arXiv subjects

Wei He

Publications and source records attributed to Wei He.

At least 235 records · Page 13Linked to original sources

arXiv2019

A new treatment for some periodic Schrödinger operators II: the wave function

Following the approach of our previous paper we continue to study the asymptotic solution of periodic Schrödinger operators. Using the eigenvalues obtained earlier the corresponding asymptotic wave functions are derived. This gives further evidence in favor of the monodromy relations for the Floquet exponent proposed in the previous paper. In particular, the large energy asymptotic wave functions are related to the instanton partition function of N=2 supersymmetric gauge theory with surface operator. A relevant number theoretic dessert is appended.

math-ph↗

arXiv2019

Non-local Meets Global: An Integrated Paradigm for Hyperspectral Denoising

Non-local low-rank tensor approximation has been developed as a state-of-the-art method for hyperspectral image (HSI) denoising. Unfortunately, with more spectral bands for HSI, while the running time of these methods significantly increases, their denoising performance benefits little. In this paper, we claim that the HSI underlines a global spectral low-rank subspace, and the spectral subspaces of each full band patch groups should underlie this global low-rank subspace. This motivates us to propose a unified spatial-spectral paradigm for HSI denoising. As the new model is hard to optimize, we further propose an efficient algorithm for optimization, which is motivated by alternating minimization. This is done by first learning a low-dimensional projection and the related reduced image from the noisy HSI. Then, the non-local low-rank denoising and iterative regularization are developed to refine the reduced image and projection, respectively. Finally, experiments on synthetic and both real datasets demonstrate the superiority against the other state-of-the-arts HSI denoising methods.

cs.CV↗

arXiv2018

Global instability of long separation bubbles in a laminar boundary layer

This work aims to numerically investigate the linear global instability of long separation bubbles origin from the changes in the adverse pressure gradient inside a laminar flat plate boundary layer disturbed by placing a bluff body with small clearances. The boundary layer starts at $Re_{δ^\ast}$=121.7 and is steady in the computational extent after being disturbed by a NACA 4415 airfoil and a cylinder with clearance $h\leqslant 0.2$, respectively. It is found that the dominant stability is associated with two- and three-dimensional stationary eigenmodes. There is a less damped oscillation mode at a small wavenumbers range. The strong effect of three-dimensionality is further confirmed in a non-modal analysis framework. Transient growth analysis shows that the most strongest optimal perturbation is three-dimensional in the range of parameters studied. The modal analysis reveals that the instability grows with reducing the clearance, while the optimal perturbations grow with increasing the clearance. It is found that the separated boundary layer stability analysis is less dependent on the bluff body geometries.

physics.flu-dyn↗

arXiv2018

The quantum entropic uncertainty relation and entanglement witness in the two-atom system coupling with the non-Markovian environments

The quantum entropic uncertainty relation and entanglement witness in the two-atom system coupling with the non-Markovian environments are studied by the time-convolutionless master-equation approach. The influence of non-Markovian effect and detuning on the lower bound of the quantum entropic uncertainty relation and entanglement witness is discussed in detail. The results show that, only if the two non-Markovian reservoirs are identical, increasing detuning and non-Markovian effect can reduce the lower bound of the entropic uncertainty relation, lengthen the time region during which the entanglement can be witnessed, and effectively protect the entanglement region witnessed by the lower bound of the entropic uncertainty relation. The results can be applied in quantum measurement, quantum cryptography task and quantum information processing.

quant-ph↗

arXiv2018

Isotropic non-local Gilbert damping driven by spin currents in epitaxial Pd/Fe/MgO(001) films

Although both theoretical predications and experimental observations demonstrated that the damping factor is anisotropic at ferromagnet/semiconductor interface with robust interfacial spin-orbit coupling, it is not well understood whether non-local Gilbert damping driven by spin currents in heavy metal/ferromagnetic metal (HM/FM) bilayers is anisotropic or not. Here, we investigated the in-plane angular- and frequency- dependence of magnetic relaxation of epitaxial Fe/MgO(001) films with different capping layers of Pd and Cu. After disentangling the parasitic contributions, such as two-magnon scattering (TMS), mosaicity, and field-dragging effect, we unambiguously observed that both local and non-local Gilbert damping are isotropic in Fe(001) plane, suggesting that the pure spin currents absorption is independent of Fe magnetization orientation in the epitaxial Pd/Fe heterostructure. First principles calculation reveals that the effective spin mixing conductance of Pd/Fe interface is nearly invariant for different magnetization directions in good agreement with the experimental observations. These results offer a valuable insight into the transmission and absorption of pure spin currents, and facilitate us to utilize next-generation spintronic devices.

cond-mat.mtrl-sci↗

arXiv2018

Active Damping of a DC Network with a Constant Power Load: An Adaptive Passivity-based Control Approach

This paper proposes a nonlinear, adaptive controller to increase the stability margin of a direct-current (DC) small-scale electrical network containing a constant power load, whose value is unknown. Due to their negative incremental impedance, constant power loads are known to reduce the effective damping of a network, leading to voltage oscillations and even to network collapse. To tackle this problem, we consider the incorporation of a controlled DC-DC power converter between the feeder and the constant power load. The design of the control law for the converter is based on the use of standard Passivity-Based Control and Immersion and Invariance theories. The good performance of the controller is evaluated with numerical simulations.

eess.SY↗

arXiv2018

Surpassing the Ambient Packing Limit of Energetic Crystals: Squeezing Effect of Molecular Level "Net-fishing"

High energy density is always a key goal for developments of energy storage or energetic materials (EMs). Except exploring novel EMs with high chemical energy, it is also desirable if the traditional EMs could be assembled at a higher density. However, it is very difficult to surpass their theoretical maximum molecular packing density under ambient conditions, even though a higher density could be achieved under ultra-high pressure (Gigapascals). Such solid-state phase changes are reversible, and hence this high density packing is not able to maintain under ambient conditions. Alternatively, in this research, we demonstrated a molecular level compression effect by stacking of 2-D TAGP, resulting in a higher density packing of the HMX molecules with changed conformation. The HMX crystal formed under compression in the solvent has a unit cell parameter very close to the reported one observed under pressure of 0.2 GPa. It shows that the compressed HMX molecules are trapped in the TAGP layers, resulting in a higher density (e.g. 2.13 g cm-3) and also higher heat of formation. The resulted HMX crystals are free of defects, and unlike the pristine HMX, no polymorphic transition and melting point were observed upon heating. Experiments and relevant calculations show that the best resulted hybrid HMX crystal has a detonation velocity of 10.40 km s-1 and pressure of 53.9 GPa, respectively. Its ground specific impulse reaches about 292 s, even much better than CL-20, making it a promising propellant component for future space explorations.

cond-mat.mtrl-sci↗

arXiv2018

Synthesis-Dependent Properties of Barlowite and Zn-Substituted Barlowite

The mineral barlowite, Cu$_4$(OH)$_6$FBr, has been the focus of recent attention due to the possibility of substituting the interlayer Cu$^{2+}$ site with non-magnetic ions to develop new quantum spin liquid materials. We re-examine previous methods of synthesizing barlowite and describe a novel hydrothermal synthesis method that produces large single crystals of barlowite and Zn-substituted barlowite (Cu$_3$Zn$_x$Cu$_{1-x}$(OH)$_6$FBr). The two synthesis techniques yield barlowite with indistinguishable crystal structures and spectroscopic properties at room temperature; however, the magnetic ordering temperatures differ by 4 K and the thermodynamic properties are clearly different. The dependence of properties upon synthetic conditions implies that the defect chemistry of barlowite and related materials is complex and significant. Zn-substituted barlowite exhibits a lack of magnetic order down to T = 2 K, characteristic of a quantum spin liquid, and we provide a synthetic route towards producing large crystals suitable for neutron scattering.

cond-mat.str-el↗

arXiv2018

Highly Scalable Deep Learning Training System with Mixed-Precision: Training ImageNet in Four Minutes

Synchronized stochastic gradient descent (SGD) optimizers with data parallelism are widely used in training large-scale deep neural networks. Although using larger mini-batch sizes can improve the system scalability by reducing the communication-to-computation ratio, it may hurt the generalization ability of the models. To this end, we build a highly scalable deep learning training system for dense GPU clusters with three main contributions: (1) We propose a mixed-precision training method that significantly improves the training throughput of a single GPU without losing accuracy. (2) We propose an optimization approach for extremely large mini-batch size (up to 64k) that can train CNN models on the ImageNet dataset without losing accuracy. (3) We propose highly optimized all-reduce algorithms that achieve up to 3x and 11x speedup on AlexNet and ResNet-50 respectively than NCCL-based training on a cluster with 1024 Tesla P40 GPUs. On training ResNet-50 with 90 epochs, the state-of-the-art GPU-based system with 1024 Tesla P100 GPUs spent 15 minutes and achieved 74.9\% top-1 test accuracy, and another KNL-based system with 2048 Intel KNLs spent 20 minutes and achieved 75.4\% accuracy. Our training system can achieve 75.8\% top-1 test accuracy in only 6.6 minutes using 2048 Tesla P40 GPUs. When training AlexNet with 95 epochs, our system can achieve 58.7\% top-1 test accuracy within 4 minutes, which also outperforms all other existing systems.

cs.LG↗

arXiv2018

Multi-temporal Sentinel-1 and -2 Data Fusion for Optical Image Simulation

In this paper, we present the optical image simulation from a synthetic aperture radar (SAR) data using deep learning based methods. Two models, i.e., optical image simulation directly from the SAR data and from multi-temporal SARoptical data, are proposed to testify the possibilities. The deep learning based methods that we chose to achieve the models are a convolutional neural network (CNN) with a residual architecture and a conditional generative adversarial network (cGAN). We validate our models using the Sentinel-1 and -2 datasets. The experiments demonstrate that the model with multi-temporal SAR-optical data can successfully simulate the optical image, meanwhile, the model with simple SAR data as input failed. The optical image simulation results indicate the possibility of SARoptical information blending for the subsequent applications such as large-scale cloud removal, and optical data temporal superresolution. We also investigate the sensitivity of the proposed models against the training samples, and reveal possible future directions.

cs.CV↗

arXiv2018

DuReader: a Chinese Machine Reading Comprehension Dataset from Real-world Applications

This paper introduces DuReader, a new large-scale, open-domain Chinese ma- chine reading comprehension (MRC) dataset, designed to address real-world MRC. DuReader has three advantages over previous MRC datasets: (1) data sources: questions and documents are based on Baidu Search and Baidu Zhidao; answers are manually generated. (2) question types: it provides rich annotations for more question types, especially yes-no and opinion questions, that leaves more opportunity for the research community. (3) scale: it contains 200K questions, 420K answers and 1M documents; it is the largest Chinese MRC dataset so far. Experiments show that human performance is well above current state-of-the-art baseline systems, leaving plenty of room for the community to make improvements. To help the community make these improvements, both DuReader and baseline systems have been posted online. We also organize a shared competition to encourage the exploration of more models. Since the release of the task, there are significant improvements over the baselines.

cs.CL↗

arXiv2018

Multi-Passage Machine Reading Comprehension with Cross-Passage Answer Verification

Machine reading comprehension (MRC) on real web data usually requires the machine to answer a question by analyzing multiple passages retrieved by search engine. Compared with MRC on a single passage, multi-passage MRC is more challenging, since we are likely to get multiple confusing answer candidates from different passages. To address this problem, we propose an end-to-end neural model that enables those answer candidates from different passages to verify each other based on their content representations. Specifically, we jointly train three modules that can predict the final answer based on three factors: the answer boundary, the answer content and the cross-passage answer verification. The experimental results show that our method outperforms the baseline by a large margin and achieves the state-of-the-art performance on the English MS-MARCO dataset and the Chinese DuReader dataset, both of which are designed for MRC in real-world settings.

cs.CL↗

arXiv2018

Bilateral Teleoperation of Multiple Robots under Scheduling Communication

In this paper, bilateral teleoperation of multiple slaves coupled to a single master under scheduling communication is investigated. The sampled-data transmission between the master and the multiple slaves is fulfilled over a delayed communication network, and at each sampling instant, only one slave is allowed to transmit its current information to the master side according to some scheduling protocols. To achieve the master-slave synchronization, Round-Robin scheduling protocol and Try-Once-Discard scheduling protocol are employed, respectively. By designing a scheduling-communication-based controller, some sufficient stability criteria related to the controller gain matrices, sampling intervals, and communication delays are obtained for the closed-loop teleoperation system under Round-Robin and Try-Once-Discard scheduling protocols, respectively. Finally, simulation studies are given to validate the effectiveness of the proposed results.

eess.SY↗

arXiv2017

An Adaptive Passivity-Based Controller of a Buck-Boost Converter With a Constant Power Load

This paper addresses the problem of regulating the output voltage of a DC-DC buck-boost converter feeding a constant power load,which is a problem of current practical interest. Designing a stabilising controller is theoretically challenging because its average model is a bilinear second order system that, due to the presence of the constant power load,is non- minimum phase with respect to both states.Moreover,to design a high performance controller, the knowledge of the extracted load power, which is difficult to measure in industrial applications, is required. In this paper, an adaptive interconnection and damping assignment passivity based control that incorporates the immersion and invariance parameter estimator for the load power is proposed to solve the problem. Some detailed simulations are provided to validate the transient behaviour of the proposed controller and compare it with the performance of a classical PD scheme.

math.OC↗

arXiv2017

Unifying ultrafast demagnetization and intrinsic Gilbert damping in Co/Ni bilayers with electronic relaxation near the Fermi surface

The ability to controllably manipulate the laser-induced ultrafast magnetic dynamics is a prerequisite for future high speed spintronic devices. The optimization of devices requires the controllability of the ultrafast demagnetization time, , and intrinsic Gilbert damping, . In previous attempts to establish the relationship between and , the rare-earth doping of a permalloy film with two different demagnetization mechanism is not a suitable candidate. Here, we choose Co/Ni bilayers to investigate the relations between and by means of time-resolved magneto-optical Kerr effect (TRMOKE) via adjusting the thickness of the Ni layers, and obtain an approximately proportional relation between these two parameters. The remarkable agreement between TRMOKE experiment and the prediction of breathing Fermi-surface model confirms that a large Elliott-Yafet spin-mixing parameter is relevant to the strong spin-orbital coupling at the Co/Ni interface. More importantly, a proportional relation between and in such metallic films or heterostructures with electronic relaxation near Fermi surface suggests the local spin-flip scattering domains the mechanism of ultrafast demagnetization, otherwise the spin-current mechanism domains. It is an effective method to distinguish the dominant contributions to ultrafast magnetic quenching in metallic heterostructures by investigating both the ultrafast demagnetization time and Gilbert damping simultaneously. Our work can open a novel avenue to manipulate the magnitude and efficiency of Terahertz emission in metallic heterostructures such as the perpendicular magnetic anisotropic Ta/Pt/Co/Ni/Pt/Ta multilayers, and then it has an immediate implication of the design of high frequency spintronic devices.

cond-mat.mes-hall↗

arXiv2017

The Necessity of Nowhere Equivalence

We prove some regularity properties (convexity, closedness, compactness and preservation of upper hemicontinuity) for distribution and regular conditional distribution of correspondences under the nowhere equivalence condition. We show the necessity of such a condition for any of these properties to hold. As an application, we demonstrate that the nowhere equivalence condition is satisfied on the underlying agent space if and only if pure-strategy Nash equilibria exist in general large games with any fixed uncountable compact action space.

math.PR↗

arXiv2017

High pressure x-ray study of spin-Peierls physics in the quantum spin chain material TiOCl

The application of pressure can induce transitions between unconventional quantum phases in correlated materials. The inorganic compound TiOCl, composed of chains of S=1/2 Ti ions, is an ideal realization of a spin-Peierls system with a relatively simple unit cell. At ambient pressure, it is an insulator due to strong electronic interactions (a Mott insulator). Its resistivity shows a sudden decrease with increasing pressure, indicating a transition to a more metallic state which may coincide with the emergence of charge density wave order. Therefore, high pressure studies of the structure with x-rays are crucial in determining the ground-state physics in this quantum magnet. In ambient pressure, TiOCl exhibits a transition to an incommensurate nearly dimerized state at $T_{c2}=92$ K and to a commensurate dimerized state at $T_{c1}=66$ K. Here, we discover a rich phase diagram as a function of temperature and pressure using x-ray diffraction on a single crystal in a diamond anvil cell down to $T=4$ K and pressures up to 14.5 GPa. Remarkably, the magnetic interaction scale increases dramatically with increasing pressure, as indicated by the high onset temperature of the spin-Peierls phase. At $\sim$7 GPa, the extrapolated onset of the spin-Peierls phase occurs above $T=300$ K, indicating a quantum singlet state exists at room temperature. Further comparisons are made with the phase diagrams of related spin-Peierls systems that display metallicity and superconductivity under pressure.

cond-mat.str-el↗

arXiv2017

Multiple Fault Attack on PRESENT with a Hardware Trojan Implementation in FPGA

Internet of Things connects lots of small constrained devices to the Internet. As in any other environment, communication security is important and cryptographic algorithms are one of many elements that we use in order to keep messages secure. Because of the constrained nature of these environments, it is necessary to use algorithms that do not require high computational power. Lightweight ciphers are therefore ideal candidates for this purpose. In this paper, we explore a possibility of attacking an ultra-lightweight cipher PRESENT by using a multiple fault attack. Utilizing the Differential Fault Analysis technique, we were able to recover the secret key with two faulty encryptions and an exhaustive search of 2^16 remaining key bits. Our attack aims at four nibbles in the penultimate round of the cipher, causing faulty output in all nibbles of the output. We also provide a practical attack scenario by exploiting Hardware Trojan (HT) technique for the proposed fault injection in a Xilinx Spartan-6 FPGA.

cs.CR↗