arXiv subjects

Gang Chen

Publications and source records attributed to Gang Chen.

At least 649 records · Page 36Linked to original sources

arXiv2015

Generalized K-fan Multimodal Deep Model with Shared Representations

Multimodal learning with deep Boltzmann machines (DBMs) is an generative approach to fuse multimodal inputs, and can learn the shared representation via Contrastive Divergence (CD) for classification and information retrieval tasks. However, it is a 2-fan DBM model, and cannot effectively handle multiple prediction tasks. Moreover, this model cannot recover the hidden representations well by sampling from the conditional distribution when more than one modalities are missing. In this paper, we propose a K-fan deep structure model, which can handle the multi-input and muti-output learning problems effectively. In particular, the deep structure has K-branch for different inputs where each branch can be composed of a multi-layer deep model, and a shared representation is learned in an discriminative manner to tackle multimodal tasks. Given the deep structure, we propose two objective functions to handle two multi-input and multi-output tasks: joint visual restoration and labeling, and the multi-view multi-calss object recognition tasks. To estimate the model parameters, we initialize the deep model parameters with CD to maximize the joint distribution, and then we use backpropagation to update the model according to specific objective function. The experimental results demonstrate that the model can effectively leverages multi-source information and predict multiple tasks well over competitive baselines.

cs.LG↗

arXiv2015

Ab initio study of electron-phonon interaction in phosphorene

The monolayer of black phosphorous, or phosphorene, has recently emerged as a new 2D semiconductor with intriguing highly anisotropic transport properties. Existing calculations of its intrinsic phonon-limited electronic transport properties so far rely on the deformation potential approximation, which is in general not directly applicable to anisotropic materials since the deformation along one specific direction can scatter electrons traveling in all directions. We perform a first-principles calculation of the electron-phonon interaction in phosphorene based on density functional perturbation theory and Wannier interpolation. Our calculation reveals that 1) the high anisotropy provides extra phase space for electron-phonon scattering, and 2) optical phonons have appreciable contributions. Both effects cannot be captured by the deformation potential calculations.

cond-mat.mtrl-sci↗

arXiv2015

DGCC:A New Dependency Graph based Concurrency Control Protocol for Multicore Database Systems

Multicore CPUs and large memories are increasingly becoming the norm in modern computer systems. However, current database management systems (DBMSs) are generally ineffective in exploiting the parallelism of such systems. In particular, contention can lead to a dramatic fall in performance. In this paper, we propose a new concurrency control protocol called DGCC (Dependency Graph based Concurrency Control) that separates concurrency control from execution. DGCC builds dependency graphs for batched transactions before executing them. Using these graphs, contentions within the same batch of transactions are resolved before execution. As a result, the execution of the transactions does not need to deal with contention while maintaining full equivalence to that of serialized execution. This better exploits multicore hardware and achieves higher level of parallelism. To facilitate DGCC, we have also proposed a system architecture that does not have certain centralized control components yielding better scalability, as well as supports a more efficient recovery mechanism. Our extensive experimental study shows that DGCC achieves up to four times higher throughput compared to that of state-of-the-art concurrency control protocols for high contention workloads.

cs.DB↗

arXiv2015

Local field topology behind light localization and metamaterial topological transitions

We revisit the mechanisms governing the sub-wavelength spatial localization of light in surface plasmon polariton (SPP) modes by investigating both local and global features in optical powerflow at SPP frequencies. Close inspection of the instantaneous Poynting vector reveals formation of optical vortices - localized areas of cyclic powerflow - at the metal-dielectric interface. As a result, optical energy circulates through a subwavelength-thick 'conveyor belt' between the metal and dielectric where it creates a high density of optical states (DOS), tight optical energy localization, and low group velocity associated with SPP waves. The formation of bonding and anti-bonding SPP modes in metal-dielectric-metal waveguides can also be conveniently explained in terms of different spatial arrangements of localized powerflow vortices between two metal interfaces. Finally, we investigate the underlying mechanisms of global topological transitions in metamaterials composed of multiple metal and dielectric films, i.e., transitions of their iso-frequency surfaces from ellipsoids to hyperboloids, which are not accompanied by the breaking of lattice symmetry. Our analysis reveals that such global topological transitions are governed by the dynamic local re-arrangement of local topological features of the optical interference field, such as vortices and saddle points, which reconfigures global optical powerflow within the metamaterial. These new insights into plasmonic light localization and DOS manipulation not only help to explain the well-known properties of SPP waves but also provide useful guidelines for the design of plasmonic components and materials for a variety of practical applications.

physics.optics↗

arXiv2015

Note on off-shell relations in nonlinear sigma model

In this note, we investigate relations between tree-level off-shell currents in nonlinear sigma model. Under Cayley parametrization, all odd-point currents vanish. We propose and prove a generalized $U(1)$ identity for even-point currents. The off-shell $U(1)$ identity given in [1] is a special case of the generalized identity studied in this note. The on-shell limit of this identity is equivalent with the on-shell KK relation. Thus this relation provides the full off-shell correspondence of tree-level KK relation in nonlinear sigma model.

hep-th↗

arXiv2015

Thin-film 'Thermal Well' Emitters and Absorbers for High-Efficiency Thermophotovoltaics

A new approach is introduced to significantly improve the performance of thermophotovoltaic (TPV) systems by using low-dimensional thermal emitters and photovoltaic (PV) cells. By reducing the thickness of both the emitter and the PV cell, strong spectral selectivity in both thermal emission and absorption can be achieved by confining photons in trapped waveguide modes inside the thin-films that act as thermal analogs to quantum wells. Simultaneously, photo-excited carriers travel shorter distances across the thin-films reducing bulk recombination losses resulting in a lower saturation current in the PV cell. We predict a TPV efficiency enhancement with near-field coupling between the thermal emitter and the PV cell of up to 38.7% using a germanium (Ge) emitter at 1000 K and a gallium antimonide (GaSb) cell with optimized thicknesses separated by 100 nm. Even in the far-field limit, the efficiency is predicted to reach 31.5%, which is an order of magnitude higher than the Shockley Queisser limit of 1.6% for a bulk GaSb cell and a blackbody emitter at 1000 K. The proposed design approach does not require nanoscale patterning of the emitter and PV cell surfaces, but instead offers a simple low-cost solution to improve the performance of a thermophotovoltaic system.

physics.optics↗

arXiv2015

First-Principles Simulation of Electron Mean-Free-Path Spectra and Thermoelectric Properties in Silicon

The mean-free-paths (MFPs) of energy carriers are of critical importance to the nano-engineering of better thermoelectric materials. Despite significant progress in the first-principles-based understanding of the spectral distribution of phonon MFPs in recent years, the spectral distribution of electron MFPs remains unclear. In this work, we compute the energy dependent electron scatterings and MFPs in silicon from first-principles. The electrical conductivity accumulation with respect to electron MFPs is compared to that of the phonon thermal conductivity accumulation to illustrate the quantitative impact of nanostructuring on electron and phonon transport. By combining all electron and phonon transport properties from first-principles, we predict the thermoelectric properties of the bulk and nanostructured silicon, and find that silicon with 20 nm nanograins can result in more than five times enhancement in their thermoelectric figure of merit as the grain boundaries scatter phonons more significantly than that of electrons due to their disparate MFP distributions.

cond-mat.mtrl-sci↗

arXiv2015

Two-axis spin squeezing in two cavities

Ultracold atoms in an ultrahigh-finesse optical cavity are a powerful platform to produce spin squeezing since photon of cavity mode can induce nonlinear spin-spin interaction and thus generate a one-axis twisting Hamiltonian $H_{\text{OAT}}=qJ_{x}^{2}$, whose corresponding maximal squeezing factor scales as $N^{-2/3}$, where $N$ is the atomic number. On the contrary, for the other two-axis twisting Hamiltonian $H_{\text{TAT}}=q(J_{x}^{2}-J_{y}^{2})$, the maximal squeezing factor scales as $N^{-1}$, approaching the Heisenberg limit. In this paper, inspired by recent experiments of cavity-assisted Raman transitions, we propose a scheme, in which an ensemble of ultracold six-level atoms interacts with two quantized cavity fields and two pairs of Raman lasers, to realize a tunable two-axis spin Hamiltonian $%H=q(J_{x}^{2}+χJ_{y}^{2})+ω_{0}J_{z}$. For proper parameters, the above one- and two- axis twisting Hamiltonians are recovered, and the scaling of $N^{-1}$ of the maximal squeezing factor can occur naturally. On the other hand, in the two-axis twisting Hamiltonian, spin squeezing is usually reduced when increasing the effective atomic resonant frequency $ω_{0}$. Surprisingly, we find that by combined with the dimensionless parameter $χ(>-1)$, the effective atomic resonant frequency $ω_{0}$ can enhance spin squeezing largely. These results are benefit for achieving the required spin squeezing in experiments.

quant-ph↗

arXiv2015

Deep Transductive Semi-supervised Maximum Margin Clustering

Semi-supervised clustering is an very important topic in machine learning and computer vision. The key challenge of this problem is how to learn a metric, such that the instances sharing the same label are more likely close to each other on the embedded space. However, little attention has been paid to learn better representations when the data lie on non-linear manifold. Fortunately, deep learning has led to great success on feature learning recently. Inspired by the advances of deep learning, we propose a deep transductive semi-supervised maximum margin clustering approach. More specifically, given pairwise constraints, we exploit both labeled and unlabeled data to learn a non-linear mapping under maximum margin framework for clustering analysis. Thus, our model unifies transductive learning, feature learning and maximum margin techniques in the semi-supervised clustering framework. We pretrain the deep network structure with restricted Boltzmann machines (RBMs) layer by layer greedily, and optimize our objective function with gradient descent. By checking the most violated constraints, our approach updates the model parameters through error backpropagation, in which deep features are learned automatically. The experimental results shows that our model is significantly better than the state of the art on semi-supervised clustering.

cs.LG↗

arXiv2015

Synthetic gauge fields stabilize a chiral spin liquid phase

We calculate the phase diagram of the SU($N$) Hubbard model describing fermionic alkaline earth atoms in a square optical lattice with on-average one atom per site, using a slave-rotor mean-field approximation. We find that the chiral spin liquid predicted for $N\ge5$ and large interactions passes through a fractionalized state with a spinon Fermi surface as interactions are decreased before transitioning to a weakly interacting metal. We also show that by adding an artificial uniform magnetic field with flux per plaquette $2π/N$, the chiral spin liquid becomes the ground state for all $N\ge 3$ at large interactions, persists to weaker interactions, and its spin gap increases, suggesting that the spin liquid physics will persist to higher temperatures. We discuss potential methods to realize the artificial gauge fields and detect the predicted phases.

cond-mat.quant-gas↗

arXiv2015

Deep Learning with Nonparametric Clustering

Clustering is an essential problem in machine learning and data mining. One vital factor that impacts clustering performance is how to learn or design the data representation (or features). Fortunately, recent advances in deep learning can learn unsupervised features effectively, and have yielded state of the art performance in many classification problems, such as character recognition, object recognition and document categorization. However, little attention has been paid to the potential of deep learning for unsupervised clustering problems. In this paper, we propose a deep belief network with nonparametric clustering. As an unsupervised method, our model first leverages the advantages of deep learning for feature representation and dimension reduction. Then, it performs nonparametric clustering under a maximum margin framework -- a discriminative clustering model and can be trained online efficiently in the code space. Lastly model parameters are refined in the deep belief network. Thus, this model can learn features for clustering and infer model complexity in an unified framework. The experimental results show the advantage of our approach over competitive baselines.

cs.LG↗

arXiv2014

Fractionalized charge excitations in a spin liquid on partially-filled pyrochlore lattice

We study the Mott transition from a metal to cluster Mott insulators in the 1/4- and 1/8-filled pyrochlore lattice systems. It is shown that such Mott transitions can arise due to charge localization in clusters or in tetrahedron units, driven by the nearest-neighbor repulsive interaction. The resulting cluster Mott insulator is a quantum spin liquid with a spinon Fermi surface, but at the same time a novel fractionalized charge liquid with charge excitations carrying half the electron charge. There exist two emergent U(1) gauge fields or "photons" that mediate interactions between spinons and charge excitations, and between fractionalized charge excitations themselves, respectively. In particular, it is suggested that the emergent photons associated with the fractionalized charge excitations can be measured in X-ray scattering experiments. Various other experimental signatures of the exotic cluster Mott insulator are discussed in light of candidate materials with partially-filled bands on the pyrochlore lattice.

cond-mat.str-el↗

arXiv2014

Compositional Structure Learning for Action Understanding

The focus of the action understanding literature has predominately been classification, how- ever, there are many applications demanding richer action understanding such as mobile robotics and video search, with solutions to classification, localization and detection. In this paper, we propose a compositional model that leverages a new mid-level representation called compositional trajectories and a locally articulated spatiotemporal deformable parts model (LALSDPM) for fully action understanding. Our methods is advantageous in capturing the variable structure of dynamic human activity over a long range. First, the compositional trajectories capture long-ranging, frequently co-occurring groups of trajectories in space time and represent them in discriminative hierarchies, where human motion is largely separated from camera motion; second, LASTDPM learns a structured model with multi-layer deformable parts to capture multiple levels of articulated motion. We implement our methods and demonstrate state of the art performance on all three problems: action detection, localization, and recognition.

cs.CV↗

arXiv2014

The energy dependence of antiparticle to particle ratios in high energy pp collisions

The energy dependence of antiparticle to particle ratios in pp collisions of high energy is studied using the PACIEA and DCPC model. The yield ratios of antimatter and matter for different masses are measured at various energies. It is found that the yield ratios of antimatter and matter increase with the increase of the c.m energy of pp collisions until they gradually approach to 1 after the c.m energy is more than 200 GeV. The distribution of transverse momentum also has signiffcant dependence on the energy and mass, i.e the average transverse momentum increase when the c.m energy of pp collisions increase. The model results are compatible with the STAR preliminary datum.

nucl-th↗

arXiv2014

Significant reduction of lattice thermal conductivity by electron-phonon interaction in silicon with high carrier concentrations: a first-principles study

Electron-phonon interaction has been well known to create major resistance to electron transport in metals and semiconductors, whereas less studies were directed to its effect on the phonon transport, especially in semiconductors. We calculate the phonon lifetimes due to scattering with electrons (or holes), combine them with the intrinsic lifetimes due to the anharmonic phonon-phonon interaction, all from first-principles, and evaluate the effect of the electron-phonon interaction on the lattice thermal conductivity of silicon. Unexpectedly, we find a significant reduction of the lattice thermal conductivity at room temperature as the carrier concentration goes above 1e19 cm-3 (the reduction reaches up to 45% in p-type silicon at around 1e21 cm-3), a range of great technological relevance to thermoelectric materials.

cond-mat.mtrl-sci↗

arXiv2014

Reconstructing phonon mean free path contributions to thermal conductivity using nanoscale membranes

Knowledge of the mean free path distribution of heat-carrying phonons is key to understanding phonon-mediated thermal transport. We demonstrate that thermal conductivity measurements of thin membranes spanning a wide thickness range can be used to characterize how bulk thermal conductivity is distributed over phonon mean free paths. A non-contact transient thermal grating technique was used to measure the thermal conductivity of suspended Si membranes ranging from 15 to 1500 nm in thickness. A decrease in the thermal conductivity from 74% to 13% of the bulk value is observed over this thickness range, which is attributed to diffuse phonon boundary scattering. Due to the well-defined relation between the membrane thickness and phonon mean free path suppression, combined with the range and accuracy of the measurements, we can reconstruct the bulk thermal conductivity accumulation vs. phonon mean free path, and compare with theoretical models.

cond-mat.mes-hall↗

arXiv2014

Spin susceptibility anomaly in cluster Mott insulators on a partially-filled anisotropic Kagome lattice: applications to LiZn2Mo3O8

Motivated by recent experiments on the quantum-spin-liquid candidate material LiZn2Mo3O8, we study a single-band extended Hubbard model on an anisotropic Kagome lattice with the 1/6 electron filling. Due to the partial filling of the lattice, the inter-site repulsive interaction is necessary to generate Mott insulators, where electrons are localized in clusters, rather than at lattice sites. We provide examples of such cluster Mott insulators and study the phase transitions between metallic states and cluster Mott insulators on an anisotropic Kagome lattice. It is shown that these cluster Mott insulators are generally U(1) quantum spin liquids with spinon Fermi surfaces. However, the nature of charge excitations in different cluster Mott insulators could be quite different and we show that there exists a novel cluster Mott insulator where charge fluctuations around the hexagonal cluster induce a plaquette charge order (PCO). The spinon excitation spectrum in this spin-liquid cluster Mott insulator is reconstructed due to the PCO so that only 1/3 of the total spinon excitations are magnetically active. The strong coupling limit of the same model is also analyzed via a Kugel-Khomskii-like model. Based on these results, we propose that the anomalous behavior of the finite-temperature spin-susceptibility in LiZn2Mo3O8 may be explained by finite-temperature properties of the cluster Mott insulator with the PCO as well as fractionalized spinon excitations. Existing and possible future experiments on LiZn2Mo3O8, and other Mo-based cluster magnets are discussed in light of these theoretical predictions.

cond-mat.str-el↗

arXiv2014

SU(2) Ginzburg-Landau theory for degenerate Fermi gases with synthetic non-Abelian gauge fields

The non-Abelian gauge fields play a key role in achieving novel quantum phenomena in condensed-matter and high-energy physics. Recently, the synthetic non-Abelian gauge fields have been created in the neutral degenerate Fermi gases, and moreover, generate many exotic effects. All the previous predictions can be well understood by the microscopic Bardeen-Cooper-Schrieffer theory. In this work, we establish an SU(2) Ginzburg-Landau theory for degenerate Fermi gases with the synthetic non-Abelian gauge fields. We firstly address a fundamental problem how the non-Abelian gauge fields, imposing originally on the Fermi atoms, affect the pairing field with no extra electric charge by a local gauge-field theory,and then obtain the first and second SU(2) Ginzburg-Landau equations. Based on these obtained SU(2) Ginzburg-Landau equations, we find that the superfluid critical temperature of the intra- (inter-) band pairing increases (decreases) linearly, when increasing the strength of the synthetic non-Abelian gauge fields. More importantly, we predict a novel SU(2) non-Abelian Josephson effect, which can be used to design a new atomic superconducting quantum interference device.

cond-mat.quant-gas↗