arXiv subjects

Jie Liu

Publications and source records attributed to Jie Liu.

At least 595 records · Page 33Linked to original sources

arXiv2020

Non-Abelian braiding of Dirac fermionic modes using topological corner states in higher-order topological insulator

We numerically demonstrate that the topological corner states residing in the corners of higher-order topological insulator possess non-Abelian braiding properties. Such topological corner states are Dirac fermionic modes other than Majorana zero-modes. We claim that Dirac fermionic modes protected by nontrivial topology also support non-Abelian braiding. An analytical description on such non-Abelian braiding is conducted based on the vortex-induced Dirac-type fermionic modes. The braiding operator for Dirac fermionic modes is also analytically derivated and compared with the Majorana zero-modes. Experimentally, such non-Abelian braiding operation on Dirac fermionic modes is proposed to be testified through topological electric circuit.

cond-mat.mes-hall↗

arXiv2019

Balancing truncation and round-off errors in practical FEM: one-dimensional analysis

In finite element methods (FEMs), the accuracy of the solution cannot increase indefinitely because the round-off error increases when the number of degrees of freedom (DoFs) is large enough. This means that the accuracy that can be reached is limited. A priori information of the highest attainable accuracy is therefore of great interest. In this paper, we devise an innovative method to obtain the highest attainable accuracy. In this method, the truncation error is extrapolated when it converges at the analytical rate, for which only a few primary $h$-refinements are required, and the bound of the round-off error is provided through extensive numerical experiments. The highest attainable accuracy is obtained by minimizing the sum of these two types of errors. We validate this method using a one-dimensional Helmholtz equation in space. It shows that the highest attainable accuracy can be accurately predicted, and the CPU time required is much less compared with that using the successive $h$-refinement.

math.NA↗

arXiv2019

Understanding and Improving Proximity Graph based Maximum Inner Product Search

The inner-product navigable small world graph (ip-NSW) represents the state-of-the-art method for approximate maximum inner product search (MIPS) and it can achieve an order of magnitude speedup over the fastest baseline. However, to date it is still unclear where its exceptional performance comes from. In this paper, we show that there is a strong norm bias in the MIPS problem, which means that the large norm items are very likely to become the result of MIPS. Then we explain the good performance of ip-NSW as matching the norm bias of the MIPS problem - large norm items have big in-degrees in the ip-NSW proximity graph and a walk on the graph spends the majority of computation on these items, thus effectively avoids unnecessary computation on small norm items. Furthermore, we propose the ip-NSW+ algorithm, which improves ip-NSW by introducing an additional angular proximity graph. Search is first conducted on the angular graph to find the angular neighbors of a query and then the MIPS neighbors of these angular neighbors are used to initialize the candidate pool for search on the inner-product proximity graph. Experiment results show that ip-NSW+ consistently and significantly outperforms ip-NSW and provides more robust performance under different data distributions.

cs.IR↗

arXiv2019

Model Embedded DRL for Intelligent Greenhouse Control

Greenhouse environment is the key to influence crops production. However, it is difficult for classical control methods to give precise environment setpoints, such as temperature, humidity, light intensity and carbon dioxide concentration for greenhouse because it is uncertain nonlinear system. Therefore, an intelligent close loop control framework based on model embedded deep reinforcement learning (MEDRL) is designed for greenhouse environment control. Specifically, computer vision algorithms are used to recognize growing periods and sex of crops, followed by the crop growth models, which can be trained with different growing periods and sex. These model outputs combined with the cost factor provide the setpoints for greenhouse and feedback to the control system in real-time. The whole MEDRL system has capability to conduct optimization control precisely and conveniently, and costs will be greatly reduced compared with traditional greenhouse control approaches.

cs.LG↗

arXiv2019

Flow Rate Control in Smart District Heating Systems Using Deep Reinforcement Learning

At high latitudes, many cities adopt a centralized heating system to improve the energy generation efficiency and to reduce pollution. In multi-tier systems, so-called district heating, there are a few efficient approaches for the flow rate control during the heating process. In this paper, we describe the theoretical methods to solve this problem by deep reinforcement learning and propose a cloud-based heating control system for implementation. A real-world case study shows the effectiveness and practicability of the proposed system controlled by humans, and the simulated experiments for deep reinforcement learning show about 1985.01 gigajoules of heat quantity and 42276.45 tons of water are saved per hour compared with manual control.

eess.SY↗

arXiv2019

Norm-Explicit Quantization: Improving Vector Quantization for Maximum Inner Product Search

Vector quantization (VQ) techniques are widely used in similarity search for data compression, fast metric computation and etc. Originally designed for Euclidean distance, existing VQ techniques (e.g., PQ, AQ) explicitly or implicitly minimize the quantization error. In this paper, we present a new angle to analyze the quantization error, which decomposes the quantization error into norm error and direction error. We show that quantization errors in norm have much higher influence on inner products than quantization errors in direction, and small quantization error does not necessarily lead to good performance in maximum inner product search (MIPS). Based on this observation, we propose norm-explicit quantization (NEQ) --- a general paradigm that improves existing VQ techniques for MIPS. NEQ quantizes the norms of items in a dataset explicitly to reduce errors in norm, which is crucial for MIPS. For the direction vectors, NEQ can simply reuse an existing VQ technique to quantize them without modification. We conducted extensive experiments on a variety of datasets and parameter configurations. The experimental results show that NEQ improves the performance of various VQ techniques for MIPS, including PQ, OPQ, RQ and AQ.

cs.IR↗

arXiv2019

Creating Auxiliary Representations from Charge Definitions for Criminal Charge Prediction

Charge prediction, determining charges for criminal cases by analyzing the textual fact descriptions, is a promising technology in legal assistant systems. In practice, the fact descriptions could exhibit a significant intra-class variation due to factors like non-normative use of language, which makes the prediction task very challenging, especially for charge classes with too few samples to cover the expression variation. In this work, we explore to use the charge definitions from criminal law to alleviate this issue. The key idea is that the expressions in a fact description should have corresponding formal terms in charge definitions, and those terms are shared across classes and could account for the diversity in the fact descriptions. Thus, we propose to create auxiliary fact representations from charge definitions to augment fact descriptions representation. The generated auxiliary representations are created through the interaction of fact description with the relevant charge definitions and terms in those definitions by integrated sentence- and word-level attention scheme. Experimental results on two datasets show that our model achieves significant improvement than baselines, especially for classes with few samples.

cs.CL↗

arXiv2019

Possibility to construct a clock on non 0-0 transition

In this letter, we discuss the possibility of non 0-0 atomic lines as the reference transitions to design a new atomic clock. The proposed clock operates in high C-field regime offering an interesting alternative to conventional clock based on 0-0 transition as the former can also eliminate the influence of the first-order Zeeman effect. Using the Breit-Rabi formula, we theoretically calculate the magnetically insensitive frequency and the corresponding C-field value for different alkali atom species. We also provide an estimate of the uncertainty of frequency shift to change of C-field. We show that a high C-field clock based on Cs: 3,-1 - 4,0, 3,0 - 4,-1, and 3,-1 - 4,-1, can achieve a bias of 10-14 with a C-field of 10-6 uncertainty. The free first-order Zeeman effect C-field for 3,-1 - 4,0 or 3,0 - 4,-1 transition is about one half that for 3,-1 - 4,-1 transition.

physics.atom-ph↗

arXiv2019

Performance Analysis and Characterization of Training Deep Learning Models on Mobile Devices

Training deep learning models on mobile devices recently becomes possible, because of increasing computation power on mobile hardware and the advantages of enabling high user experiences. Most of the existing work on machine learning at mobile devices is focused on the inference of deep learning models (particularly convolutional neural network and recurrent neural network), but not training. The performance characterization of training deep learning models on mobile devices is largely unexplored, although understanding the performance characterization is critical for designing and implementing deep learning models on mobile devices. In this paper, we perform a variety of experiments on a representative mobile device (the NVIDIA TX2) to study the performance of training deep learning models. We introduce a benchmark suite and tools to study performance of training deep learning models on mobile devices, from the perspectives of memory consumption, hardware utilization, and power consumption. The tools can correlate performance results with fine-grained operations in deep learning models, providing capabilities to capture performance variance and problems at a fine granularity. We reveal interesting performance problems and opportunities, including under-utilization of heterogeneous hardware, large energy consumption of the memory, and high predictability of workload characterization. Based on the performance analysis, we suggest interesting research directions.

cs.LG↗

arXiv2019

On the Constantin-Lax-Majda Model with Convection

The well-known Constantin-Lax-Majda (CLM) equation, an important toy model of the 3D Euler equations without convection, can develop finite time singularities [5]. De Gregorio modified the CLM model by adding a convective term [6], which is known important for fluid dynamics [10,14]. Presented are two results on the De Gregorio model. The first one is the global well-posedness of such a model for general initial data with non-negative (or non-positive) vorticity which is based on a newly discovered conserved quantity. This verifies the numerical observations for such class of initial data. The second one is an exponential stability result of ground states, which is similar to the recent significant work of Jia, Steward and Sverak [11], with the zero mean constraint on the initial data being removable. The novelty of the method is the introduction of the new solution space $\mathcal{H}_{DW}$ together with a new basis and an effective inner product of $\mathcal{H}_{DW}$.

math.AP↗

arXiv2019

Majorana zero modes by engineering topological kink states in two dimensional electron gas

Majorana zero modes (MZMs)--bearing potential applications for topological quantum computing--are verified in quasi-one-dimensional (1D) Fermion systems, including semiconductor nanowires, magnetic atomic chains, planar Josephson junctions. However, the existence of multi-bands in these systems makes the MZMs fragile to the influence of disorder. Moreover, in practical perspective, the proximity induced superconductivity may be difficult and restricted for 1D systems. Here, we propose a flexible route to realize MZMs through 1D topological kink states by engineering a 2D electron gas with antidot lattices, in which both the aforementioned issues can be avoided owing to the robustness of kink states and the intrinsically attainable superconductivity in high-dimensional systems. The MZMs are verified to be quite robust against disorders and the bending of kink states, and can be conveniently tuned by varying the Rashba spin-orbit coupling strength. Our proposal provides an experimental feasible platform for MZMs with systematic manipulability and assembleability based on the present techniques in 2D electron gas system.

cond-mat.mes-hall↗

arXiv2019

AI for Earth: Rainforest Conservation by Acoustic Surveillance

Saving rainforests is a key to halting adverse climate changes. In this paper, we introduce an innovative solution built on acoustic surveillance and machine learning technologies to help rainforest conservation. In particular, We propose new convolutional neural network (CNN) models for environmental sound classification and achieved promising preliminary results on two datasets, including a public audio dataset and our real rainforest sound dataset. The proposed audio classification models can be easily extended in an automated machine learning paradigm and integrated in cloud-based services for real world deployment.

cs.SD↗

arXiv2019

Bound state and non-Markovian dynamics of a quantum emitter around a surface plasmonic nanostructure

A bound state between a quantum emitter (QE) and surface plasmon polaritons (SPPs) can be formed, where the QE is partially stabilized in its excited state. We put forward a general approach for calculating the energy level shift at a negative frequency $ω$, which is just the negative of the nonresonant part for the energy level shift at positive frequency $-ω$. We also propose an efficient formalism for obtaining the long-time value of the excited-state population without calculating the eigenfrequency of the bound state or performing a time evolution of the system, in which the probability amplitude for the excited state in the steady limit is equal to one minus the integral of the evolution spectrum over the positive frequency range. With the above two quantities obtained, we show that the non-Markovian decay dynamics in the presence of a bound state can be obtained by the method based on the Green's function expression for the evolution operator. A general criterion for identifying the existence of a bound state is presented. These are numerically demonstrated for a QE located around a nanosphere and in a gap plasmonic nanocavity. These findings are instructive in the fields of coherent light-matter interactions.

cond-mat.mes-hall↗

arXiv2019

Spectral Numerical Mode Matching Method for 3D Layered Multi-Region Structures

The spectral numerical mode-matching (SNMM) method is developed to simulate the 3D layered multi-region structures. The SNMM method is a semi-analytical solver having the properties of dimensionality reduction to reduce computational costs; it is especially useful for microwave and optical integrated circuits where fabrication is often done in a layered structure. Furthermore, at some layer interfaces, very thin surfaces such as good conductor surfaces and metasurfaces can be deposited to achieve desired properties such as high absorbance and/or anomalous reflection/refraction. In this work, the 3D SNMM method is further extended from a single interface to multiple layers so that the electromagnetic propagation and scattering in the longitudinal direction is treated analytically through reflection and transmission matrices by using the eigenmode expansions in the transverse directions. Therefore, it effectively reduces the original 3D problem into a series of 2D eigenvalue problems for periodic structures. We apply this method to characterize metasurfaces and lithography models, and show that the SNMM method is especially efficient when the longitudinal layer thicknesses are large compared with wavelength. Numerical experiments indicate that the SNMM method is highly efficient and accurate for the metasurfaces and the lithography models.

physics.optics↗

arXiv2019

Single-Path Mobile AutoML: Efficient ConvNet Design and NAS Hyperparameter Optimization

Can we reduce the search cost of Neural Architecture Search (NAS) from days down to only few hours? NAS methods automate the design of Convolutional Networks (ConvNets) under hardware constraints and they have emerged as key components of AutoML frameworks. However, the NAS problem remains challenging due to the combinatorially large design space and the significant search time (at least 200 GPU-hours). In this work, we alleviate the NAS search cost down to less than 3 hours, while achieving state-of-the-art image classification results under mobile latency constraints. We propose a novel differentiable NAS formulation, namely Single-Path NAS, that uses one single-path over-parameterized ConvNet to encode all architectural decisions based on shared convolutional kernel parameters, hence drastically decreasing the search overhead. Single-Path NAS achieves state-of-the-art top-1 ImageNet accuracy (75.62%), hence outperforming existing mobile NAS methods in similar latency settings (~80ms). In particular, we enhance the accuracy-runtime trade-off in differentiable NAS by treating the Squeeze-and-Excitation path as a fully searchable operation with our novel single-path encoding. Our method has an overall cost of only 8 epochs (24 TPU-hours), which is up to 5,000x faster compared to prior work. Moreover, we study how different NAS formulation choices affect the performance of the designed ConvNets. Furthermore, we exploit the efficiency of our method to answer an interesting question: instead of empirically tuning the hyperparameters of the NAS solver (as in prior work), can we automatically find the hyperparameter values that yield the desired accuracy-runtime trade-off? We open-source our entire codebase at: https://github.com/dstamoulis/single-path-nas.

cs.LG↗

arXiv2019

Flux-induced Topological Superconductor in Planar Josephson Junction

A planar Josephson junction with a normal metal attached on its top surface will form a hollow nanowire structure due to its three dimensional nature. In such hollow nanowire structure, the magnetic flux induced by a small magnetic field (about 0.01T) will tune the system into topologically non-trivial phase and therefore two Majorana zero-modes will form at the ends of the nanowire. Through tuning the chemical potential of the normal metal, the topologically non-trivial phase can be obtained for almost all energy within the band. Furthermore, the system can be conveniently tuned between the topologically trivial and non-trivial phases via the phase difference between the superconductors. Such device, manipulable through flux, can be conveniently fabricated into desired 2D networks. Finally, we also propose a cross-shaped junction realizing the braiding of Majorana zero-modes through manipulating the phase differences.

cond-mat.supr-con↗

arXiv2019

Fine structure of swift heavy ion track in rutile TiO2

We report on the first observation of fine structure of latent tracks in rutile TiO2, which changes from cylinder to dumbbell-shape and then to sandglass-shape as a function of the ion path length. Based on inelastic thermal spike model, we show that Hagen-Poiseuille flow of molten phase produces the hillocks on surface and the void-rich zone near surface after epitaxial recrystallization due to material deficit, while at a deep depth, the lack of efficient outflow and recrystallization result in the absence of tracks. We propose that core-shell duration of transient molten phase induced by swift heavy ion and parabolic distribution of fluid velocity are radial-dependent. Moreover, the various morphologies of tracks are a consequence of the molten phase outflow and recrystallization during rapid cooling down. Our perspective provides a new interpretation in the track formation.

physics.app-ph↗

arXiv2019

Single-Path NAS: Device-Aware Efficient ConvNet Design

Can we automatically design a Convolutional Network (ConvNet) with the highest image classification accuracy under the latency constraint of a mobile device? Neural Architecture Search (NAS) for ConvNet design is a challenging problem due to the combinatorially large design space and search time (at least 200 GPU-hours). To alleviate this complexity, we propose Single-Path NAS, a novel differentiable NAS method for designing device-efficient ConvNets in less than 4 hours. 1. Novel NAS formulation: our method introduces a single-path, over-parameterized ConvNet to encode all architectural decisions with shared convolutional kernel parameters. 2. NAS efficiency: Our method decreases the NAS search cost down to 8 epochs (30 TPU-hours), i.e., up to 5,000x faster compared to prior work. 3. On-device image classification: Single-Path NAS achieves 74.96% top-1 accuracy on ImageNet with 79ms inference latency on a Pixel 1 phone, which is state-of-the-art accuracy compared to NAS methods with similar latency (<80ms).

cs.LG↗