arXiv subjects

Wei He

Publications and source records attributed to Wei He.

At least 235 records · Page 13Linked to original sources

arXiv2019

Tuning Non-Gilbert-type damping in FeGa films on MgO(001) via oblique deposition

The ability to tailor the damping factor is essential for spintronic and spin-torque applications. Here, we report an approach to manipulate the damping factor of FeGa/MgO(001) films by oblique deposition. Owing to the defects at the surface or interface in thin films, two-magnon scattering (TMS) acts as a non-Gilbert damping mechanism in magnetization relaxation. In this work, the contribution of TMS was characterized by in-plane angular dependent ferromagnetic resonance (FMR). It is demonstrated that the intrinsic Gilbert damping is isotropic and invariant, while the extrinsic mechanism related to TMS is anisotropic and can be tuned by oblique deposition. Furthermore, the two and fourfold TMS related to the uniaxial magnetic anisotropy (UMA) and magnetocrystalline anisotropy were discussed. Our results open an avenue to manipulate magnetization relaxation in spintronic devices.

physics.app-ph↗

arXiv2019

Voltage regulation in buck--boost coniverters feeding an unknown constant power load: an adaptive passivity-based control

Rapid developments in power distribution systems and renewable energy have widened the applications of dc--dc buck--boost converters in dc voltage regulation. Applications include vehicular power systems, renewable energy sources that generate power at a low voltage, and dc microgrids. It is noted that the cascade--connection of converters in these applications may cause instability due to the fact that converters acting as loads have a constant power load (CPL) behavior. In this paper, the output voltage regulation problem of a buck--boost converter feeding a CPL is addressed. The construction of the feedback controller is based on the interconnection and damping assignment control technique. Additionally, an immersion and invariance parameter estimator is proposed to compute online the extracted load power, which is difficult to measure in practical applications. It is ensured through the design that the desired operating point is (locally) asymptotically stable with a guaranteed domain of attraction. The approach is validated via computer simulations and experimental prototyping.

eess.SY↗

arXiv2019

On the formation of three-dimensional flows over finite-aspect-ratio wings under tip effects

We perform DNS of flow over finite-aspect-ratio NACA 0015 wings to characterize the tip effects on the wake dynamics. This study focuses on the development of three-dimensional separated flow over the wing, and discuss flow structures formed on the wing surface as well as in the far field wake. Vorticity is introduced into the flow in a predominantly two-dimensional manner. The vortex sheet from the wing tip rolls up around the free end to form the tip vortex. At its inception, the tip vortex is weak and its effect is spatially confined. As the flow around the tip separates, the tip effects extend farther in the spanwise direction, generating noticeable three dimensionality in the wake. For low-aspect-ratio wings, the wake remains stable due to the strong downwash over the entire span. Increasing the aspect ratio allows unsteady vortical flow to emerge away from the tip at sufficiently high angles of attack. These unsteady vortices shed and form closed vortical loops. For higher-aspect-ratio wings, the tip effects retard the near-tip shedding process, which desynchronizes from the two-dimensional shedding over the mid-span region, yielding vortex dislocation. At high angles of attack, the tip vortex exhibits noticeable undulations due to the strong interaction with the unsteady shedding vortices. The spanwise distribution of force coefficients is related to the three-dimensional wake dynamics and the tip effects. Vortical elements in the wake that are responsible for the generation of lift and drag forces are identified through the force element analysis. We note that at high angles of attack, a stationary vortical structure forms at the leading edge near the tip, giving rise to locally high lift and drag forces. The analysis performed here reveals how the vortical flow around the tip influences the separation physics, the global wake dynamics, and the spanwise force distributions.

physics.flu-dyn↗

arXiv2019

Enhancement of ultrafast demagnetization rate and Gilbert damping driven by femtosecond laser-induced spin currents in Fe81Ga19/Ir20Mn80 bilayers

In spintronics applications, ultrafast spin dynamics have to be controlled at femtosecond (fs) timescales via fs-laser radiation. At such ultrafast timescales, the effect of the Gilbert damping factor α on ultrafast demagnetization time should be considered. In previous explorations for the relationship between these two parameters, it was found that the theoretical calculations based on the local spin-flip scattering model do not agree with the experimental results. Here, we find that in Fe81Ga19(FeGa)/Ir20Mn80(IrMn) bilayers, the unconventional IrMn thickness dependence of α results from the competition between spin currents pumped from the ferromagnetic (FM) FeGa layer to the antiferromagnetic (AFM) IrMn layer and those pumped from the AFM layer to the FM layer. More importantly, we establish a proportional relationship between the change of the ultrafast demagnetization rate and the enhancement of Gilbert damping induced by the spin currents via interfacial spin chemical potential . Our work builds a bridge to connect the ultrafast demagnetization time and Gilbert damping in ultrafast photo-induced spin currents dominated systems, which not only explains the disagreement between experimental and theoretical results in the relation of τ_M with α, but provides further insight into ultrafast spin dynamics as well.

cond-mat.mtrl-sci↗

arXiv2019

Invariant Detection with Program Verification Tools

Compilers can specialize programs having invariants for performance improvement. Detecting program invariants that span large and complex code, however, is difficult for compilers. Traditional compilers do not perform very expensive analysis and thus only identify limited invariants, which limits the potential of subsequent optimizations. We would like to address the invariant detection problem via more sophisticated analyses using program verification tools. In this paper, we reveal pitfalls of choosing program verification tools for invariant detection, identify challenges of modeling program behavior using one of these tools---CVC4, and propose some ideas about how to address the challenges.

cs.PL↗

arXiv2019

A new treatment for some periodic Schrödinger operators I: the eigenvalue

We study the problem of how the Floquet property manifests for periodic Schrödinger operators which are known to have multiple of asymptotic spectral solutions. The main conclusions are made for elliptic potentials, we demonstrate that for each period of the elliptic function there is a relation about the Floquet exponent and the monodromy of wave function. Among them there are two relations not explained by the classical Floquet theory. These relations produce both old and new asymptotic solutions consistent with results already known.

math-ph↗

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↗