Searcharxiv⌕ Search

arXiv subjects

Fei Xue

Publications and source records attributed to Fei Xue.

At least 109 records · Page 6Linked to original sources
arXiv2019

Local Supports Global: Deep Camera Relocalization with Sequence Enhancement

We propose to leverage the local information in image sequences to support global camera relocalization. In contrast to previous methods that regress global poses from single images, we exploit the spatial-temporal consistency in sequential images to alleviate uncertainty due to visual ambiguities by incorporating a visual odometry (VO) component. Specifically, we introduce two effective steps called content-augmented pose estimation and motion-based refinement. The content-augmentation step focuses on alleviating the uncertainty of pose estimation by augmenting the observation based on the co-visibility in local maps built by the VO stream. Besides, the motion-based refinement is formulated as a pose graph, where the camera poses are further optimized by adopting relative poses provided by the VO component as additional motion constraints. Thus, the global consistency can be guaranteed. Experiments on the public indoor 7-Scenes and outdoor Oxford RobotCar benchmark datasets demonstrate that benefited from local information inherent in the sequence, our approach outperforms state-of-the-art methods, especially in some challenging cases, e.g., insufficient texture, highly repetitive textures, similar appearances, and over-exposure.

cs.CV↗
arXiv2019

Individualized Multi-directional Variable Selection

In this paper we propose a heterogeneous modeling framework which achieves individual-wise feature selection and individualized covariates' effects subgrouping simultaneously. In contrast to conventional model selection approaches, the new approach constructs a separation penalty with multi-directional shrinkages, which facilitates individualized modeling to distinguish strong signals from noisy ones and selects different relevant variables for different individuals. Meanwhile, the proposed model identifies subgroups among which individuals share similar covariates' effects, and thus improves individualized estimation efficiency and feature selection accuracy. Moreover, the proposed model also incorporates within-individual correlation for longitudinal data to gain extra efficiency. We provide a general theoretical foundation under a double-divergence modeling framework where the number of individuals and the number of individual-wise measurements can both diverge, which enables inference on both an individual level and a population level. In particular, we establish strong oracle property for the individualized estimator to ensure its optimal large sample property under various conditions. An efficient ADMM algorithm is developed for computational scalability. Simulation studies and applications to post-trauma mental disorder analysis with genetic variation and an HIV longitudinal treatment study are illustrated to compare the new approach to existing methods.

stat.ME↗
arXiv2019

Beyond Tracking: Selecting Memory and Refining Poses for Deep Visual Odometry

Most previous learning-based visual odometry (VO) methods take VO as a pure tracking problem. In contrast, we present a VO framework by incorporating two additional components called Memory and Refining. The Memory component preserves global information by employing an adaptive and efficient selection strategy. The Refining component ameliorates previous results with the contexts stored in the Memory by adopting a spatial-temporal attention mechanism for feature distilling. Experiments on the KITTI and TUM-RGBD benchmark datasets demonstrate that our method outperforms state-of-the-art learning-based methods by a large margin and produces competitive results against classic monocular VO approaches. Especially, our model achieves outstanding performance in challenging scenarios such as texture-less regions and abrupt motions, where classic VO algorithms tend to fail.

cs.CV↗
arXiv2018

Guided Feature Selection for Deep Visual Odometry

We present a novel end-to-end visual odometry architecture with guided feature selection based on deep convolutional recurrent neural networks. Different from current monocular visual odometry methods, our approach is established on the intuition that features contribute discriminately to different motion patterns. Specifically, we propose a dual-branch recurrent network to learn the rotation and translation separately by leveraging current Convolutional Neural Network (CNN) for feature representation and Recurrent Neural Network (RNN) for image sequence reasoning. To enhance the ability of feature selection, we further introduce an effective context-aware guidance mechanism to force each branch to distill related information for specific motion pattern explicitly. Experiments demonstrate that on the prevalent KITTI and ICL_NUIM benchmarks, our method outperforms current state-of-the-art model- and learning-based methods for both decoupled and joint camera pose recovery.

cs.CV↗
arXiv2018

TRPL+K: Thick-Restart Preconditioned Lanczos+K Method for Large Symmetric Eigenvalue Problems

The Lanczos method is one of the standard approaches for computing a few eigenpairs of a large, sparse, symmetric matrix. It is typically used with restarting to avoid unbounded growth of memory and computational requirements. Thick-restart Lanczos is a popular restarted variant because of its simplicity and numerically robustness. However, convergence can be slow for highly clustered eigenvalues so more effective restarting techniques and the use of preconditioning is needed. In this paper, we present a thick-restart preconditioned Lanczos method, TRPL+K, that combines the power of locally optimal restarting (+K) and preconditioning techniques with the efficiency of the thick-restart Lanczos method. TRPL+K employs an inner-outer scheme where the inner loop applies Lanczos on a preconditioned operator while the outer loop augments the resulting Lanczos subspace with certain vectors from the previous restart cycle to obtain eigenvector approximations with which it thick restarts the outer subspace. We first identify the differences from various relevant methods in the literature. Then, based on an optimization perspective, we show an asymptotic global quasi-optimality of a simplified TRPL+K method compared to an unrestarted global optimal method. Finally, we present extensive experiments showing that TRPL+K either outperforms or matches other state-of-the-art eigenmethods in both matrix-vector multiplications and computational time.

math.NA↗
arXiv2018

Time-Reversal Symmetry-Breaking Nematic Insulators near Quantum Spin Hall Phase Transitions

We study the phase diagram of a model quantum spin Hall system as a function of band inversion and band-coupling strength, demonstrating that when band hybridization is weak, an interaction-induced nematic insulator state emerges over a wide range of band inversion. This property is a consequence of the long-range Coulomb interaction, which favors interband phase coherence that is weakly dependent on momentum and therefore frustrated by the single-particle Hamiltonian at the band inversion point. For weak band hybridization, interactions convert the continuous gap closing topological phase transition at inversion into a pair of continuous phase transitions bounding a state with broken time-reversal and rotational symmetries. At intermediate band hybridization, the topological phase transition proceeds instead via a quantum anomalous Hall insulator state, whereas at strong hybridization interactions play no role. We comment on the implications of our findings for InAs/GaSb and HgTe/CdTe quantum spin Hall systems.

cond-mat.mes-hall↗
arXiv2018

Boltzmann scaling of spontaneous Hall current and nonequilibrium spin-polarization

We extend the semiclassical Boltzmann formalism for the anomalous Hall effect (AHE) in nondegenerate multiband electron systems to the spin Hall effect (SHE) and unconventional Edelstein effect (UEE, cannot be accounted for by the conventional Boltzmann equation, unlike the conventional Edelstein effect). This extension is confirmed by extending the Kohn-Luttinger density-matrix transport theory in the weak disorder-potential regime. By performing Kubo linear response calculations in a prototypical multiband model, the Boltzmann scaling for the AHE/SHE and UEE is found to be valid only if the disorder-broadening of bands is quite smaller than the minimal intrinsic energy-scale around the Fermi level. Discussions on this criterion in various multiband systems are also presented. A qualitative phase diagram is proposed to show the influences of changing independently the impurity density and strength of disorder potential on the AHE/SHE and UEE.

cond-mat.mes-hall↗
arXiv2018

Boltzmann approach to spin-orbit-induced transport in effective quantum theories

In model studies of the spin/anomalous Hall effect, effective Hamiltonians often serve as the starting point. However, a complete effective quantum theory contains not only the effective Hamiltonian but also the relation linking the physical observables to the canonical ones. We construct the semiclassical Boltzmann (SB) transport framework in the weak disorder-potential regime directly in the level of the effective quantum theory, and confirm this construction by formulating a generalized Kohn-Luttinger density matrix transport theory also in this level. The link and difference between the present SB theory and previous phenomenological Boltzmann, quantum kinetic and usual Kubo-Streda theories are clarified. We also present the slightly generalized Kubo-Streda formula in the level of the effective quantum theory. In this level, it is the generalized Kubo-Streda formula rather than the usual one that leads to the same physical interpretations as the present SB theory. In the application to a Rashba 2D effective model, a nonzero spin Hall effect important in the case of strong Rashba coupling but neglected in previous theories is found.

cond-mat.mes-hall↗
arXiv2017

Instability and topological robustness of Weyl semimetals against Coulomb interaction

There is a close connection between various new phenomena in Weyl semimetals and the existence of linear band crossings in the single particle description. We show, by a full self-consistent mean-field calculation, how this picture is modified in the presence of long-range Coulomb interactions. The chiral symmetry breaking occurs at strong enough interactions and the internode interband excitonic pairing channel is found to be significant, which determines the gap-opened band profile varying with interaction strength. Remarkably, in the resultant interacting phase, finite band Chern number jumps in the three-dimensional momentum space are retained, indicating the robustness of the topologically nontrivial features.

cond-mat.mes-hall↗
arXiv2017

Determining the direction of a nanowire's flexural vibrations by micro-lens optical fiber interferometer

Nanowires are perfect transducers for ultra-sensitive detections of force and mass. For small mass sensing, recent advances in detecting more properties than their masses with high sensitivity for absorbed particles onto a nanowire rely on identifications of two orthogonal flexural vibration modes of the nanowire. The directions of these orthogonal flexural vibrations with respect to measurement direction are crucial parameters of vibration modes. However, previous method, which determines a nanowire's vibration direction using thermal vibrations, requires simultaneously detecting of two orthogonal flexural vibrations of the nanowire with sufficient sensitivity. In this work, we propose and realize a method for the determination of directions of a nanowire's flexural vibrations by micro-lens optical fiber interferometer. Our method combines the light interference and light scattering of the nanowire. It does not require detecting a pair of degenerated orthogonal flexural vibration modes. Therefore, our method is expected to have wide usages in characterizing nanowire's vibrations and their applications.

cond-mat.mes-hall↗
arXiv2017

Cooling Flexural Modes of a Mechanical Oscillator by Magnetic Trapped Bose-Einstein Condensate Atoms

We theoretically study cooling of flexural modes of a mechanical oscillator by Bose-Einstein Condensate (BEC) atoms (Rb87) trapped in a magnetic trap. The mechanical oscillator with a tiny magnet attached on one of its free ends produces an oscillating magnetic field. When its oscillating frequency matches certain hyperfine Zeeman energy of Rb87 atoms, the trapped BEC atoms are coupled out of the magnetic trap by the mechanical oscillator, flying away from the trap with stolen energy from the mechanical oscillator. Thus the mode temperature of the mechanical oscillator is reduced. The mode temperature of the steady state of a mechanical oscillator, measured by the mean steady-state phonon number in the flexural mode of the mechanical oscillator, is analyzed. It is found that ground state (phonon number less than 1) may be accessible with optimal parameters of the hybrid system of a mechanical oscillator and trapped BEC atoms.

quant-ph↗
arXiv2017

Mechanical Switching of Nanoscale Multiferroic Phase Boundaries

Tuning the lattice degree of freedom in nanoscale functional crystals is critical to exploit the emerging functionalities such as piezoelectricity, shape-memory effect, or piezomagnetism, which are attributed to the intrinsic lattice-polar or lattice-spin coupling. Here it is reported that a mechanical probe can be a dynamic tool to switch the ferroic orders at the nanoscale multiferroic phase boundaries in BiFeO 3 with a phase mixture, where the material can be reversibly transformed between the "soft" tetragonal-like and the "hard" rhombohedrallike structures. The microscopic origin of the nonvolatile mechanical switching of the multiferroic phase boundaries, coupled with a reversible 180° rotation of the in-plane ferroelectric polarization, is the nanoscale pressure-induced elastic deformation and reconstruction of the spontaneous strain gradient across the multiferroic phase boundaries. The reversible control of the room-temperature multiple ferroic orders using a pure mechanical stimulus may bring us a new pathway to achieve the potential energy conversion and sensing applications.

cond-mat.mtrl-sci↗
arXiv2017

Strain induced incommensurate phases in hexagonal manganites

An incommensurate phase refers to a solid state in which the period of a superstructure is incommensurable with the primitive unit cell. Recently the incommensurate phase is induced by applying an in-plane strain to hexagonal manganites, which demonstrates single chiral modulation of six domain variants. Here we employ Landau theory in combination with the phase-field method to investigate the incommensurate phase in hexagonal manganites. It is shown that the equilibrium wave length of the incommensurate phase is determined by temperature and the magnitude of the applied strain, and a temperature-strain phase diagram is constructed for the stability of the incommensurate phase. Temporal evolution of domain structures reveals that the applied strain not only produces the force pulling the vortices and anti-vortices in opposite directions, but also results in the creation and annihilation of vortex-antivortex pairs.

cond-mat.mtrl-sci↗
arXiv2017

Three-dimensional vortex structures and dynamics in hexagonal manganites

Hexagonal manganites REMnO3 (RE, rare earths) have attracted significant attention due to their potential applications as multiferroic materials and the intriguing physics associated with the topological defects. The two-dimensional (2D) and 3D domain and vortex structure evolution of REMnO3 is predicted using the phase-field method based on a thermodynamic potential constructed from first-principles calculations. In 3D spaces, vortex lines show three types of topological changes, i.e. shrinking, coalescence, and splitting, with the latter two caused by the interaction and exchange of vortex loops. Compared to the coarsening rate of the isotropic XY model, the six-fold degeneracy gives rise to negligible differences with the vortex-antivortex annihilation controlling the scaling dynamics, whereas the anisotropy of interfacial energy results in a deviation. The temporal evolution of domain and vortex structures serves as a platform to fully explore the mesoscale mechanisms for the 0-D and 1-D topological defects.

cond-mat.mtrl-sci↗
arXiv2017

On the Banach-Mazur Distance between the Cube and the Crosspolytope

In this note we study the Banach-Mazur distance between the $n$-dimensional cube and the crosspolytope. Previous work shows that the distance has order $\sqrt{n}$, and here we will prove some explicit bounds improving on former results. Even in dimension 3 the exact distance is not known, and based on computational results it is conjectured to be $\frac{9}{5}$. Here we will also present computerbased potential optimal results in dimension $4$ to $8$.

math.MG↗
arXiv2017

Theory of Two-Dimensional Spatially Indirect Equilibrium Exciton Condensates

We present a theory of bilayer two-dimensional electron systems that host a spatially indirect exciton condensate when in thermal equilibrium. Equilibrium bilayer exciton condensates (BXCs) are expected to form when two nearby semiconductor layers are electrically isolated, and when the conduction band of one layer is brought close to degeneracy with the valence band of a nearby layer by varying bias or gate voltages. BXCs are characterized by spontaneous inter-layer phase coherence and counterflow superfluidity. The bilayer system we consider is composed of two transition metal dichalcogenide monolayers separated and surrounded by hexagonal boron nitride. We use mean-field-theory and a bosonic weakly interacting exciton model to explore the BXC phase diagram, and time-dependent mean-field theory to address condensate collective mode spectra and quantum fluctuations. We find that a phase transition occurs between states containing one and two condensate components as the layer separation and the exciton density are varied, and derive simple approximate expressions for the exciton-exciton interaction strength which we show can be measured capacitively.

cond-mat.mes-hall↗