• In this paper, we show that the peeling property still holds for Bondi-Sachs metrics with nonzero cosmological constant under the boundary condition given by Sommerfeld's radiation condition together with three nontrivial $\Lambda$-independent functions $B$, $a$, $b$. This should indicate the new boundary condition is natural. Moreover, we construct some nonstationary vacuum Bondi-Sachs metrics without Bondi news, which Newmann-Penrose quantities fall faster than usual. This provides a new feature of gravitational waves for nonzero cosmological constant.
  • Unmanned Aerial Vehicle (UAV) networks have emerged as a promising technique to rapidly provide wireless coverage to a geographical area, where a flying UAV can be fast deployed to serve as cell site. Existing work on UAV-enabled wireless networks overlook the fast UAV deployment for wireless coverage, and such deployment problems have only been studied recently in sensor networks. Unlike sensors, UAVs should be deployed to the air and they are generally different in flying speed, operating altitude and wireless coverage radius. By considering such UAV heterogeneity to cover the whole target area, this paper studies two fast UAV deployment problems: one is to minimize the maximum deployment delay among all UAVs (min-max) for fairness consideration, and the other is to minimize the total deployment delay (min-sum) for efficiency consideration. We prove both min-max and min-sum problems are NP-complete in general. When dispatching UAVs from the same location, we present an optimal algorithm of low computational complexity $O(n^2)$ for the min-max problem. When UAVs are dispatched from different locations, we propose to preserve their location order during deployment and successfully design a fully polynomial time approximation scheme (FPTAS) of computation complexity $O(n^2 \log \frac{1}{\epsilon})$ to arbitrarily approach the global optimum with relative error $\epsilon$. The min-sum problem is more challenging. When UAVs are dispatched from the same initial location, we present an approximation algorithm of linear time. As for the general case, we further reformulate it as a dynamic program and propose a pseudo polynomial-time algorithm to solve it optimally.
  • This short paper describes our solution to the 2018 IEEE World Congress on Computational Intelligence One-Minute Gradual-Emotional Behavior Challenge, whose goal was to estimate continuous arousal and valence values from short videos. We designed four base regression models using visual and audio features, and then used a spectral approach to fuse them to obtain improved performance.
  • Quantum Hall (QH) states are arguably the most ubiquitous examples of nontrivial topological order, requiring no special symmetry and elegantly characterized by the first Chern number. Their higher dimension generalizations are particularly interesting from both mathematical and phenomenological perspectives, and have attracted recent attention due to a few high profile experimental realizations. In this work, we derive from first principles the electromagnetic response of QH systems in arbitrary number of dimensions, and elaborate on the crucial roles played by their modified phase space density of states under the simultaneous presence of magnetic field and Berry curvature. We provide new mathematical results relating this phase space modification to the non-commutativity of phase space, and show how they are manifested as a Hall conductivity quantized by a higher Chern number. When a Fermi surface is present, additional response currents unrelated to these Chern numbers also appear. This unconventional response can be directly investigated through a few minimal models with specially chosen fluxes. These models, together with more generic 6D QH systems, can be realized in realistic 3D experimental setups like cold atom systems through possibly entangled synthetic dimensions.
  • This work proposes an automated algorithm, called NetAdapt, that adapts a pre-trained deep neural network to a mobile platform given a resource budget. While many existing algorithms simplify networks based on the number of MACs or the number of parameters, optimizing those indirect metrics may not necessarily reduce the direct metrics, such as latency and energy consumption. To solve this problem, NetAdapt incorporates direct metrics into its adaptation algorithm. These direct metrics are evaluated using empirical measurements, so that detailed knowledge of the platform and toolchain is not required. NetAdapt automatically and progressively simplifies a pre-trained network until the resource budget (e.g., latency) is met while maximizing the accuracy. Experiment results show that NetAdapt achieves better accuracy versus latency trade-offs on both mobile CPU and mobile GPU, compared with the state-of-the-art automated network simplification algorithms. For image classification on the ImageNet dataset, NetAdapt achieves up to a 1.66$\times$ speedup in measured inference latency with higher accuracy.
  • It is a fundamental, but still elusive question whether methods based on quantum mechanics, in particular on quantum entanglement, can be used for classical information processing and machine learning. Even partial answer to this question would bring important insights to both fields of both machine learning and quantum mechanics. In this work, we implement simple numerical experiments, related to pattern/images classification, in which we represent the classifiers by quantum matrix product states (MPS). Classical machine learning algorithm is then applied to these quantum states. We explicitly show how quantum features (i.e., single-site and bipartite entanglement) can emerge in such represented images; entanglement characterizes here the importance of data, and this information can be practically used to improve the learning procedures. Thanks to the low demands on the dimensions and number of the unitary matrices, necessary to construct the MPS, we expect such numerical experiments could open new paths in classical machine learning, and shed at same time lights on generic quantum simulations/computations.
  • We revisit the inductive matrix completion problem that aims to recover a rank-$r$ matrix with ambient dimension $d$ given $n$ features as the side prior information. The goal is to make use of the known $n$ features to reduce sample and computational complexities. We present and analyze a new gradient-based non-convex optimization algorithm that converges to the true underlying matrix at a linear rate with sample complexity only linearly depending on $n$ and logarithmically depending on $d$. To the best of our knowledge, all previous algorithms either have a quadratic dependency on the number of features in sample complexity or a sub-linear computational convergence rate. In addition, we provide experiments on both synthetic and real world data to demonstrate the effectiveness of our proposed algorithm.
  • Reading and understanding text is one important component in computer aided diagnosis in clinical medicine, also being a major research problem in the field of NLP. In this work, we introduce a question-answering task called MedQA to study answering questions in clinical medicine using knowledge in a large-scale document collection. The aim of MedQA is to answer real-world questions with large-scale reading comprehension. We propose our solution SeaReader--a modular end-to-end reading comprehension model based on LSTM networks and dual-path attention architecture. The novel dual-path attention models information flow from two perspectives and has the ability to simultaneously read individual documents and integrate information across multiple documents. In experiments our SeaReader achieved a large increase in accuracy on MedQA over competing models. Additionally, we develop a series of novel techniques to demonstrate the interpretation of the question answering process in SeaReader.
  • Recently, there has been a drive towards the realization of topological phases beyond conventional electronic materials, including phases defined in more than three dimensions. We propose a versatile and experimentally realistic approach of realizing a large variety of multi-component quantum Hall phases in 2D photonic crystals with quasi-periodically modulated defects. With a length scale introduced by a background resonator lattice, the defects are found to host various effective orbitals of $s$, $p$ and $d$-type symmetries, thus providing a monolithic platform for realizing multi-component topological states without requiring separate internal degrees of freedom in the physical setup. Notably, by coupling the defect modulations diagonally, we report the novel realization of "entangled" 4D QH phase which cannot be factorized into two copies of 2D QH phases each described by the 1st Chern number. The structure of this non-factorizability can be quantified by a classical entanglement entropy inspired by quantum information theory.
  • We propose a unified framework to solve general low-rank plus sparse matrix recovery problems based on matrix factorization, which covers a broad family of objective functions satisfying the restricted strong convexity and smoothness conditions. Based on projected gradient descent and the double thresholding operator, our proposed generic algorithm is guaranteed to converge to the unknown low-rank and sparse matrices at a locally linear rate, while matching the best-known robustness guarantee (i.e., tolerance for sparsity). At the core of our theory is a novel structural Lipschitz gradient condition for low-rank plus sparse matrices, which is essential for proving the linear convergence rate of our algorithm, and we believe is of independent interest to prove fast rates for general superposition-structured models. We illustrate the application of our framework through two concrete examples: robust matrix sensing and robust PCA. Experiments on both synthetic and real datasets corroborate our theory.
  • We have investigated the critical behavior of a shandite-type half-metal ferromagnet Co3Sn2S2. It exhibits a second-order paramagnetic-ferromagnetic phase transition with TC = 174 K. To investigate the nature of the magnetic phase transition, a detailed critical exponent study has been performed. The critical components beta, gamma, and delta determined using the modified Arrott plot, the Kouvel-Fisher method as well as the critical isotherm analysis are match reasonably well and follow the scaling equation, confirming that the exponents are unambiguous and intrinsic to the material. The determined exponents of Co3Sn2S2 deviates from theoretical estimated short-range universal models. Instead, Co3Sn2S2 exhibits long-range order in the nature of magnetic interaction with the spin decay as J(r) ~ 1/r^[-(d + sigma)] with sigma = 1.28.
  • Ionic liquid gating has a number of advantages over solid-state gating, especially for flexible or transparent devices and for applications requiring high carrier densities. However, the large number of charged ions near the channel inevitably results in Coulomb scattering, which limits the carrier mobility in otherwise clean systems. We develop a model for this Coulomb scattering. We validate our model experimentally using ionic liquid gating of graphene across varying thicknesses of hexagonal boron nitride, demonstrating that disorder in the bulk ionic liquid often dominates the scattering.
  • Weyl semimetals are novel topological conductors that host Weyl fermions as emergent quasiparticles. While the Weyl fermions in high-energy physics are strictly defined as the massless solution of the Dirac equation and uniquely fixed by Lorentz symmetry, there is no such constraint for a topological metal in general. Specifically, the Weyl quasiparticles can arise by breaking either the space-inversion ($\mathcal{I}$) or time-reversal ($\mathcal{T}$) symmetry. They can either respect Lorentz symmetry (type-I) or strongly violate it (type-II). To date, different types of Weyl fermions have been predicted to occur only in different classes of materials. In this paper, we present a significant materials breakthrough by identifying a large class of Weyl materials in the RAlX (R=Rare earth, Al, X=Ge, Si) family that can realize all different types of emergent Weyl fermions ($\mathcal{I}$-breaking, $\mathcal{T}$-breaking, type-I or type-II), depending on a suitable choice of the rare earth elements. Specifically, RAlX can be ferromagnetic, nonmagnetic or antiferromagnetic and the electronic band topology and topological nature of the Weyl fermions can be tuned. The unparalleled tunability and the large number of compounds make the RAlX family of compounds a unique Weyl semimetal class for exploring the wide-ranging topological phenomena associated with different types of emergent Weyl fermions in transport, spectroscopic and device-based experiments.
  • We consider a distributed optimal control problem governed by an elliptic PDE, and propose an embedded discontinuous Galerkin (EDG) method to approximate the solution. We derive optimal a priori error estimates for the state, dual state, the optimal control, and suboptimal estimates for the fluxes. We present numerical experiments to confirm our theoretical results.
  • In the first part of this work, we analyzed a Dirichlet boundary control problem for an elliptic convection diffusion PDE and proposed a new hybridizable discontinuous Galerkin (HDG) method to approximate the solution. For the case of a 2D polygonal domain, we also proved an optimal superlinear convergence rate for the control under certain assumptions on the domain and on the target state. In this work, we revisit the convergence analysis without these assumptions; in this case, the solution can have low regularity and we use a different analysis approach. We again prove an optimal convergence rate for the control, and present numerical results to illustrate the convergence theory.
  • We consider the robust phase retrieval problem of recovering the unknown signal from the magnitude-only measurements, where the measurements can be contaminated by both sparse arbitrary corruption and bounded random noise. We propose a new nonconvex algorithm for robust phase retrieval, namely Robust Wirtinger Flow to jointly estimate the unknown signal and the sparse corruption. We show that our proposed algorithm is guaranteed to converge linearly to the unknown true signal up to a minimax optimal statistical precision in such a challenging setting. Compared with existing robust phase retrieval methods, we achieve an optimal sample complexity of $O(n)$ in both noisy and noise-free settings. Thorough experiments on both synthetic and real datasets corroborate our theory.
  • We propose an embedded discontinuous Galerkin (EDG) method to approximate the solution of a distributed control problem governed by convection diffusion PDEs, and obtain optimal a priori error estimates for the state, dual state, their fluxes, and the control. Moreover, we prove the optimize-then-discretize (OD) and discrtize-then-optimize (DO) approaches coincide. Numerical results confirm our theoretical results.
  • The non-equilibrium boron nitride (BN) phase of zinc oxide (ZnO) has been reported for thin films and nanostructures, however, its properties are not well understood due to a persistent controversy that prevents reconciling experimental and first-principles results for its atomic coordinates. We use first-principles theoretical spectroscopy to accurately compute electronic and optical properties, including single-quasiparticle and excitonic effects: Band structures and densities of states are computed using density functional theory, hybrid functionals, and the $GW$ approximation. Accurate optical absorption spectra and exciton binding energies are computed by solving the Bethe-Salpeter equation for the optical polarization function. Using this data we show that the band-gap difference between BN-ZnO and wurtzite (WZ) ZnO agrees very well with experiment when the theoretical lattice geometry is used, but significantly disagrees for the experimental atomic coordinates. We also show that the optical anisotropy of BN-ZnO differs significantly from that of WZ-ZnO, allowing to optically distinguish both polymorphs. By using the transfer-matrix method to solve Maxwell's equations for thin films composed of both polymorphs, we illustrate that this opens up a promising route for tuning optical properties.
  • We report the physical properties of orthorhombic o-CuMnAs single crystal, which is predicted to be a topological Dirac semimetal with magnetic ground state and inversion symmetry broken. o-CuMnAs exhibits an antiferromagnetic transition with TN ~ 312 K. Further characterizations of magnetic properties suggest that the AFM order may be canted with the spin orientation in the bc plane. Small isotropic MR and linearly field-dependent Hall resistivity with positive slope indicate that single hole-type carries with high density and low mobility dominate the transport properties of o-CuMnAs. Furthermore, the result of low-temperature heat capacity shows that the effective mass of carriers is much larger than those in typical topological semimetals. These results imply that the carriers in o-CuMnAs exhibit remarkably different features from those of Dirac fermions predicted in theory.
  • Despite intense interest in realizing topological phases across a variety of electronic, photonic and mechanical platforms, the detailed microscopic origin of topological behavior often remains elusive. To bridge this conceptual gap, we show how hallmarks of topological modes - boundary localization and chirality - emerge from Newton's laws in mechanical topological systems. We first construct a gyroscopic lattice with analytically solvable edge modes, and show how the Lorentz and spring restoring forces conspire to support very robust "dangling bond" boundary modes. The chirality and locality of these modes intuitively emerges from microscopic balancing of restoring forces and cyclotron tendencies. Next, we introduce the highlight of this work, a very experimentally realistic mechanical non-equilibrium (Floquet) Chern lattice driven by AC electromagnets. Through appropriate synchronization of the AC driving protocol, the Floquet lattice is "pushed around" by a rotating potential analogous to an object washed ashore by water waves. Besides hosting "dangling bond" chiral modes analogous to the gyroscopic boundary modes, our Floquet Chern lattice also supports peculiar half-period chiral modes with no static analog. With key parameters controlled electronically, our setup has the advantage of being dynamically tunable for applications involving arbitrary Floquet modulations. The physical intuition gleaned from our two prototypical topological systems are applicable not just to arbitrarily complicated mechanical systems, but also photonic and electrical topological setups.
  • In apparel recognition, specialized models (e.g. models trained for a particular vertical like dresses) can significantly outperform general models (i.e. models that cover a wide range of verticals). Therefore, deep neural network models are often trained separately for different verticals. However, using specialized models for different verticals is not scalable and expensive to deploy. This paper addresses the problem of learning one unified embedding model for multiple object verticals (e.g. all apparel classes) without sacrificing accuracy. The problem is tackled from two aspects: training data and training difficulty. On the training data aspect, we figure out that for a single model trained with triplet loss, there is an accuracy sweet spot in terms of how many verticals are trained together. To ease the training difficulty, a novel learning scheme is proposed by using the output from specialized models as learning targets so that L2 loss can be used instead of triplet loss. This new loss makes the training easier and make it possible for more efficient use of the feature space. The end result is a unified model which can achieve the same retrieval accuracy as a number of separate specialized models, while having the model complexity as one. The effectiveness of our approach is shown in experiments.
  • This paper proposes an automatic spatially-aware concept discovery approach using weakly labeled image-text data from shopping websites. We first fine-tune GoogleNet by jointly modeling clothing images and their corresponding descriptions in a visual-semantic embedding space. Then, for each attribute (word), we generate its spatially-aware representation by combining its semantic word vector representation with its spatial representation derived from the convolutional maps of the fine-tuned network. The resulting spatially-aware representations are further used to cluster attributes into multiple groups to form spatially-aware concepts (e.g., the neckline concept might consist of attributes like v-neck, round-neck, etc). Finally, we decompose the visual-semantic embedding space into multiple concept-specific subspaces, which facilitates structured browsing and attribute-feedback product retrieval by exploiting multimodal linguistic regularities. We conducted extensive experiments on our newly collected Fashion200K dataset, and results on clustering quality evaluation and attribute-feedback product retrieval task demonstrate the effectiveness of our automatically discovered spatially-aware concepts.
  • In this paper, the influence of the fractional dimensions of the L\'evy path under the Earth's gravitational field is studied, and the phase transitions of energy and wave functions are obtained: the energy changes from discrete to continuous and wave functions change from non-degenerate to degenerate when dimension of L\'evy path becomes from integer to non-integer. By analyzing the phase transitions, we solve two popular problems. First, we find an exotic way to produce the bound states in the continuum (BICs), our approach only needs a simple potential, and does not depend on interactions between particles. Second, we address the continuity of the energy will become strong when the mass of the particle becomes small. By deeply analyze, it can provide a way to distinguish ultralight particles from others types in the Earth's gravitational field, and five popular particles are discussed. In addition, we obtain analytical expressions for the wave functions and energy in the Earth's gravitational field in the circumstance of a fractional fractal dimensional L\'evy path. Moreover, to consider the influence of the minimal length, we analyze the phase transitions and the BICs in the presence of the minimal length. We find the phenomenon energy shift do not exist, which is a common phenomenon in the presence of the minimal length, and hence such above phenomena can still be found. Finally, relations between our results and existing results are discussed.
  • We address propagation of light in nonlinear twisted multi-core fibers with alternating amplifying and absorbing cores that are arranged into the PT - symmetric structure. In this structure, the coupling strength between neighboring cores and global energy transport can be controlled not only by the nonlinearity strength, but also by gain and losses and by the fiber twisting rate. The threshold level of gain/losses, at which PT -symmetry breaking occurs, is a non-monotonic function of the fiber twisting rate and it can be reduced nearly to zero or, instead, notably increased just by changing this rate. Nonlinearity usually leads to the monotonic reduction of the symmetry breaking threshold in such fibers.
  • The emergence of two-dimensional (2D) materials has attracted a great deal of attention due to their fascinating physical properties and potential applications for future nanoelectronic devices. Since the first isolation of graphene, a Dirac material, a large family of new functional 2D materials have been discovered and characterized, including insulating 2D boron nitride, semiconducting 2D transition metal dichalcogenides and black phosphorus, and superconducting 2D bismuth strontium calcium copper oxide, molybdenum disulphide and niobium selenide, etc. Here, we report the identification of ferromagnetic thin flakes of Cr2Ge2Te6 (CGT) with thickness down to a few nanometers, which provides a very important piece to the van der Waals structures consisting of various 2D materials. We further demonstrate the giant modulation of the channel resistance of 2D CGT devices via electric field effect. Our results illustrate the gate voltage tunability of 2D CGT and the potential of CGT, a ferromagnetic 2D material, as a new functional quantum material for applications in future nanoelectronics and spintronics.