arXiv subjects

Lei Xu

Publications and source records attributed to Lei Xu.

At least 253 records · Page 14Linked to original sources

arXiv2017

Large valley splitting in monolayer WS$_2$ by proximity coupling to an insulating antiferromagnetic substrate

Lifting the valley degeneracy is an efficient way to achieve valley polarization for further valleytronics operations. In this work, we demonstrate that a large valley splitting can be obtained in monolayer transition metal dichalcogenides by magnetic proximity coupling to an insulating antiferromagnetic substrate. As an example, we perform first-principles calculations to investigate the electronic structures of monolayer WS$_2$ on the MnO(111) surface. Our calculation results suggest that a large valley splitting of 214 meV, which corresponds to a Zeeman magnetic field of 1516 T, is induced in the valence band of monolayer WS$_2$. The magnitude of valley splitting relies on the strength of interfacial orbital hybridization, and can be continually tuned by applying an external out-of-plane pressure and in-plane strain. More interestingly, we find that both spin and valley index will flip when the magnetic ordering of MnO is reversed. Besides, owing to the sizeable Berry curvature and time-reversal symmetry breaking in the WS$_2$/MnO heterostructure, a spin and valley polarized anomalous Hall current can be generated in the presence of an in-plane electric field, which allow one to detect valleys by the electrical approach. Our results shed light on the realization of valleytronic devices using the antiferromagnetic insulator as the substrate.

cond-mat.mtrl-sci↗

arXiv2017

Sketch-pix2seq: a Model to Generate Sketches of Multiple Categories

Sketch is an important media for human to communicate ideas, which reflects the superiority of human intelligence. Studies on sketch can be roughly summarized into recognition and generation. Existing models on image recognition failed to obtain satisfying performance on sketch classification. But for sketch generation, a recent study proposed a sequence-to-sequence variational-auto-encoder (VAE) model called sketch-rnn which was able to generate sketches based on human inputs. The model achieved amazing results when asked to learn one category of object, such as an animal or a vehicle. However, the performance dropped when multiple categories were fed into the model. Here, we proposed a model called sketch-pix2seq which could learn and draw multiple categories of sketches. Two modifications were made to improve the sketch-rnn model: one is to replace the bidirectional recurrent neural network (BRNN) encoder with a convolutional neural network(CNN); the other is to remove the Kullback-Leibler divergence from the objective function of VAE. Experimental results showed that models with CNN encoders outperformed those with RNN encoders in generating human-style sketches. Visualization of the latent space illustrated that the removal of KL-divergence made the encoder learn a posterior of latent space that reflected the features of different categories. Moreover, the combination of CNN encoder and removal of KL-divergence, i.e., the sketch-pix2seq model, had better performance in learning and generating sketches of multiple categories and showed promising results in creativity tasks.

cs.CV↗

arXiv2017

A Comparative Study of Joint-SNVs Analysis Methods and Detection of Susceptibility Genes for Gastric Cancer in Korean Population

Many joint-SNVs (single-nucleotide variants) analysis methods were proposed to tackle the "missing heritability" problem, which emphasizes that the joint genetic variants can explain more heritability of traits and diseases. However, there is still lack of a systematic comparison and investigation on the relative strengths and weaknesses of these methods. In this paper, we evaluated their performance on extensive simulated data generated by varying sample size, linkage disequilibrium (LD), odds ratios (OR), and minor allele frequency (MAF), which aims to cover almost all scenarios encountered in practical applications. Results indicated that a method called Statistics-space Boundary Based Test (S-space BBT) showed stronger detection power than other methods. Results on a real dataset of gastric cancer for Korean population also validate the effectiveness of the S-space BBT method.

stat.ME↗

arXiv2017

Artificial Light Harvesting by Dimerized Mobius Ring

We theoretically study artificial light harvesting by a dimerized Mobius ring. When the donors in the ring are dimerized, the energies of the donor ring are splitted into two sub-bands. Because of the nontrivial Mobius boundary condition, both the photon and acceptor are coupled to all collectiveexcitation modes in the donor ring. Therefore, the quantum dynamics in the light harvesting are subtly influenced by the dimerization in the Mobius ring. It is discovered that energy transfer is more efficient in a dimerized ring than that in an equally-spaced ring. This discovery is also confirmed by the calculation with the perturbation theory, which is equivalent to the Wigner-Weisskopf approximation. Our findings may be benificial to the optimal design of artificial light harvesting.

quant-ph↗

arXiv2017

Input Convex Neural Networks

This paper presents the input convex neural network architecture. These are scalar-valued (potentially deep) neural networks with constraints on the network parameters such that the output of the network is a convex function of (some of) the inputs. The networks allow for efficient inference via optimization over some inputs to the network given others, and can be applied to settings including structured prediction, data imputation, reinforcement learning, and others. In this paper we lay the basic groundwork for these models, proposing methods for inference, optimization and learning, and analyze their representational power. We show that many existing neural network architectures can be made input-convex with a minor modification, and develop specialized optimization algorithms tailored to this setting. Finally, we highlight the performance of the methods on multi-label prediction, image completion, and reinforcement learning problems, where we show improvement over the existing state of the art in many cases.

cs.LG↗

arXiv2017

Direct detection of pixel-level myocardial infarction areas via a deep-learning algorithm

Accurate detection of the myocardial infarction (MI) area is crucial for early diagnosis planning and follow-up management. In this study, we propose an end-to-end deep-learning algorithm framework (OF-RNN ) to accurately detect the MI area at the pixel level. Our OF-RNN consists of three different function layers: the heart localization layers, which can accurately and automatically crop the region-of-interest (ROI) sequences, including the left ventricle, using the whole cardiac magnetic resonance image sequences; the motion statistical layers, which are used to build a time-series architecture to capture two types of motion features (at the pixel-level) by integrating the local motion features generated by long short-term memory-recurrent neural networks and the global motion features generated by deep optical flows from the whole ROI sequence, which can effectively characterize myocardial physiologic function; and the fully connected discriminate layers, which use stacked auto-encoders to further learn these features, and they use a softmax classifier to build the correspondences from the motion features to the tissue identities (infarction or not) for each pixel. Through the seamless connection of each layer, our OF-RNN can obtain the area, position, and shape of the MI for each patient. Our proposed framework yielded an overall classification accuracy of 94.35% at the pixel level, from 114 clinical subjects. These results indicate the potential of our proposed method in aiding standardized MI assessments.

cs.CV↗

arXiv2017

Mechanism of Contact between a Droplet and an Atomically Smooth Substrate

When a droplet gently lands on an atomically smooth substrate, it will most likely contact the underlying surface in about 0.1 s. However, theoretical estimation from fluid mechanics predicts a contact time of 10 to 100 s. What causes this large discrepancy, and how does nature speed up contact by 2 orders of magnitude? To probe this fundamental question, we prepare atomically smooth substrates by either coating a liquid film on glass or using a freshly cleaved mica surface, and visualize the droplet contact dynamics with 30 nm resolution. Interestingly, we discover two distinct speed-up approaches: (1) droplet skidding due to even minute perturbations breaks rotational symmetry and produces early contact at the thinnest gap location, and (2) for the unperturbed situation with rotational symmetry, a previously unnoticed boundary flow around only 0.1 mm/s expedites air drainage by over 1 order of magnitude. Together, these two mechanisms universally explain general contact phenomena on smooth substrates. The fundamental discoveries shed new light on contact and drainage research.

physics.flu-dyn↗

arXiv2016

Covalency and vibronic couplings make a nonmagnetic j=3/2 ion magnetic

For 4$d^1$ and 5$d^1$ spin-orbit-coupled electron configurations, the notion of nonmagnetic j=3/2 quartet ground state discussed in classical textbooks is at odds with the observed variety of magnetic properties. Here we throw fresh light on the electronic structure of 4$d^1$ and 5$d^1$ ions in molybdenum- and osmium-based double-perovskite systems and reveal different kinds of on-site many-body physics in the two families of compounds: while the sizable magnetic moments and $g$ factors measured experimentally are due to both metal $d$-ligand $p$ hybridization and dynamic Jahn-Teller interactions for 4$d$ electrons, it is essentially $d$-$p$ covalency for the 5$d^1$ configuration. These results highlight the subtle interplay of spin-orbit interactions, covalency and electron-lattice couplings as the major factor in deciding the nature of the magnetic ground states of 4$d$ and 5$d$ quantum materials. Cation charge imbalance in the double-perovskite structure is further shown to allow a fine tuning of the gap between the $t_{2g}$ and $e_g$ levels, an effect of much potential in the context of orbital engineering in oxide electronics.

cond-mat.str-el↗

arXiv2016

Topic Sensitive Neural Headline Generation

Neural models have recently been used in text summarization including headline generation. The model can be trained using a set of document-headline pairs. However, the model does not explicitly consider topical similarities and differences of documents. We suggest to categorizing documents into various topics so that documents within the same topic are similar in content and share similar summarization patterns. Taking advantage of topic information of documents, we propose topic sensitive neural headline generation model. Our model can generate more accurate summaries guided by document topics. We test our model on LCSTS dataset, and experiments show that our method outperforms other baselines on each topic and achieves the state-of-art performance.

cs.CL↗

arXiv2016

Probing the role of mobility in the collective motion of non-equilibrium systems

By systematically varying the mobility of self-propelled particles in a two-dimensional (2D) lattice, we experimentally study the influence of particle mobility on system's collective motion. Our system is intrinsically non-equilibrium due to the lack of energy equipartition. By constructing the covariance matrix of spatial fluctuations and solving for its eigenmodes, we obtain the collective motions of the system with various magnitudes. Interestingly, our structurally ordered non-equilibrium system exhibits almost identical properties as disordered glassy systems under thermal equilibrium: the modes with large overall motions are spatially correlated and quasilocalized while the modes with small collective motions are highly localized, resembling the low- and high-frequency modes in glass. More surprisingly, a peak similar to the boson peak forms in our non-equilibrium system as the number of mobile particles increases, revealing the possible origin of the boson peak from a dynamic aspect. We further illustrate that the spatially-correlated large-movement modes can be produced by the cooperation of highly-active particles above a threshold fraction, while the localized small-movement modes can be created by adding individual inactive particles. Our study clarifies the role of mobility in collective motions, and further suggests a promising possibility of extending the powerful mode analysis approach to non-equilibrium systems.

cond-mat.soft↗

arXiv2016

Feedback from Nature: Simple Randomised Distributed Algorithms for Maximal Independent Set Selection and Greedy Colouring

We propose distributed algorithms for two well-established problems that operate efficiently under extremely harsh conditions. Our algorithms achieve state-of-the-art performance in a simple and novel way. Our algorithm for maximal independent set selection operates on a network of identical anonymous processors. The processor at each node has no prior information about the network. At each time step, each node can only broadcast a single bit to all its neighbours, or remain silent. Each node can detect whether one or more neighbours have broadcast, but cannot tell how many of its neighbours have broadcast, or which ones. We build on recent work of Afek et al. which was inspired by studying the development of a network of cells in the fruit fly~\cite{Afek2011a}. However we incorporate for the first time another important feature of the biological system: varying the probability value used at each node based on local feedback from neighbouring nodes. Given any $n$-node network, our algorithm achieves the optimal expected time complexity of $O(\log n)$ rounds and the optimal expected message complexity of $O(1)$ single-bit messages broadcast by each node.We also show that the previous approach, without feedback, cannot achieve better than $Ω(\log^2 n)$ expected time complexity, whatever global scheme is used to choose the probabilities. Our algorithm for distributed greedy colouring works under similar harsh conditions: each identical node has no prior information about the network, can only broadcast a single message to all neighbours at each time step representing a desired colour, and can only detect whether at least one neighbour has broadcast each colour value. We show that our algorithm has an expected time complexity of $O(Δ+\log n)$, where $Δ$ is the maximum degree of the network, and expected message complexity of $O(1)$ messages broadcast by each node.

cs.DC↗

arXiv2015

Kelvin Helmholtz instability in an ultrathin air film causes drop splashing on smooth surfaces

When a fast-moving drop impacts onto a smooth substrate, splashing will be produced at the edge of the expanding liquid sheet. This ubiquitous phenomenon lacks a fundamental understanding. Combining experiment with model, we illustrate that the ultrathin air film trapped under the expanding liquid front triggers splashing. Because this film is thinner than the mean free path of air molecules, the interior airflow transfers momentum with an unusually high velocity comparable to the speed of sound and generates a stress 10 times stronger than the airflow in common situations. Such a large stress initiates Kelvin Helmholtz instabilities at small length scales and effectively produces splashing. Our model agrees quantitatively with experimental verifications and brings a fundamental understanding to the ubiquitous phenomenon of drop splashing on smooth surfaces.

physics.flu-dyn↗

arXiv2015

A Herschel Study of 24 micron-Selected AGNs and Their Host Galaxies

We present a sample of 290 24-micron-selected active galactic nuclei (AGNs) mostly at z ~ 0.3 -- 2.5, within 5.2 square degrees distributed as 25' X 25' fields around each of 30 galaxy clusters in the Local Cluster Substructure Survey (LoCuSS). The sample is nearly complete to 1 mJy at 24 microns, and has a rich multi-wavelength set of ancillary data; 162 are detected by Herschel. We use spectral templates for AGNs, stellar populations, and infrared emission by star forming galaxies to decompose the spectral energy distributions (SEDs) of these AGNs and their host galaxies, and estimate their star formation rates (SFRs), AGN luminosities, and host galaxy stellar masses. The set of templates is relatively simple: a standard Type-1 quasar template; another for the photospheric output of the stellar population; and a far infrared star-forming template. For the Type-2 AGN SEDs, we substitute templates including internal obscuration, and some Type-1 objects require a warm component (T > 50 K). The individually Herschel- detected Type-1 AGNs and a subset of 17 Type-2 ones typically have luminosities > 10^{45} ergs/s, and supermassive black holes of ~ 3 X 10^8 Msun emitting at ~ 10% of the Eddington rate. We find them in about twice the numbers of AGN identified in SDSS data in the same fields, i.e., they represent typical high luminosity AGN, not an infrared-selected minority. These AGNs and their host galaxies are studied further in an accompanying paper.

astro-ph.GA↗

arXiv2015

The Relation between Luminous AGNs and Star Formation in Their Host Galaxies

We study the relation of active galactic nuclei (AGNs) to star formation in their host galaxies. Our sample includes 205 Type-1 and 85 Type-2 AGNs, 162 detected with Herschel, from fields surrounding 30 galaxy clusters in the Local Cluster Substructure Survey (LoCuSS). The sample is identified by optical line widths and ratios after selection to be brighter than 1 mJy at 24 microns. We show that Type-2 AGN [OIII]5007 line fluxes at high z can be contaminated by their host galaxies with typical spectrograph entrance apertures (but our sample is not compromised in this way). We use spectral energy distribution (SED) templates to decompose the galaxy SEDs and estimate star formation rates, AGN luminosities, and host galaxy stellar masses (described in an accompanying paper). The AGNs arise from massive black holes (~ 3 X 10^8 Msun) accreting at ~ 10% of the Eddington rate and residing in galaxies with stellar mass > 3 X 10^{10} Msun; those detected with Herschel have IR luminosity from star formation in the range of 10^{10} -- 10^{12} Lsun. We find that: 1.) the specific star formation rates in the host galaxies are generally consistent with those of normal star-forming (main sequence) galaxies; 2.) there is a strong correlation between the luminosities from star formation and the AGN; and 3.) however, the correlation may not result from a causal connection, but could arise because the black hole mass (and hence AGN Eddington luminosity) and star formation are both correlated with the galaxy mass.

astro-ph.GA↗

arXiv2015

Effect of Exchange Interaction on Magnetic Thermal Fluctuation and Spin Susceptibility

The expression of the thermal fluctuation parameter in the stochastic Landau-Lifshitz-Gilbert equation has been derived from a fundamental quantum theory of spins and phonons, in which the exchange interaction between nearest atoms has been included. Our studies show that the thermal fluctuation decreases exponentially with increasing exchange interaction. The non-uniform fluctuation of local spins make the spin susceptibility much different from the result derived by the macro-spin model or single spin model. The related spin susceptibility depends not only on the strength of exchange interaction, but also on the lattice structure. The non-uniform fluctuation can lead to an extra broadening of the resonance line width along with the broadening arisen from the Gilbert damping.

cond-mat.mes-hall↗

arXiv2014

Visualizing kinetic pathways of homogeneous nucleation in colloidal crystallization

When a system undergoes a transition from a liquid to a solid phase, it passes through multiple intermediate structures before reaching the final state. However, our knowledge on the exact pathways of this process is limited, mainly due to the difficulty of realizing direct observations. Here, we experimentally study the evolution of symmetry and density for various colloidal systems during liquid-to-solid phase transitions, and visualize kinetic pathways with single-particle resolution. We observe the formation of relatively-ordered precursor structures with different symmetries, which then convert into metastable solids. During this conversion, two major cross-symmetry pathways always occur, regardless of the final state and the interaction potential. In addition, we find a broad decoupling of density variation and symmetry development, and discover that nucleation rarely starts from the densest regions. These findings hold for all our samples, suggesting the possibility of finding a unified picture for the complex crystallization kinetics in colloidal systems.

cond-mat.soft↗

arXiv2013

Structuring Causal Tree Models with Continuous Variables

This paper considers the problem of invoking auxiliary, unobservable variables to facilitate the structuring of causal tree models for a given set of continuous variables. Paralleling the treatment of bi-valued variables in [Pearl 1986], we show that if a collection of coupled variables are governed by a joint normal distribution and a tree-structured representation exists, then both the topology and all internal relationships of the tree can be uncovered by observing pairwise dependencies among the observed variables (i.e., the leaves of the tree). Furthermore, the conditions for normally distributed variables are less restrictive than those governing bi-valued variables. The result extends the applications of causal tree models which were found useful in evidential reasoning tasks.

cs.AI↗

arXiv2013

Eliminating cracking during drying

When colloidal suspensions dry, stresses build up and cracks often occur - a phenomenon undesirable for important industries such as paint and ceramics. We demonstrate an effective method which can completely eliminate cracking during drying: by adding emulsion droplets into colloidal suspensions, we can systematically decrease the amount of cracking, and eliminate it completely above a critical droplet concentration. Since the emulsion droplets eventually also evaporate, our technique achieves an effective function while making little changes to the component of final product, and may therefore serve as a promising approach for cracking elimination. Furthermore, adding droplets also varies the speed of air invasion and provides a powerful method to adjust drying rate. With the effective control over cracking and drying rate, our study may find important applications in many drying and cracking related industrial processes.

physics.flu-dyn↗