom of electrons has profound effects on the properties of materials. Such texture in real space can manifest as an emergent magnetic field (or Berry curvature), which is expected to induce interesting valley/spin-related transport phenomena. Moiré pattern, which emerges as a spatial variation at the interface of 2D atomic crystals, provides a natural platform for investigating such real space Berry curvature effects. Here we study moiré structures formed in homobilayer transition metal dichalcogenides (TMDs) due to twisting, various uniform strain profiles, and their combinations, where electrons can reside in either layer with the layer index serving as an internal degree of freedom. The layer pseudo-spin exhibits vortex/antivortex textures in the moiré supercell, leading to a giant geometric magnetic field and a scalar potential. Within a geometric picture, the moiré magnetic field is found as the cross product of the gradients of the out-of-plane pseudo-spin and the in-plane pseudo-spin orientation respectively. We discover dual roles of uniform strain: Besides being a cause of the moire atomic texture in the homobilayer, it also contributes a pseudo-gauge potential that modifies the local phase of interlayer coupling. Consequently, strain can be employed to tune the in-plane pseudo-spin texture, while interlayer bias tunes the out-of-plane pseudo-spin, and we show how the moiré magnetic field's spatial profile, intensity, and flux per supercell can be engineered. Through the geometric scalar correction, the landscape of the scalar potential can also be engineered along with the moiré magnetic field, forming distinct effective lattice structures. These properties render TMD moiré structures promising to build tunable flux lattices for transport and topological material applications. △ Less

Submitted 3 June, 2020; originally announced June 2020.

Comments: Accepted by Phys. Rev. Materials

Journal ref: Phys. Rev. Materials 4, 094002 (2020)

Rectified Meta-Learning from Noisy Labels for Robust Image-based Plant Disease Diagnosis

Authors: Ruifeng Shi, Deming Zhai, Xianming Liu, Junjun Jiang, Wen Gao

Abstract: Plant diseases serve as one of main threats to food security and crop production. It is thus valuable to exploit recent advances of artificial intelligence to assist plant disease diagnosis. One popular approach is to transform this problem as a leaf image classification task, which can be then addressed by the powerful convolutional neural networks (CNNs). However, the performance of CNN-based cl… ▽ More Plant diseases serve as one of main threats to food security and crop production. It is thus valuable to exploit recent advances of artificial intelligence to assist plant disease diagnosis. One popular approach is to transform this problem as a leaf image classification task, which can be then addressed by the powerful convolutional neural networks (CNNs). However, the performance of CNN-based classification approach depends on a large amount of high-quality manually labeled training data, which are inevitably introduced noise on labels in practice, leading to model overfitting and performance degradation. To overcome this problem, we propose a novel framework that incorporates rectified meta-learning module into common CNN paradigm to train a noise-robust deep network without using extra supervision information. The proposed method enjoys the following merits: i) A rectified meta-learning is designed to pay more attention to unbiased samples, leading to accelerated convergence and improved classification accuracy. ii) Our method is free on assumption of label noise distribution, which works well on various kinds of noise. iii) Our method serves as a plug-and-play module, which can be embedded into any deep models optimized by gradient descent based method. Extensive experiments are conducted to demonstrate the superior performance of our algorithm over the state-of-the-arts. △ Less

Submitted 17 March, 2020; v1 submitted 17 March, 2020; originally announced March 2020.

A novel semi-supervised multi-view clustering framework for screening Parkinson's disease

Authors: Xiaobo Zhang, Donghai Zhai, Yan Yang, Yiling Zhang, Chunlin Wang

Abstract: In recent years, there are many research cases for the diagnosis of Parkinson's disease (PD) with the brain magnetic resonance imaging (MRI) by utilizing the traditional unsupervised machine learning methods and the supervised deep learning models. However, unsupervised learning methods are not good at extracting accurate features among MRIs and it is difficult to collect enough data in the field… ▽ More In recent years, there are many research cases for the diagnosis of Parkinson's disease (PD) with the brain magnetic resonance imaging (MRI) by utilizing the traditional unsupervised machine learning methods and the supervised deep learning models. However, unsupervised learning methods are not good at extracting accurate features among MRIs and it is difficult to collect enough data in the field of PD to satisfy the need of training deep learning models. Moreover, most of the existing studies are based on single-view MRI data, of which data characteristics are not sufficient enough. In this paper, therefore, in order to tackle the drawbacks mentioned above, we propose a novel semi-supervised learning framework called Semi-supervised Multi-view learning Clustering architecture technology (SMC). The model firstly introduces the sliding window method to grasp different features, and then uses the dimensionality reduction algorithms of Linear Discriminant Analysis (LDA) to process the data with different features. Finally, the traditional single-view clustering and multi-view clustering methods are employed on multiple feature views to obtain the results. Experiments show that our proposed method is superior to the state-of-art unsupervised learning models on the clustering effect. As a result, it may be noted that, our work could contribute to improving the effectiveness of identifying PD by previous labeled and subsequent unlabeled medical MRI data in the realistic medical environment. △ Less

Submitted 9 March, 2020; originally announced March 2020.

Comments: 17 pages, 8 figures, article

ADRN: Attention-based Deep Residual Network for Hyperspectral Image Denoising

Authors: Yongsen Zhao, Deming Zhai, Junjun Jiang, Xianming Liu

Abstract: Hyperspectral image (HSI) denoising is of crucial importance for many subsequent applications, such as HSI classification and interpretation. In this paper, we propose an attention-based deep residual network to directly learn a map** from noisy HSI to the clean one. To jointly utilize the spatial-spectral information, the current band and its $K$ adjacent bands are simultaneously exploited as t… ▽ More Hyperspectral image (HSI) denoising is of crucial importance for many subsequent applications, such as HSI classification and interpretation. In this paper, we propose an attention-based deep residual network to directly learn a map** from noisy HSI to the clean one. To jointly utilize the spatial-spectral information, the current band and its $K$ adjacent bands are simultaneously exploited as the input. Then, we adopt convolution layer with different filter sizes to fuse the multi-scale feature, and use shortcut connection to incorporate the multi-level information for better noise removal. In addition, the channel attention mechanism is employed to make the network concentrate on the most relevant auxiliary information and features that are beneficial to the denoising process best. To ease the training procedure, we reconstruct the output through a residual mode rather than a straightforward prediction. Experimental results demonstrate that our proposed ADRN scheme outperforms the state-of-the-art methods both in quantitative and visual evaluations. △ Less

Submitted 4 March, 2020; originally announced March 2020.

Structural Controllability of Networked Relative Coupling Systems

Authors: Yuan Zhang, Yuanqing Xia, Dihua Zhai

Abstract: This paper studies the controllability of networked relative coupling systems (NRCSs), in which subsystems are of fixed high-order linear dynamics and coupled through relative variables depending on their neighbors, from a structural perspective. The purpose is to explore conditions for subsystem dynamics and network topologies under which for almost all weights of the subsystem interaction links,… ▽ More This paper studies the controllability of networked relative coupling systems (NRCSs), in which subsystems are of fixed high-order linear dynamics and coupled through relative variables depending on their neighbors, from a structural perspective. The purpose is to explore conditions for subsystem dynamics and network topologies under which for almost all weights of the subsystem interaction links, the corresponding numerical NRCSs are controllable, which is called structurally controllable. Three types of subsystem interaction fashions are considered: 1) each subsystem is single-input-single-output (SISO), 2) each subsystem is multiple-input-multiple-output (MIMO), and the weights for all channels between two subsystems are identical, and 3) each subsystem is MIMO, but different channels between two subsystems can be weighted differently. {We show that all parameter-dependent modes of the NRCSs are generically controllable under some necessary connectivity conditions. We then derive necessary and/or sufficient conditions for structural controllability depending on subsystem dynamics and network topologies' connectivity in a decoupled form for all the three interaction fashions.} We also extend our results to handle certain subsystem heterogeneities and demonstrate their direct applications on some practical systems, including the mass-spring-damper system and the power network. △ Less

Submitted 26 January, 2021; v1 submitted 14 November, 2019; originally announced November 2019.

Comments: to appear in Automatica as a brief paper

Journal ref: Automatica, 2021

Electron dynamics in strained graphene

Authors: Dawei Zhai, Nancy Sandler

Abstract: The paper presents a theoretical description of the effects of strain induced by out-of-plane deformations on charge distributions and transport on graphene. A review of a continuum model for electrons using the Dirac formalism is complemented with elasticity theory to represent strain fields. The resulting model is cast in terms of scalar and pseudo-magnetic fields that control electron dynamics.… ▽ More The paper presents a theoretical description of the effects of strain induced by out-of-plane deformations on charge distributions and transport on graphene. A review of a continuum model for electrons using the Dirac formalism is complemented with elasticity theory to represent strain fields. The resulting model is cast in terms of scalar and pseudo-magnetic fields that control electron dynamics. Two distinct geometries, a bubble, and a fold are chosen to represent the most commonly observed deformations in experimental settings. It is shown that local charge accumulation regions appear in deformed areas, with a peculiar charge distribution that favors the occupation of one sublattice only. This unique phenomenon that allows distinguishing each carbon atom in the unit cell, is the manifestation of a sublattice symmetry broken phase. For specific parameters, resonant states appear in localized charged regions, as shown by the emergence of discrete levels in band structure calculations. These findings are presented in terms of intuitive pictures that exploit analogies with confinement produced by square barriers. In addition, electron currents through strained regions are spatially separated into their valley components, making possible the manipulation of electrons with different valley indices. The degree of valley filtering (or polarization) for a specific system can be controlled by properly designing the strained area. The comparison between efficiencies of filters built with this type of geometries identifies extended deformations as better valley filters. A proposal for their experimental implementations as a component of devices and a discussion for potential observation of novel physics in strained structures are presented at the end of the article. △ Less

Submitted 12 October, 2019; originally announced October 2019.

Comments: Brief Review Article (35 pages)

Journal ref: Modern Physics Letters B. 1930001 (2019)

Linking interlayer twist angle to geometrical parameters of self-assembled folded graphene structures

Authors: Johannes C. Rode, Dawei Zhai, Christopher Belke, Sung J. Hong, Hennrik Schmidt, Nancy Sandler, Rolf J. Haug

Abstract: Thin adhesive films can be removed from substrates, torn, and folded in distinct geometries under external driving forces. In two-dimensional materials, however, these processes can be self-driven as shown in previous studies on folded twisted bilayer graphene nanoribbons produced by spontaneous tearing and peeling from a substrate. Here, we use atomic force microscopy techniques to generate and c… ▽ More Thin adhesive films can be removed from substrates, torn, and folded in distinct geometries under external driving forces. In two-dimensional materials, however, these processes can be self-driven as shown in previous studies on folded twisted bilayer graphene nanoribbons produced by spontaneous tearing and peeling from a substrate. Here, we use atomic force microscopy techniques to generate and characterize the geometrical structure of naturally self-grown folded nanoribbon structures. Measurements of nanoribbon width and interlayer separation reveal similar twist-angle dependences possibly caused by the anisotropy in the bilayer potential. In addition, analysis of the data shows an unexpected correlation between the height of the folded arc edge -parameterized by a radius R-, and the ribbon width, suggestive of a self-growth process driven by a variable cross-sectional shape. These observations are well described by an energy minimization model that includes the bilayer adhesion energy density as represented by a distance dependent Morse potential. We obtain an analytical expression for the radius R versus the ribbon width that predicts a renormalized bending rigidity and stands in good agreement with experimental observations. The newly found relation between these geometrical parameters suggests a mechanism for tailored growth of folded twisted bilayer graphene -- a platform for many intriguing physics phenomena. △ Less

Submitted 11 April, 2019; originally announced April 2019.

Journal ref: 2D Mater. 6, 015021 (2019)

Sublattice symmetry breaking and Kondo-effect enhancement in strained graphene

Authors: Dawei Zhai, Kevin Ingersent, Sergio E. Ulloa, Nancy Sandler

Abstract: Kondo physics in doped monolayer graphene is predicted to exhibit unusual features due to the linear vanishing of the pristine material's density of states at the Dirac point. Despite several attempts, conclusive experimental observation of the phenomenon remains elusive. One likely obstacle to identification is a very small Kondo temperature scale $T_K$ in situations where the chemical potential… ▽ More Kondo physics in doped monolayer graphene is predicted to exhibit unusual features due to the linear vanishing of the pristine material's density of states at the Dirac point. Despite several attempts, conclusive experimental observation of the phenomenon remains elusive. One likely obstacle to identification is a very small Kondo temperature scale $T_K$ in situations where the chemical potential lies near the Dirac point. We propose tailored mechanical deformations of monolayer graphene as a means of revealing unique fingerprints of the Kondo effect. Inhomogeneous strains are known to produce specific alternating changes in the local density of states (LDOS) away from the Dirac point that signal sublattice symmetry breaking effects. Small LDOS changes can be amplified in an exponential increase or decrease of $T_K$ for magnetic impurities attached at different locations. We illustrate this behavior in two deformation geometries: a circular 'bubble' and a long fold, both described by Gaussian displacement profiles. We calculate the LDOS changes for modest strains and analyze the relevant Anderson impurity model describing a magnetic atom adsorbed in either a 'top-site' or a 'hollow-site' configuration. Numerical renormalization-group solutions of the impurity model suggest that higher expected $T_K$ values, combined with distinctive spatial patterns under variation of the point of graphene attachment, make the top-site configuration the more promising for experimental observation of signatures of the Kondo effect. The strong strain sensitivity of $T_K$ may lift top-site Kondo physics into the range experimentally accessible using local probes such as scanning tunneling microscopy. △ Less

Submitted 25 March, 2019; v1 submitted 24 August, 2018; originally announced August 2018.

Comments: 19 pages, 7 figures (added Figs. 6 and 7 to version 1)

Journal ref: Phys. Rev. B 99, 195410 (2019)

Exact solution to an extremal problem on graphic sequences with a realization containing every $2$-tree on $k$ vertices

Authors: De-Yan Zeng, Dong-Yang Zhai, Jian-Hua Yin

Abstract: A simple graph $G$ is an {\it 2-tree} if $G=K_3$, or $G$ has a vertex $v$ of degree 2, whose neighbors are adjacent, and $G-v$ is an 2-tree. Clearly, if $G$ is an 2-tree on $n$ vertices, then $|E(G)|=2n-3$. A non-increasing sequence $π=(d_1,\ldots,d_n)$ of nonnegative integers is a {\it graphic sequence} if it is realizable by a simple graph $G$ on $n$ vertices. Yin and Li (Acta Mathematica Sinica… ▽ More A simple graph $G$ is an {\it 2-tree} if $G=K_3$, or $G$ has a vertex $v$ of degree 2, whose neighbors are adjacent, and $G-v$ is an 2-tree. Clearly, if $G$ is an 2-tree on $n$ vertices, then $|E(G)|=2n-3$. A non-increasing sequence $π=(d_1,\ldots,d_n)$ of nonnegative integers is a {\it graphic sequence} if it is realizable by a simple graph $G$ on $n$ vertices. Yin and Li (Acta Mathematica Sinica, English Series, 25(2009)795--802) proved that if $k\ge 2$, $n\ge \frac{9}{2}k^2+\frac{19}{2}k$ and $π=(d_1,\ldots,d_n)$ is a graphic sequence with $\sum\limits_{i=1}^n d_i>(k-2)n$, then $π$ has a realization containing every 1-tree (the usual tree) on $k$ vertices. Moreover, the lower bound $(k-2)n$ is the best possible. This is a variation of a conjecture due to Erdős and Sós. In this paper, we investigate an analogue problem for $2$-trees and prove that if $k\ge 3$ is an integer with $k\equiv i(\mbox{mod }3)$, $n\geq20\lfloor\frac{k}{3}\rfloor^2+31\lfloor\frac{k}{3}\rfloor+12$ and $π=(d_1,\ldots,d_n)$ is a graphic sequence with $\sum\limits_{i=1}^n d_i>\max\{(k-1)(n-1),2\lfloor\frac{2k}{3}\rfloor n-2n-\lfloor\frac{2k}{3}\rfloor^2+\lfloor\frac{2k}{3}\rfloor+1-(-1)^i\}$, then $π$ has a realization containing every 2-tree on $k$ vertices. Moreover, the lower bound $\max\{(k-1)(n-1),2\lfloor\frac{2k}{3}\rfloor n-2n-\lfloor\frac{2k}{3}\rfloor^2+\lfloor\frac{2k}{3}\rfloor+1-(-1)^i\}$ is the best possible. This result implies a conjecture due to Zeng and Yin (Discrete Math. Theor. Comput. Sci., 17(3)(2016), 315--326). △ Less

Submitted 2 July, 2018; originally announced July 2018.

Comments: 31 page

Local versus extended deformed graphene geometries for valley filtering

Authors: Dawei Zhai, Nancy Sandler

Abstract: The existence of two-inequivalent valleys in the band structure of graphene has motivated the search of mechanisms that allow their separation and control for potential device applications. Among the several schemes proposed in the literature, strain-induced out-of-plane deformations (occurring naturally or intentionally designed in graphene samples), ranks among the best candidates to produce sep… ▽ More The existence of two-inequivalent valleys in the band structure of graphene has motivated the search of mechanisms that allow their separation and control for potential device applications. Among the several schemes proposed in the literature, strain-induced out-of-plane deformations (occurring naturally or intentionally designed in graphene samples), ranks among the best candidates to produce separation of valley currents. Because valley filtering properties in these structures is, however, highly dependent on the type of deformation and setups considered, it is important to identify the relevant factors determining optimal operation and detection of valley currents. In this paper we present a comprehensive comparison of two typical deformations commonly found in graphene samples: local centro-symmetric bubbles and extended folds/wrinkles. Using the Dirac model for graphene and the second-order Born approximation we characterize the scattering properties of the bubble deformation, while numerical transmission matrix methods are used for the fold-like deformations. In both cases, we obtain the dependence of valley polarization on the geometrical parameters of deformations, and discuss their possible experimental realizations. Our study reveals that extended deformations act as better valley filters in broader energy ranges and present more robust features against variations of geometrical parameters and incident current directions. △ Less

Submitted 25 July, 2018; v1 submitted 28 June, 2018; originally announced June 2018.

Comments: 17 pages, 16 figures, figures were adjusted, added a few references, accepted by PRB

Journal ref: Phys. Rev. B 98, 165437 (2018)

Quantum Wires and Waveguides Formed in Graphene by Strain

Authors: Y. Wu, D. Zhai, C. Pan, B. Cheng, T. Taniguchi, K. Watanabe, N. Sandler, M. Bockrath

Abstract: Confinement of electrons in graphene to make devices has proven to be a challenging task. Electrostatic methods fail because of Klein tunneling, while etching into nanoribbons requires extreme control of edge terminations, and bottom-up approaches are limited in size to a few nanometers. Fortunately, its mechanical flexibility raises the possibility of using strain to alter graphene's properties a… ▽ More Confinement of electrons in graphene to make devices has proven to be a challenging task. Electrostatic methods fail because of Klein tunneling, while etching into nanoribbons requires extreme control of edge terminations, and bottom-up approaches are limited in size to a few nanometers. Fortunately, its mechanical flexibility raises the possibility of using strain to alter graphene's properties and create novel straintronic devices. Here, we report transport studies of nanowires created by linearly-shaped strained regions resulting from individual folds formed by layer transfer onto hexagonal boron nitride. Conductance measurements across the folds reveal Coulomb blockade signatures, indicating confined charges within these structures, which act as quantum dots. Along folds, we observe sharp features in traverse resistivity measurements, attributed to an amplification of the dot conductance modulations by a resistance bridge incorporating the device. Our data indicates ballistic transport up to ~1 um along the folds. Calculations using the Dirac model including strain are consistent with measured bound state energies and predict the existence of valley-polarized currents. Our results show that graphene folds can act as straintronic quantum wires. △ Less

Submitted 31 March, 2018; originally announced April 2018.

Journal ref: Nano Lett., 2018, 18, 64