arXiv subjects

Kun Zhou

Publications and source records attributed to Kun Zhou.

303 records · Page 17Linked to original sources

arXiv2018

H-CNN: Spatial Hashing Based CNN for 3D Shape Analysis

We present a novel spatial hashing based data structure to facilitate 3D shape analysis using convolutional neural networks (CNNs). Our method well utilizes the sparse occupancy of 3D shape boundary and builds hierarchical hash tables for an input model under different resolutions. Based on this data structure, we design two efficient GPU algorithms namely hash2col and col2hash so that the CNN operations like convolution and pooling can be efficiently parallelized. The spatial hashing is nearly minimal, and our data structure is almost of the same size as the raw input. Compared with state-of-the-art octree-based methods, our data structure significantly reduces the memory footprint during the CNN training. As the input geometry features are more compactly packed, CNN operations also run faster with our data structure. The experiment shows that, under the same network structure, our method yields comparable or better benchmarks compared to the state-of-the-art while it has only one-third memory consumption. Such superior memory performance allows the CNN to handle high-resolution shape analysis.

cs.GR↗

arXiv2018

DeepWarp: DNN-based Nonlinear Deformation

DeepWarp is an efficient and highly re-usable deep neural network (DNN) based nonlinear deformable simulation framework. Unlike other deep learning applications such as image recognition, where different inputs have a uniform and consistent format (e.g. an array of all the pixels in an image), the input for deformable simulation is quite variable, high-dimensional, and parametrization-unfriendly. Consequently, even though DNN is known for its rich expressivity of nonlinear functions, directly using DNN to reconstruct the force-displacement relation for general deformable simulation is nearly impossible. DeepWarp obviates this difficulty by partially restoring the force-displacement relation via warping the nodal displacement simulated using a simplistic constitutive model -- the linear elasticity. In other words, DeepWarp yields an incremental displacement fix based on a simplified (therefore incorrect) simulation result other than returning the unknown displacement directly. We contrive a compact yet effective feature vector including geodesic, potential and digression to sort training pairs of per-node linear and nonlinear displacement. DeepWarp is robust under different model shapes and tessellations. With the assistance of deformation substructuring, one DNN training is able to handle a wide range of 3D models of various geometries including most examples shown in the paper. Thanks to the linear elasticity and its constant system matrix, the underlying simulator only needs to perform one pre-factorized matrix solve at each time step, and DeepWarp is able to simulate large models in real time.

cs.GR↗

arXiv2018

A First-Principles Study on the Adsorption of Small Molecules on Antimonene: Oxidation Tendency and Stability

Antimonene, a new group-VA 2D semiconducting material beyond phosphorene, was recently synthesized through various approaches and was shown to exhibit a good structural integrity in ambient conditions and various interesting properties. In this work, we perform systematical first-principles investigations on the interactions of antimonene with the small molecules CO, NO, NO2, H2O, O2, NH3 and H2. It is found that NO, NO2, H2O, O2, and NH3 serve as charge acceptors, while CO shows a negligible charge transfer. H2 acts as a charge donor to antimonene with the amount of charge transfer being ten times that of H2 on phosphorene. The interaction of the O2 molecule with antimonene is much stronger than that with phosphorene. Surprisingly, the kinetic barrier for the splitting of the O2 molecule on antimonene is low (~0.40 eV), suggesting that pristine antimonene may undergo oxidation in ambient conditions, especially at elevated temperatures. Fortunately, the acceptor role of H2O on antimonene, opposite to a donor role in phosphorene, helps to suppress further structural degradation of the oxidized antimonene by preventing the proton transfer between water molecules and oxygen species to form acids.

cond-mat.mtrl-sci↗

arXiv2018

Interaction of phonons with discrete breather in strained graphene

We numerically analyze the interaction of small-amplitude phonon waves with standing gap discrete breather (DB) in strained graphene. To make the system support gap DB, strain is applied to create a gap in the phonon spectrum. We only focus on the in-plane phonons and DB, so the issue is investigated under a quasi-one-dimensional setup. It is found that, for the longitudinal sound waves having frequencies below 6 THz, DB is transparent and thus no radiation of energy from DB takes place; whereas for those sound waves with higher frequencies within the acoustic (optical) phonon band, phonon is mainly transmitted (reflected) by DB, and concomitantly, DB radiates its energy when interacting with phonons. The latter case is supported by the fact that, the sum of the transmitted and reflected phonon energy densities is noticeably higher than that of the incident wave. Our results here may provide insight into energy transport in graphene when the spatially localized nonlinear vibration modes are presented.

cond-mat.mes-hall↗

arXiv2018

Graphene Nanoribbon as an Elastic Damper

Heterostructures composed of dissimilar two-dimensional nanomaterials can have nontrivial physical and mechanical properties promising for many applications. Interestingly, in some cases, it is possible to create heterostructures composed of weakly and strongly stretched domains with the same chemical composition, as it has been demonstrated for some polymer chains, DNA, and intermetallic nanowires supporting this effect of two-phase stretching. These materials at relatively strong tension forces split into domains with smaller and larger tensile strain. Within this region, average strain increases at constant tensile force due to the growth of the domain with the larger strain in the expense of the domain with smaller strain. Here the two-phase stretching phenomenon is described for graphene nanoribbons with the help of molecular dynamics simulations. This unprecedented feature of graphene revealed in our study is related to the peculiarities of nucleation and motion of the domain walls separating the domains with different elastic strain. It turns out that the loading-unloading curves exhibit a hysteresis-like behavior due to the energy dissipation during the domain wall nucleation and motion. Here, we originally put forward the idea of implementing graphene nanoribbons as elastic dampers, efficiently converting mechanical strain energy into heat during cyclic loading-unloading through elastic extension where domains with larger and smaller strain coexist. Furthermore, in the regime of two-phase stretching, graphene nanoribbon is a heterostructure for which the fraction of domains with larger and smaller strain, and consequently its physical and mechanical properties, can be tuned in a controllable manner by applying elastic strain and/or heat.

cond-mat.mtrl-sci↗

arXiv2017

Exploring the Charge Localization and Band Gap Opening of Borophene: A First-Principles Study

Recently synthesized two-dimensional (2D) boron, borophene, exhibits a novel metallic behavior rooted in the s-p orbital hybridization, distinctively different from other 2D materials such as sulfides/selenides and semi-metallic graphene. This unique feature of borophene implies new routes for charge delocalization and band gap opening. Herein, using first-principles calculations, we explore the routes to localize the carriers and open the band gap of borophene via chemical functionalization, ribbon construction, and defect engineering. The metallicity of borophene is found to be remarkably robust against H- and F-functionalization and the presence of vacancies. Interestingly, a strong odd-even oscillation of the electronic structure with width is revealed for H-functionalized borophene nanoribbons, while an ultra-high work function (~ 7.83 eV) is found for the F-functionalized borophene due to its strong charge transfer to the atomic adsorbates.

cond-mat.mtrl-sci↗

arXiv2017

Atomic-scale mechanisms of defect- and light-induced oxidation and degradation of InSe

Layered indium selenide (InSe), a new two-dimensional (2D) material with a hexagonal structure and semiconducting characteristic, is gaining increasing attention owing to its intriguing electronic properties. Here, by using first-principles calculations, we reveal that perfect InSe possesses a high chemical stability against oxidation, superior to MoS2. However, the presence of intrinsic Se vacancy (VSe) and light illumination can markedly affect the surface activity. In particular, the excess electrons associated with the exposed In atoms at the VSe site under illumination are able to remarkably reduce the dissociation barrier of O2 to ~0.2 eV. Moreover, at ambient conditions, the splitting of O2 enables the formation of substitutional (apical) oxygen atomic species, which further cause the trapping and subsequent rapid splitting of H2O molecules and ultimately the formation of hydroxyl groups. Our findings uncover the causes and underlying mechanisms of InSe surface degradation via the defect-photo promoted oxidations. Such results will be beneficial in developing strategies for the storage of InSe material and its applications for surface passivation with boron nitride, graphene or In-based oxide layers.

cond-mat.mtrl-sci↗

arXiv2017

Intrinsic Light Field Images

We present a method to automatically decompose a light field into its intrinsic shading and albedo components. Contrary to previous work targeted to 2D single images and videos, a light field is a 4D structure that captures non-integrated incoming radiance over a discrete angular domain. This higher dimensionality of the problem renders previous state-of-the-art algorithms impractical either due to their cost of processing a single 2D slice, or their inability to enforce proper coherence in additional dimensions. We propose a new decomposition algorithm that jointly optimizes the whole light field data for proper angular coherence. For efficiency, we extend Retinex theory, working on the gradient domain, where new albedo and occlusion terms are introduced. Results show our method provides 4D intrinsic decompositions difficult to achieve with previous state-of-the-art algorithms. We further provide a comprehensive analysis and comparisons with existing intrinsic image/video decomposition methods on light field images.

cs.CV↗

arXiv2017

Strain and Water Effects on the Electronic Structure and Chemical Activity of In-Plane Graphene/Silicene Heterostructure

By using first-principles calculations, the electronic structure of planar and strained in-plane graphene/silicene heterostructure is studied. The heterostructure is found to be metallic in a strain range from -7% (compression) to +7% (tension). The effect of compressive/tensile strain on the chemical activity of the in-plane graphene/silicene heterostructure is examined by studying its interaction with the H2O molecule. It shows that compressive/tensile strain is able to increase the binding energy of H2O compared with the adsorption on a planar surface, and the charge transfer between the water molecule and the graphene/silicene sheet can be modulated by strain. Moreover, the presence of the boron-nitride (BN)-substrate significantly influences the chemical activity of the graphene/silicene heterostructure upon its interaction with the H2O molecule and may cause an increase/decrease of the charge transfer between the H2O molecule and the heterostructure. These findings provide insights into the modulation of electronic properties of the in-plane free-standing/substrate-supported graphene/silicene heterostructure, and render possible ways to control its electronic structure, carrier density and redox characteristics, which may be useful for its potential applications in nanoelectronics and gas sensors.

cond-mat.mtrl-sci↗

arXiv2016

Resonant interaction of $ϕ^4$ kink with spatially periodic $\mathcal{PT}$-symmetric perturbation

The resonant interaction of the $ϕ^4$ kink with a periodic $\mathcal{PT}$-symmetric perturbation is observed in the frame of the continuum model and with the help of a two degree of freedom collective variable model derived in PRA 89, 010102(R). When the kink interacts with the perturbation, the kink's internal mode is excited with the amplitude varying in time quasiperiodically. The maximal value of the amplitude was found to grow when the kink velocity is such that it travels one period of perturbation is nearly one period of the kink's internal mode. It is also found that the kink's translational and vibrational modes are coupled in a way that an increase in the kink's internal mode amplitude results in a decrease in kink velocity. The results obtained with the collective variable method are in a good qualitative agreement with the numerical simulations for the continuum model. The results of the present study suggest that kink dynamics in open systems with balanced gain and loss can have new features in comparison with the case of conservative systems.

nlin.PS↗

arXiv2016

Large Electronic Anisotropy and Enhanced Chemical Activity of Highly Rippled Phosphorene

We investigate the electronic structure and chemical activity of rippled phosphorene induced by large compressive strains via first-principles calculation. It is found that phosphorene is extraordinarily bendable, enabling the accommodation of ripples with large curvatures. Such highly rippled phosphorene shows a strong anisotropy in electronic properties. For ripples along the armchair direction, the band gap changes from 0.84 to 0.51 eV for the compressive strain up to -20% and further compression shows no significant effect, for ripples along the zigzag direction, semiconductor to metal transition occurs. Within the rippled phosphorene, the local electronic properties, such as the modulated band gap and the alignments of frontier orbitals, are found to be highly spatially dependent, which may be used for modulating the injection and confinement of carriers for optical and photovoltaic applications. The examination of the interaction of a physisorbed NO molecule with the rippled phosphorene under different compressive strains shows that the chemical activities of the phosphorene are significantly enhanced at the top and bottom peaks of the ripples, indicated by the enhanced adsorption and charge transfer between them. All these features can be ascribed to the effect of curvatures, which modifies the orbital coupling between atoms at the ripple peaks.

cond-mat.mtrl-sci↗

arXiv2016

The Role of H2O and O2 Molecules and Phosphorus Vacancies in the Structure Instability of Phosphorene

The poor structural stability of phosphorene in air was commonly ascribed to humidity and oxygen molecules. Recent exfoliation of phosphorene in deoxygenated water promotes the need to re-examine the role of H2O and O2 molecules. Considering the presence of high population of vacancies in phosphorene, we investigate the interaction of H2O and O2 molecules with vacancy-contained phosphorene using first-principles calculations. In contrast to the common notion that physisorbed molecules tend to have a stronger adsorption at vacancy sites, we show that H2O has nearly the same adsorption energy at the vacancy site as that at the perfect one. Charge transfer analysis shows that O2 is a strong electron scavenger, which transfers the lone-pair electrons of the phosphorus atoms to the 2π* antibonding orbital of O2. As a result, the barrier for the O-O bond splitting to form O-P bonds is reduced from 0.81 eV at the perfect site to 0.59 eV at the defect site, leading to an about 5000 faster oxidizing rate at the defect site than at the perfect site at room temperature. Hence, our work reveals that the vacancy in phosphorene shows a stronger oxygen affinity than the perfect phosphorene lattice site. Structural degradation of phosphorene due to oxidization may occur rapidly at edges and grain boundaries where vacancies tend to agglomerate.

cond-mat.mtrl-sci↗

arXiv2016

High energy density in the collision of $N$ kinks in the $ϕ^4$ model

Recently for the sine-Gordon equation it has been established that during collisions of $N$ slow kinks maximal energy density increases as $N^2$. In this numerical study, the same scaling rule is established for the non-integrable $ϕ^4$ model for $N\le 5$. For odd (even) $N$ the maximal energy density is in the form of potential (kinetic) energy density. Maximal elastic strain is also calculated. In addition, the effect of the kink's internal modes on the maximal energy density is analysed for $N=1, 2, 3$. Our results suggest that in multi-soliton collisions very high energy density can be achieved in a controllable manner.

nlin.PS↗

arXiv2016

Segmentation Rectification for Video Cutout via One-Class Structured Learning

Recent works on interactive video object cutout mainly focus on designing dynamic foreground-background (FB) classifiers for segmentation propagation. However, the research on optimally removing errors from the FB classification is sparse, and the errors often accumulate rapidly, causing significant errors in the propagated frames. In this work, we take the initial steps to addressing this problem, and we call this new task \emph{segmentation rectification}. Our key observation is that the possibly asymmetrically distributed false positive and false negative errors were handled equally in the conventional methods. We, alternatively, propose to optimally remove these two types of errors. To this effect, we propose a novel bilayer Markov Random Field (MRF) model for this new task. We also adopt the well-established structured learning framework to learn the optimal model from data. Additionally, we propose a novel one-class structured SVM (OSSVM) which greatly speeds up the structured learning process. Our method naturally extends to RGB-D videos as well. Comprehensive experiments on both RGB and RGB-D data demonstrate that our simple and effective method significantly outperforms the segmentation propagation methods adopted in the state-of-the-art video cutout systems, and the results also suggest the potential usefulness of our method in image cutout system.

cs.CV↗

arXiv2013

Superior thermal conductivity and extremely high mechanical strength in polyethylene chains from {\it ab initio} calculation

The upper limit of the thermal conductivity and the mechanical strength are predicted for the polyethylene chain, by performing the {\it ab initio} calculation and applying the quantum mechanical non-equilibrium Green's function approach. Specially, there are two main findings from our calculation: (1). the thermal conductivity can reach a high value of 310 W/K/m in a 100 nm polyethylene chain at room temperature; (2). the Young's modulus in the polyethylene chain is as high as 374.5 GPa, and the polyethylene chain can sustain $32.85% \pm 0.05%$ (ultimate) strain before undergoing structural phase transition into gaseous ethylene.

cond-mat.mtrl-sci↗