arXiv subjects

Chang Liu

Publications and source records attributed to Chang Liu.

At least 1,081 records · Page 60Linked to original sources

arXiv2019

Unified Gas-kinetic Wave-Particle Methods III: Multiscale Photon Transport

In this paper, we extend the unified gas-kinetic wave-particle (UGKWP) method to the multiscale photon transport. In this method, the photon free streaming and scattering processes are treated in an un-splitting way. The duality descriptions, namely the simulation particle and distribution function, are utilized to describe the photon. By accurately recovering the governing equations of the unified gas-kinetic scheme (UGKS), the UGKWP preserves the multiscale dynamics of photon transport from optically thin to optically thick regime. In the optically thin regime, the UGKWP becomes a Monte Carlo type particle tracking method, while in the optically thick regime, the UGKWP becomes a diffusion equation solver. The local photon dynamics of the UGKWP, as well as the proportion of wave-described and particle-described photons are automatically adapted according to the numerical resolution and transport regime. Compared to the $S_n$ -type UGKS, the UGKWP requires less memory cost and does not suffer ray effect. Compared to the implicit Monte Carlo (IMC) method, the statistical noise of UGKWP is greatly reduced and computational efficiency is significantly improved in the optically thick regime. Several numerical examples covering all transport regimes from the optically thin to optically thick are computed to validate the accuracy and efficiency of the UGKWP method. In comparison to the $S_n$ -type UGKS and IMC method, the UGKWP method may have several-order-of-magnitude reduction in computational cost and memory requirement in solving some multsicale transport problems.

physics.comp-ph↗

arXiv2019

Unified Gas-kinetic Wave-Particle Methods II: Multiscale Simulation on Unstructured Mesh

In this paper, we present a unified gas-kinetic wave-particle (UGKWP) method on unstructured mesh for multiscale simulation of continuum and rarefied flow. Inheriting from the multicale transport in the unified gas-kinetic scheme (UGKS), the integral solution of kinetic model equation is employed in the construction of UGKWP method to model the flow physics in the cell size and time step scales. A novel wave-particle adaptive formulation is introduced in the UGKWP method to describe the flow dynamics in each control volume. The local gas evolution is constructed through the dynamical interaction of the deterministic hydrodynamic wave and the stochastic kinetic particle. Within the resolution of cell size and time step, the decomposition, interaction, and evolution of the hydrodynamic wave and the kinetic particle depend on the ratio of the time step to the local particle collision time. In the rarefied flow regime, the flow physics is mainly recovered by the discrete particles and the UGKWP method performs as a stochastic particle method. In the continuum flow regime, the flow behavior is solely followed by macroscopic variable evolution and the UGKWP method becomes a gas-kinetic hydrodynamic flow solver for the viscous and heat-conducting Navier--Stokes solutions. In different flow regimes, many numerical test cases are computed to validate the UGKWP method on unstructured mesh. The UGKWP method can get the same UGKS solutions in all Knudsen regimes without the requirement of the time step and mesh size being less than than the particle collision time and mean free path. With an automatic wave-particle decomposition, the UGKWP method becomes very efficient. For example, at Mach number 30 and Knudsen number 0.1, in comparison with UGKS several-order-of-magnitude reductions in computational cost and memory requirement have been achieved by UGKWP.

physics.comp-ph↗

arXiv2019

Neural Network Model Extraction Attacks in Edge Devices by Hearing Architectural Hints

As neural networks continue their reach into nearly every aspect of software operations, the details of those networks become an increasingly sensitive subject. Even those that deploy neural networks embedded in physical devices may wish to keep the inner working of their designs hidden -- either to protect their intellectual property or as a form of protection from adversarial inputs. The specific problem we address is how, through heavy system stack, given noisy and imperfect memory traces, one might reconstruct the neural network architecture including the set of layers employed, their connectivity, and their respective dimension sizes. Considering both the intra-layer architecture features and the inter-layer temporal association information introduced by the DNN design empirical experience, we draw upon ideas from speech recognition to solve this problem. We show that off-chip memory address traces and PCIe events provide ample information to reconstruct such neural network architectures accurately. We are the first to propose such accurate model extraction techniques and demonstrate an end-to-end attack experimentally in the context of an off-the-shelf Nvidia GPU platform with full system stack. Results show that the proposed techniques achieve a high reverse engineering accuracy and improve the one's ability to conduct targeted adversarial attack with success rate from 14.6\%$\sim$25.5\% (without network architecture knowledge) to 75.9\% (with extracted network architecture).

cs.CR↗

arXiv2019

Diffusion of Colloidal Rods in Corrugated Channels

In many natural and artificial devices diffusive transport takes place in confined geometries with corrugated boundaries. Such boundaries cause both entropic and hydrodynamic effects, which have been studied only for the case of spherical particles. Here we experimentally investigate diffusion of particles of elongated shape confined into a corrugated quasi-two-dimensional channel. Elongated shape causes complex excluded-volume interactions between particle and channel walls which reduce the accessible configuration space and lead to novel entropic free energy effects. The extra rotational degree of freedom also gives rise to a complex diffusivity matrix that depends on both the particle location and its orientation. We further show how to extend the standard Fick-Jacobs theory to incorporate combined hydrodynamic and entropic effects, so as, for instance, to accurately predict experimentally measured mean first passage times along the channel. Our approach can be used as a generic method to describe translational diffusion of anisotropic particles in corrugated channels.

cond-mat.soft↗

arXiv2019

Electric Field-Tuned Topological Phase Transition in Ultra-Thin Na3Bi - Towards a Topological Transistor

The electric field induced quantum phase transition from topological to conventional insulator has been proposed as the basis of a topological field effect transistor [1-4]. In this scheme an electric field can switch 'on' the ballistic flow of charge and spin along dissipationless edges of the two-dimensional (2D) quantum spin Hall insulator [5-9], and when 'off' is a conventional insulator with no conductive channels. Such as topological transistor is promising for low-energy logic circuits [4], which would necessitate electric field-switched materials with conventional and topological bandgaps much greater than room temperature, significantly greater than proposed to date [6-8]. Topological Dirac semimetals(TDS) are promising systems in which to look for topological field-effect switching, as they lie at the boundary between conventional and topological phases [3,10-16]. Here we use scanning probe microscopy/spectroscopy (STM/STS) and angle-resolved photoelectron spectroscopy (ARPES) to show that mono- and bilayer films of TDS Na3Bi [3,17] are 2D topological insulators with bulk bandgaps >400 meV in the absence of electric field. Upon application of electric field by doping with potassium or by close approach of the STM tip, the bandgap can be completely closed then re-opened with conventional gap greater than 100 meV. The large bandgaps in both the conventional and quantum spin Hall phases, much greater than the thermal energy kT = 25 meV at room temperature, suggest that ultrathin Na3Bi is suitable for room temperature topological transistor operation.

cond-mat.mtrl-sci↗

arXiv2018

A Multiscale Image Denoising Algorithm Based On Dilated Residual Convolution Network

Image denoising is a classical problem in low level computer vision. Model-based optimization methods and deep learning approaches have been the two main strategies for solving the problem. Model-based optimization methods are flexible for handling different inverse problems but are usually time-consuming. In contrast, deep learning methods have fast testing speed but the performance of these CNNs is still inferior. To address this issue, here we propose a novel deep residual learning model that combines the dilated residual convolution and multi-scale convolution groups. Due to the complex patterns and structures of inside an image, the multiscale convolution group is utilized to learn those patterns and enlarge the receptive field. Specifically, the residual connection and batch normalization are utilized to speed up the training process and maintain the denoising performance. In order to decrease the gridding artifacts, we integrate the hybrid dilated convolution design into our model. To this end, this paper aims to train a lightweight and effective denoiser based on multiscale convolution group. Experimental results have demonstrated that the enhanced denoiser can not only achieve promising denoising results, but also become a strong competitor in practical application.

cs.CV↗

arXiv2018

E-RNN: Design Optimization for Efficient Recurrent Neural Networks in FPGAs

Recurrent Neural Networks (RNNs) are becoming increasingly important for time series-related applications which require efficient and real-time implementations. The two major types are Long Short-Term Memory (LSTM) and Gated Recurrent Unit (GRU) networks. It is a challenging task to have real-time, efficient, and accurate hardware RNN implementations because of the high sensitivity to imprecision accumulation and the requirement of special activation function implementations. A key limitation of the prior works is the lack of a systematic design optimization framework of RNN model and hardware implementations, especially when the block size (or compression ratio) should be jointly optimized with RNN type, layer size, etc. In this paper, we adopt the block-circulant matrix-based framework, and present the Efficient RNN (E-RNN) framework for FPGA implementations of the Automatic Speech Recognition (ASR) application. The overall goal is to improve performance/energy efficiency under accuracy requirement. We use the alternating direction method of multipliers (ADMM) technique for more accurate block-circulant training, and present two design explorations providing guidance on block size and reducing RNN training trials. Based on the two observations, we decompose E-RNN in two phases: Phase I on determining RNN model to reduce computation and storage subject to accuracy requirement, and Phase II on hardware implementations given RNN model, including processing element design/optimization, quantization, activation implementation, etc. Experimental results on actual FPGA deployments show that E-RNN achieves a maximum energy efficiency improvement of 37.4$\times$ compared with ESE, and more than 2$\times$ compared with C-LSTM, under the same accuracy.

cs.CV↗

arXiv2018

Boosting Model Performance through Differentially Private Model Aggregation

A key factor in developing high performing machine learning models is the availability of sufficiently large datasets. This work is motivated by applications arising in Software as a Service (SaaS) companies where there exist numerous similar yet disjoint datasets from multiple client companies. To overcome the challenges of insufficient data without explicitly aggregating the clients' datasets due to privacy concerns, one solution is to collect more data for each individual client, another is to privately aggregate information from models trained on each client's data. In this work, two approaches for private model aggregation are proposed that enable the transfer of knowledge from existing models trained on other companies' datasets to a new company with limited labeled data while protecting each client company's underlying individual sensitive information. The two proposed approaches are based on state-of-the-art private learning algorithms: Differentially Private Permutation-based Stochastic Gradient Descent and Approximate Minima Perturbation. We empirically show that by leveraging differentially private techniques, we can enable private model aggregation and augment data utility while providing provable mathematical guarantees on privacy. The proposed methods thus provide significant business value for SaaS companies and their clients, specifically as a solution for the cold-start problem.

cs.LG↗

arXiv2018

Unified Gas-kinetic Wave-Particle methods I: Continuum and Rarefied Gas Flow

The unified gas-kinetic scheme (UGKS) provides a framework for simulating multiscale transport with the updates of both gas distribution function and macroscopic flow variables on the cell size and time step scales. The multiscale dynamics in UGKS is achieved through the coupled particle transport and collision in the particle evolution process within a time step. In this paper, under the UGKS framework, we propose an efficient multiscale unified gas-kinetic wave-particle (UGKWP) method. The gas dynamics in UGKWP method is described by the individual particle movement coupled with the evolution of the probability density function (PDF). During a time step, the trajectories of simulation particles are tracked until collision happens, and the post-collision particles are evolved collectively through the evolution of the corresponding distribution function. The evolution of simulation particles and distribution function is guided by evolution of macroscopic variables. The two descriptions on a gas particle, i.e. wave and particle, switch dynamically with time. A new concept of multiscale multi-efficiency preserving (MMP) method is introduced, and the UGKWP method is shown to be an MMP scheme. The UGKWP method is specially efficient for hypersonic flow simulation in all regimes in comparison with the wave-type discrete ordinate methods, and presents a much lower stochastic noise in the continuum flow regime in comparison with the particle-based Monte Carlo methods. Numerical tests for flows over a wide range of Mach and Knudsen numbers are presented. The examples include mainly the hypersonic flow passing a circular cylinder at Mach numbers $20$ and $30$ and Knudsen numbers $1$ and $10^{-4}$, low speed lid-driven cavity flow, and laminar boundary layer. These results validate the accuracy, efficiency, and multiscale property of UGKWP method.

math.NA↗

arXiv2018

Attentive Tensor Product Learning

This paper proposes a new architecture - Attentive Tensor Product Learning (ATPL) - to represent grammatical structures in deep learning models. ATPL is a new architecture to bridge this gap by exploiting Tensor Product Representations (TPR), a structured neural-symbolic model developed in cognitive science, aiming to integrate deep learning with explicit language structures and rules. The key ideas of ATPL are: 1) unsupervised learning of role-unbinding vectors of words via TPR-based deep neural network; 2) employing attention modules to compute TPR; and 3) integration of TPR with typical deep learning architectures including Long Short-Term Memory (LSTM) and Feedforward Neural Network (FFNN). The novelty of our approach lies in its ability to extract the grammatical structure of a sentence by using role-unbinding vectors, which are obtained in an unsupervised manner. This ATPL approach is applied to 1) image captioning, 2) part of speech (POS) tagging, and 3) constituency parsing of a sentence. Experimental results demonstrate the effectiveness of the proposed approach.

cs.CL↗

arXiv2018

Evolution of Photospheric Vector Magnetic Field Associated with Moving Flare Ribbons As Seen By GST

The photospheric response to solar flares, also known as coronal back reaction, is often observed as sudden flare-induced changes in vector magnetic field and sunspot motions. However, it remains obscure whether evolving flare ribbons, the flare signature closest to the photosphere, are accompanied by changes in vector magnetic field therein. Here we explore the relationship between the dynamics of flare ribbons in the chromosphere and variations of magnetic fields in the underlying photosphere, using high-resolution off-band H-alpha images and near-infrared vector magnetograms of the M6.5 flare on 2015 June 22 observed with the 1.6 m Goode Solar Telescope. We find that changes of photospheric fields occur at the arrival of the flare ribbon front, thus propagating analogously to flare ribbons. In general, the horizontal field increases and the field lines become more inclined to the surface. When ribbons sweep through regions that undergo a rotational motion, the fields transiently turn more vertical with decreased horizontal field and inclination angle, and then restore and/or become more horizontal than before the ribbon arrival. The ribbon propagation decelerates near the sunspot rotation center, where the vertical field becomes permanently enhanced. Similar magnetic field changes are discernible in magnetograms from the Helioseismic and Magnetic Imager (HMI), and an inward collapse of coronal magnetic fields is inferred from the time sequence of non-linear force-free field models extrapolated from HMI magnetograms. We conclude that photospheric fields respond nearly instantaneously to magnetic reconnection in the corona.

astro-ph.SR↗

arXiv2018

Tree-to-tree Neural Networks for Program Translation

Program translation is an important tool to migrate legacy code in one language into an ecosystem built in a different language. In this work, we are the first to employ deep neural networks toward tackling this problem. We observe that program translation is a modular procedure, in which a sub-tree of the source tree is translated into the corresponding target sub-tree at each step. To capture this intuition, we design a tree-to-tree neural network to translate a source tree into a target one. Meanwhile, we develop an attention mechanism for the tree-to-tree model, so that when the decoder expands one non-terminal in the target tree, the attention mechanism locates the corresponding sub-tree in the source tree to guide the expansion of the decoder. We evaluate the program translation capability of our tree-to-tree model against several state-of-the-art approaches. Compared against other neural translation models, we observe that our approach is consistently better than the baselines with a margin of up to 15 points. Further, our approach can improve the previous state-of-the-art program translation approaches by a margin of 20 points on the translation of real-world projects.

cs.AI↗

arXiv2018

The construction and use of LISA sensitivity curves

The Laser Interferometer Space Antenna (LISA) will open the mHz band of the gravitational wave spectrum for exploration. Sensitivity curves are a useful tool for surveying the types of sources that can be detected by the LISA mission. Here we describe how the sensitivity curve is constructed, and how it can be used to compute the signal-to-noise ratio for a wide range of binary systems. We adopt the 2018 LISA Phase-0 reference design parameters. We consider both sky-averaged sensitivities, and the sensitivity to sources at particular sky locations. The calculations are included in a publicly available {\em Python} notebook.

astro-ph.HE↗

arXiv2018

A Unified Gas-kinetic Particle Method for Multiscale Photon Transport

In this work, we present a unified gas-kinetic particle (UGKP) method for the simulation of multiscale photon transport. The multiscale nature of the particle method mainly comes from the recovery of the time evolution flux function in the unified gas-kinetic scheme (UGKS) through a coupled dynamic process of particle transport and collision. This practice improves the original operator splitting approach in the Monte Carlo method, such as the separated treatment of particle transport and collision. As a result, with the variation of the ratio between numerical time step and local photon's collision time, different transport physics can be fully captured in a single computation. In the diffusive limit, the UGKP method could recover the solution of the diffusion equation with the cell size and time step being much larger than the photon's mean free path and the mean collision time. In the free transport limit, it presents an exact particle tracking process as the original Monte Carlo method. In the transition regime, the weights of particle free transport and collision are determined by the ratio of local numerical time step to the photon's collision time. Several one-dimensional numerical examples covering all transport regimes from the optically thin to optically thick are computed to validate the accuracy and efficiency of the current scheme. In comparison with the $S_N$ discrete ordinate method, the UGKP method is based on particles and avoids the discretization of particle velocity space, which does not suffer from the ray effect.

physics.comp-ph↗

arXiv2018

Notes on Scrambling in Conformal Field Theory

The onset of quantum chaos in quantum field theory may be studied using out-of-time-order correlators at finite temperature. Recent work argued that a timescale logarithmic in the central charge emerged in the context of two-dimensional conformal field theories, provided the intermediate channel was dominated by the Virasoro identity block. This suggests a wide class of conformal field theories exhibit a version of fast scrambling. In the present work we study this idea in more detail. We begin by clarifying to what extent correlators of wavepackets built out of superpositions of primary operators may be used to quantify quantum scrambling. Subject to certain caveats, these results concur with previous work. We then go on to study the contribution of intermediate states beyond the Virasoro identity block. We find that at late times, time-ordered correlators exhibit a familiar decoupling theorem, suppressing the contribution of higher dimension operators. However this is no longer true of the out-of-time-order correlators relevant for the discussion of quantum chaos. We compute the contributions of these conformal blocks to the relevant correlators, and find they are able to dominate in many interesting limits. Interpreting these results in the context of holographic models of quantum gravity, sheds new light on the black hole information problem by exhibiting a class of correlators where bulk effective field theory does not predict its own demise.

hep-th↗

arXiv2018

Bayesian inference in high-dimensional linear models using an empirical correlation-adaptive prior

In the context of a high-dimensional linear regression model, we propose the use of an empirical correlation-adaptive prior that makes use of information in the observed predictor variable matrix to adaptively address high collinearity, determining if parameters associated with correlated predictors should be shrunk together or kept apart. Under suitable conditions, we prove that this empirical Bayes posterior concentrates around the true sparse parameter at the optimal rate asymptotically. A simplified version of a shotgun stochastic search algorithm is employed to implement the variable selection procedure, and we show, via simulation experiments across different settings and a real-data application, the favorable performance of the proposed method compared to existing methods.

stat.ME↗

arXiv2018

A Unified Gas-kinetic Scheme for Continuum and Rarefied Flows VI: Dilute Disperse Gas-Particle Multiphase System

In this paper, a unified gas kinetic scheme for multiphase dilute gas-particle system is proposed. The UGKS multiphase (UGKS-M) is a finite volume method, which captures flow physics in the regimes from collisionless multispecies transport to the two-fluid hydrodynamic Navier-Stokes (NS) solution with the variation of Knudsen number, and from granular flow regime to dusty gas dynamics with the variation of Stokes number. The main reason for preserving the multiscale nature in UGKS-M is mainly coming from the direct modeling of the flow physics in the scales of discrete cell size and time step, where the ratio of the time step over the particle collision time determines flow behavior in different regimes. For the particle phase, the integral solution of the kinetic equation is used in the construction of the numerical flux, which takes into account the particle transport, collision, and acceleration. The gas phase, which is assumed to be in the continuum flow regime, evolves numerically by the gas kinetic scheme (GKS), which is a subset of the UGKS for the Navier-Stokes (NS) solutions. The interaction between the gas and particle phase is calculated based on a velocity space mapping method, which solves accurately the kinetic acceleration process. The stability of UGKS-M is determined by the CFL condition only. With the inclusion of the material temperature evolution equation of solid particles, the UGKS-M conserves the total mass, momentum, and energy for the whole multiphase system. In the numerical tests, the UGKS-M shows good multiscale property in capturing the particle trajectory crossing (PTC), particle wall reflecting phenomena, and vortex-induced segregation of inertial particles under different Stokes numbers. The scheme is also applied to simulate shock induced fluidization problem and the simulation results agree well with experimental measurement.

physics.comp-ph↗

arXiv2018

Modified Self-Organized Task Allocation in a Group of Robots

This paper introduces a modified self-organized task allocation algorithm, where robots are assigned to pick up one of the two types of object. This paper also demonstrates both algorithms by showing the simulation results of the conventional self-organized task allocation algorithm and the simulation results of its modification.

cs.RO↗