arXiv subjects

Rui Wu

Publications and source records attributed to Rui Wu.

At least 91 records · Page 5Linked to original sources

arXiv2021

Magnetic coupling in Y$_3$Fe$_5$O$_{12}$/Gd$_3$Fe$_5$O$_{12}$ heterostructures

Ferrimagnetic Y$_3$Fe$_5$O$_{12}$ (YIG) is the prototypical material for studying magnonic properties due to its exceptionally low damping. By substituting the yttrium with other rare earth elements that have a net magnetic moment, we can introduce an additional spin degree of freedom. Here, we study the magnetic coupling in epitaxial Y$_3$Fe$_5$O$_{12}$/Gd$_3$Fe$_5$O$_{12}$ (YIG/GIG) heterostructures grown by pulsed laser deposition. From bulk sensitive magnetometry and surface sensitive spin Seebeck effect (SSE) and spin Hall magnetoresistance (SMR) measurements, we determine the alignment of the heterostructure magnetization through temperature and external magnetic field. The ferromagnetic coupling between the Fe sublattices of YIG and GIG dominates the overall behavior of the heterostructures. Due to the temperature dependent gadolinium moment, a magnetic compensation point of the total bilayer system can be identified. This compensation point shifts to lower temperatures with increasing thickness of YIG due the parallel alignment of the iron moments. We show that we can control the magnetic properties of the heterostructures by tuning the thickness of the individual layers, opening up a large playground for magnonic devices based on coupled magnetic insulators. These devices could potentially control the magnon transport analogously to electron transport in giant magnetoresistive devices.

cond-mat.mtrl-sci↗

arXiv2020

Identifying the origin of the non-monotonic thickness dependence of spin-orbit torques and interfacial Dzyaloshinskii-Moriya interaction in a ferrimagnetic insulator heterostructure

Electrical manipulation of magnetism via spin-orbit torques (SOTs) promises efficient spintronic devices. In systems comprising magnetic insulators and heavy metals, SOTs have started to be investigated only recently, especially in systems with interfacial Dzyaloshinskii-Moriya interaction (iDMI). Here, we quantitatively study the SOT efficiency and iDMI in a series of gadolinium gallium garnet (GGG) / thulium iron garnet (TmIG) / platinum (Pt) heterostructures with varying TmIG and Pt thicknesses. We find that the non-monotonic SOT efficiency as a function of the magnetic layer thickness is not consistent with the 1/thickness dependence expected from a simple interfacial SOT mechanism. Moreover, considering the insulating nature of TmIG, our results cannot be explained by the SOT mechanism established for metallic magnets where the transverse charge spin current can inject and dephase in the magnetic layers. Rather we can explain this non-monotonic behavior by a model based on the interfacial spin mixing conductance that is affected by the thickness-dependent exchange splitting energy by determining the phase difference of the reflected spin-up and spin-down electrons at the TmIG / Pt interface. By studying the Pt thickness dependence, we find that the effective DMI for GGG / TmIG / Pt does not depend on the Pt thickness, which indicates that the GGG / TmIG interface is the source of the iDMI in this system. Our work demonstrates that SOT and DMI can originate from two different interfaces, which enables independent optimization of DMI and SOT for advanced chiral spintronics with low damping magnetic insulators.

cond-mat.mtrl-sci↗

arXiv2020

Learning Heatmap-Style Jigsaw Puzzles Provides Good Pretraining for 2D Human Pose Estimation

The target of 2D human pose estimation is to locate the keypoints of body parts from input 2D images. State-of-the-art methods for pose estimation usually construct pixel-wise heatmaps from keypoints as labels for learning convolution neural networks, which are usually initialized randomly or using classification models on ImageNet as their backbones. We note that 2D pose estimation task is highly dependent on the contextual relationship between image patches, thus we introduce a self-supervised method for pretraining 2D pose estimation networks. Specifically, we propose Heatmap-Style Jigsaw Puzzles (HSJP) problem as our pretext-task, whose target is to learn the location of each patch from an image composed of shuffled patches. During our pretraining process, we only use images of person instances in MS-COCO, rather than introducing extra and much larger ImageNet dataset. A heatmap-style label for patch location is designed and our learning process is in a non-contrastive way. The weights learned by HSJP pretext task are utilised as backbones of 2D human pose estimator, which are then finetuned on MS-COCO human keypoints dataset. With two popular and strong 2D human pose estimators, HRNet and SimpleBaseline, we evaluate mAP score on both MS-COCO validation and test-dev datasets. Our experiments show that downstream pose estimators with our self-supervised pretraining obtain much better performance than those trained from scratch, and are comparable to those using ImageNet classification models as their initial backbones.

cs.CV↗

arXiv2020

Surface Charge Induced Dirac Band Splitting in a Charge Density Wave Material (TaSe4)2I

(TaSe4)2I, a quasi-one-dimensional (1D) crystal, shows a characteristic temperature-driven metal-insulator phase transition. Above the charge density wave (CDW) temperature Tc, (TaSe4)2I has been predicted to harbor a Weyl semimetal phase. Below Tc, it becomes an axion insulator. Here, we performed angle-resolved photoemission spectroscopy (ARPES) measurements on the (110) surface of (TaSe4)2I and observed two sets of Dirac-like energy bands in the first Brillion zone, which agree well with our first-principles calculations. Moreover, we found that each Dirac band exhibits an energy splitting of hundreds of meV under certain circumstances. In combination with core level measurements, our theoretical analysis showed that this Dirac band splitting is a result of surface charge polarization due to the loss of surface iodine atoms. Our findings here shed new light on the interplay between band topology and CDW order in Peierls compounds and will motivate more studies on topological properties of strongly correlated quasi-1D materials.

cond-mat.str-el↗

arXiv2020

Learning Enhanced Resolution-wise features for Human Pose Estimation

Recently, multi-resolution networks (such as Hourglass, CPN, HRNet, etc.) have achieved significant performance on pose estimation by combining feature maps of various resolutions. In this paper, we propose a Resolution-wise Attention Module (RAM) and Gradual Pyramid Refinement (GPR), to learn enhanced resolution-wise feature maps for precise pose estimation. Specifically, RAM learns a group of weights to represent the different importance of feature maps across resolutions, and the GPR gradually merges every two feature maps from low to high resolutions to regress final human keypoint heatmaps. With the enhanced resolution-wise features learnt by CNN, we obtain more accurate human keypoint locations. The efficacies of our proposed methods are demonstrated on MS-COCO dataset, achieving state-of-the-art performance with average precision of 77.7 on COCO val2017 set and 77.0 on test-dev2017 set without using extra human keypoint training dataset.

cs.CV↗

arXiv2020

Role of anion in the pairing interaction of iron-based superconductivity

High-temperature iron-based superconductivity develops in a structure with unusual lattice-orbital geometry, based on a planar layer of Fe atoms with 3d orbitals and tetrahedrally coordinated by anions. Here we elucidate the electronic role of anions in the iron-based superconductors utilizing state-of-the-art scanning tunneling microscopy. By measuring the local electronic structure, we find that As anion in Ba0.4K0.6Fe2As2 has a striking impact on the electron pairing. The superconducting electronic feature can be switched off/on by removing/restoring As atoms on Fe layer at the atomic scale. Our analysis shows that this remarkable atomic switch effect is related to the geometrical cooperation between anion mediated hopping and unconventional pairing interaction. Our results uncover that the local Fe-anion coupling is fundamental for the pairing interaction of iron-based superconductivity, and promise the potential of bottom-up engineering of electron pairing.

cond-mat.supr-con↗

arXiv2020

MCEN: Bridging Cross-Modal Gap between Cooking Recipes and Dish Images with Latent Variable Model

Nowadays, driven by the increasing concern on diet and health, food computing has attracted enormous attention from both industry and research community. One of the most popular research topics in this domain is Food Retrieval, due to its profound influence on health-oriented applications. In this paper, we focus on the task of cross-modal retrieval between food images and cooking recipes. We present Modality-Consistent Embedding Network (MCEN) that learns modality-invariant representations by projecting images and texts to the same embedding space. To capture the latent alignments between modalities, we incorporate stochastic latent variables to explicitly exploit the interactions between textual and visual features. Importantly, our method learns the cross-modal alignments during training but computes embeddings of different modalities independently at inference time for the sake of efficiency. Extensive experimental results clearly demonstrate that the proposed MCEN outperforms all existing approaches on the benchmark Recipe1M dataset and requires less computational cost.

cs.CV↗

arXiv2020

Experimental observations indicating the topological nature of the edge states on HfTe5

The topological edge states of two-dimensional topological insulators with large energy gap furnish ideal conduction channels for dissipationless current transport. Transition metal tellurides XTe5 (X=Zr, Hf) are theoretically predicted to be large-gap two-dimensional topological insulators and the experimental observations of their bulk insulating gap and in-gap edge states have been reported, but the topological nature of these edge states still remains to be further elucidated. Here, we report our low temperature scanning tunneling microscopy/spectroscopy study on single crystals of HfTe5. We demonstrate a full energy gap of ~80 meV near the Fermi level on the surface monolayer of HfTe5 and that such insulating energy gap gets filled with finite energy states when measured at the monolayer step edges. Remarkably, such states are absent at the edges of a narrow monolayer strip of one-unit-cell in width but persist at both step edges of a unit-cell wide monolayer groove. These experimental observations strongly indicate that the edge states of HfTe5 monolayers are not trivially caused by translational symmetry breaking, instead they are topological in nature protected by the 2D nontrivial bulk properties.

cond-mat.mtrl-sci↗

arXiv2020

Drying of porous media by concurrent drainage and evaporation: A pore network modeling study

Drainage and evaporation can occur simultaneously during the drying of porous media, but the interactions between these processes and their effects on drying are rarely studied. In this work, we develop a pore network model that considers drainage, evaporation, and rarefied multi-component gas transport in porous media with nanoscale pores. Using this model, we investigate the drying of a liquid solvent-saturated porous medium enabled by the flow of purge gas through it. Simulations show that drying progresses in three stages, and the solvent removal by drainage effects (evaporation effects) becomes increasingly weak (strong) as drying progresses through these stages. Interestingly, drainage can contribute considerably to solvent removal even after evaporation effects become very strong, especially when the applied pressure difference across the porous medium is low. We show that these phenomena are the results of the coupling between the drainage and evaporation effects and this coupling depends on the operating conditions and the stage of drying.

cond-mat.soft↗

arXiv2020

Data Efficient Training for Reinforcement Learning with Adaptive Behavior Policy Sharing

Deep Reinforcement Learning (RL) is proven powerful for decision making in simulated environments. However, training deep RL model is challenging in real world applications such as production-scale health-care or recommender systems because of the expensiveness of interaction and limitation of budget at deployment. One aspect of the data inefficiency comes from the expensive hyper-parameter tuning when optimizing deep neural networks. We propose Adaptive Behavior Policy Sharing (ABPS), a data-efficient training algorithm that allows sharing of experience collected by behavior policy that is adaptively selected from a pool of agents trained with an ensemble of hyper-parameters. We further extend ABPS to evolve hyper-parameters during training by hybridizing ABPS with an adapted version of Population Based Training (ABPS-PBT). We conduct experiments with multiple Atari games with up to 16 hyper-parameter/architecture setups. ABPS achieves superior overall performance, reduced variance on top 25% agents, and equivalent performance on the best agent compared to conventional hyper-parameter tuning with independent training, even though ABPS only requires the same number of environmental interactions as training a single agent. We also show that ABPS-PBT further improves the convergence speed and reduces the variance.

cs.LG↗

arXiv2020

Influence of Atomic Roughness at The Uncompensated Fe/CoO (111) Interface on Exchange Bias Effect

The effect of interface roughness of ferromagnetic and antiferromagnetic layers on exchange bias is still not well understood. In this report we have investigated the effect of surface roughness in (111)-oriented antiferromagnetic CoO films on exchange bias with ferromagnetic Fe grown on top. The surface roughness is controlled at the atomic scale, over a range below ~ 0.35 nm, by varying layer thickness of the CoO films. It is observed that both exchange bias field ($H_{E}$) and coercivity ($H_{C}$) extensively depend on the atomic scale roughness of the CoO (111) at the interface with Fe film. An opposite dependence of $H_{E}$ and $H_{C}$ on interface roughness was found, which was ascribed to partially compensated spin states induced by the atomic roughness at the fully uncompensated CoO (111) surfaces and was corroborated using the Monte Carlo simulations. Moreover, the onset temperature for $H_{C}$ is found to be up to ~ 80 K below the blocking temperature ($T_{B}$) and the temperature dependence of $H_{C}$ follows the power law with a critical exponent equal to one, which indicates that, in this system, $H_{C}$ is more of an interface-related property than $H_{E}$.

cond-mat.mtrl-sci↗

arXiv2019

Bubble formation due to capillary instability during evaporation of a porous medium

We show that during evaporation of a pore network, liquid can refill the gas occupied pores, snapping off a gas bubble, which then moves to a stable configuration. This phenomenon is induced by the capillary instability due to the wettability heterogeneity of the pore network and has a much smaller time scale as compared to the evaporation process. The capillary instability induced liquid refilling and bubble movement are explained in detail based on the analysis of the images obtained from the visualization experiment. The capillary valve effect, which hinders the movement of the gas-liquid interface and is induced by the sudden geometrical expansion between small and large pores, can be suppressed by the residual liquid in the large pore. For better understanding of the capillary instability induced gas-liquid two-phase transport during evaporation, a novel pore network model is developed, which considers not only the capillary and viscous forces but also the inertial forces that are seldom taken into account in the previous models. The pore network modeling results are in good agreement with the experimental data, demonstrating the effectiveness of the developed pore network model, which opens up a new route for better understanding of the role of inertial forces in two-phase transport in porous media.

cond-mat.soft↗

arXiv2019

Leidenfrost temperature: surface thermal diffusivity and effusivity effect

Droplet impact on hot surfaces results in either droplet-surface contact or droplet-surface non-contact, i.e., the Leidenfrost state. The Leidenfrost droplet is levitated upon its vapor, deteriorating the heat transfer. The Leidenfrost temperature depends on the thermal properties of the surface, which are usually characterized by two parameters: the thermal diffusivity and the thermal effusivity. In this paper, the effects of these two parameters on the Leidenfrost temperature are clarified experimentally by varying the one of interest while keeping the other one constant. The experimental results indicate that the Leidenfrost temperature is affected by the thermal effusivity rather than by the thermal diffusivity; the higher the thermal effusivity is, the lower the Leidenfrost temperature; and the increase of the Leidenfrost temperature with the droplet impact velocity is greater for the surface with the lower thermal effusivity. To further understand the experimental findings, a theoretical model is proposed, which considers the transient heat transfer in the surface. The theoretical analysis shows that the Leidenfrost temperature scales as the inverse of the thermal effusivity and the root square of the impact velocity, in good agreement with the experimental results.

physics.flu-dyn↗

arXiv2019

Pore network model of evaporation in porous media with continuous and discontinuous corner films

During evaporation in porous media, two types of corner films are distinguished. A continuous corner film is connected to the bulk liquid, while a discontinuous one is not. To disclose their effects on evaporation in porous media, a pore network model with both continuous and discontinuous corner films is developed, which considers the capillary and viscous forces as well as the effects of corner films on the threshold pressures of pores. The capillary valve effect induced by the sudden geometrical expansion between the small and large pores is also taken into account in the model. The developed pore network model agrees well with the evaporation experiment with a quasi 2D micro model porous medium, in terms of not only the variation of the liquid saturation in each pore but also the variation of the total evaporation rate. The pore network models that neglect the corner films or the liquid viscosity are also compared with the experiment so as to shed light on the roles of the corner films. The continuous corner films, which contribute to sustain the high evaporation rate, can be interrupted to be the discontinuous ones not only by the gas invasion into pores but also by the capillary scissors effect due to the local convex topology of the solid matrix.

physics.flu-dyn↗

arXiv2019

RecSim: A Configurable Simulation Platform for Recommender Systems

We propose RecSim, a configurable platform for authoring simulation environments for recommender systems (RSs) that naturally supports sequential interaction with users. RecSim allows the creation of new environments that reflect particular aspects of user behavior and item structure at a level of abstraction well-suited to pushing the limits of current reinforcement learning (RL) and RS techniques in sequential interactive recommendation problems. Environments can be easily configured that vary assumptions about: user preferences and item familiarity; user latent state and its dynamics; and choice models and other user response behavior. We outline how RecSim offers value to RL and RS researchers and practitioners, and how it can serve as a vehicle for academic-industrial collaboration.

cs.LG↗

arXiv2019

Exploiting Offset-guided Network for Pose Estimation and Tracking

Human pose estimation has witnessed a significant advance thanks to the development of deep learning. Recent human pose estimation approaches tend to directly predict the location heatmaps, which causes quantization errors and inevitably deteriorates the performance within the reduced network output. Aim at solving it, we revisit the heatmap-offset aggregation method and propose the Offset-guided Network (OGN) with an intuitive but effective fusion strategy for both two-stages pose estimation and Mask R-CNN. For two-stages pose estimation, a greedy box generation strategy is also proposed to keep more necessary candidates while performing person detection. For mask R-CNN, ratio-consistent is adopted to improve the generalization ability of the network. State-of-the-art results on COCO and PoseTrack dataset verify the effectiveness of our offset-guided pose estimation and tracking.

cs.CV↗

arXiv2019

Reinforcement Learning for Slate-based Recommender Systems: A Tractable Decomposition and Practical Methodology

Most practical recommender systems focus on estimating immediate user engagement without considering the long-term effects of recommendations on user behavior. Reinforcement learning (RL) methods offer the potential to optimize recommendations for long-term user engagement. However, since users are often presented with slates of multiple items - which may have interacting effects on user choice - methods are required to deal with the combinatorics of the RL action space. In this work, we address the challenge of making slate-based recommendations to optimize long-term value using RL. Our contributions are three-fold. (i) We develop SLATEQ, a decomposition of value-based temporal-difference and Q-learning that renders RL tractable with slates. Under mild assumptions on user choice behavior, we show that the long-term value (LTV) of a slate can be decomposed into a tractable function of its component item-wise LTVs. (ii) We outline a methodology that leverages existing myopic learning-based recommenders to quickly develop a recommender that handles LTV. (iii) We demonstrate our methods in simulation, and validate the scalability of decomposed TD-learning using SLATEQ in live experiments on YouTube.

cs.LG↗

arXiv2019

On the Capacity Region of Broadcast Packet Erasure Relay Networks With Feedback

We derive a new outer bound on the capacity region of broadcast traffic in multiple input broadcast packet erasure channels with feedback, and extend this outer bound to packet erasure relay networks with feedback. We show the tightness of the outer bound for various classes of networks. An important engineering implication of this work is that for network coding schemes for parallel broadcast channels, the `xor' packets should be sent over correlated broadcast subchannels.

cs.IT↗