arXiv subjects

Yue Yu

Publications and source records attributed to Yue Yu.

At least 433 records · Page 24Linked to original sources

arXiv2019

Double Quantization for Communication-Efficient Distributed Optimization

Modern distributed training of machine learning models suffers from high communication overhead for synchronizing stochastic gradients and model parameters. In this paper, to reduce the communication complexity, we propose \emph{double quantization}, a general scheme for quantizing both model parameters and gradients. Three communication-efficient algorithms are proposed under this general scheme. Specifically, (i) we propose a low-precision algorithm AsyLPG with asynchronous parallelism, (ii) we explore integrating gradient sparsification with double quantization and develop Sparse-AsyLPG, (iii) we show that double quantization can also be accelerated by momentum technique and design accelerated AsyLPG. We establish rigorous performance guarantees for the algorithms, and conduct experiments on a multi-server test-bed to demonstrate that our algorithms can effectively save transmitted bits without performance degradation.

math.OC↗

arXiv2019

DAG-GNN: DAG Structure Learning with Graph Neural Networks

Learning a faithful directed acyclic graph (DAG) from samples of a joint distribution is a challenging combinatorial problem, owing to the intractable search space superexponential in the number of graph nodes. A recent breakthrough formulates the problem as a continuous optimization with a structural constraint that ensures acyclicity (Zheng et al., 2018). The authors apply the approach to the linear structural equation model (SEM) and the least-squares loss function that are statistically well justified but nevertheless limited. Motivated by the widespread success of deep learning that is capable of capturing complex nonlinear mappings, in this work we propose a deep generative model and apply a variant of the structural constraint to learn the DAG. At the heart of the generative model is a variational autoencoder parameterized by a novel graph neural network architecture, which we coin DAG-GNN. In addition to the richer capacity, an advantage of the proposed model is that it naturally handles discrete variables as well as vector-valued ones. We demonstrate that on synthetic data sets, the proposed method learns more accurate graphs for nonlinearly generated samples; and on benchmark data sets with discrete variables, the learned graphs are reasonably close to the global optima. The code is available at \url{https://github.com/fishmoon1234/DAG-GNN}.

cs.LG↗

arXiv2019

Stochastic Bregman Parallel Direction Method of Multipliers for Distributed Optimization

Bregman parallel direction method of multipliers (BPDMM) efficiently solves distributed optimization over a network, which arises in a wide spectrum of collaborative multi-agent learning applications. In this paper, we generalize BPDMM to stochastic BPDMM, where each iteration only solves local optimization on a randomly selected subset of nodes rather than all the nodes in the network. Such generalization reduce the need for computational resources and allows applications to larger scale networks. We establish both the global convergence and the $O(1/T)$ iteration complexity of stochastic BPDMM. We demonstrate our results via numerical examples.

math.OC↗

arXiv2019

Tolling for Constraint Satisfaction in Markov Decision Process Congestion Games

Markov decision process (MDP) congestion game is an extension of classic congestion games, where a continuous population of selfish agents solves Markov decision processes with congestion: the payoff of a strategy decreases as more population uses it. We draw parallels between key concepts from capacitated congestion games and MDP. In particular, we show that population mass constraints in MDP congestion games are equivalent to imposing tolls/incentives on the reward function, which can be utilized by social planners to achieve auxiliary objectives. We demonstrate such methods in a simulated Seattle ride-share model, where tolls and incentives are enforced for two separate objectives: to guarantee minimum driver density in downtown Seattle, and to shift the game equilibrium towards a maximum social output.

cs.GT↗

arXiv2019

Exploring Fast and Communication-Efficient Algorithms in Large-scale Distributed Networks

The communication overhead has become a significant bottleneck in data-parallel network with the increasing of model size and data samples. In this work, we propose a new algorithm LPC-SVRG with quantized gradients and its acceleration ALPC-SVRG to effectively reduce the communication complexity while maintaining the same convergence as the unquantized algorithms. Specifically, we formulate the heuristic gradient clipping technique within the quantization scheme and show that unbiased quantization methods in related works [3, 33, 38] are special cases of ours. We introduce double sampling in the accelerated algorithm ALPC-SVRG to fully combine the gradients of full-precision and low-precision, and then achieve acceleration with fewer communication overhead. Our analysis focuses on the nonsmooth composite problem, which makes our algorithms more general. The experiments on linear models and deep neural networks validate the effectiveness of algorithms.

math.OC↗

arXiv2019

Fragile superconductivity in the presence of weakly disordered charge density waves

When superconducting (SC) and charge-density wave (CDW) orders compete, novel low temperature behaviors can result. From an analysis of the Landau-Ginzberg-Wilson theory of competing orders, we demonstrate the generic occurrence of a `fragile' SC phase at low temperatures and high fields in the presence of weak disorder. Here, the SC order is largely concentrated in the vicinity of dilute dislocations in the CDW order, leading to transition temperatures and critical currents that are parametrically smaller than those characterizing the zero field SC phase. This may provide the outline of an explanation of the recently discovered `resilient' superconducting phase at high fields in underdoped YBa$_2$Cu$_3$O$_{6+δ}$.

cond-mat.supr-con↗

arXiv2019

Effect of strain inhomogeneity on a chiral p-wave superconductor

Motivated by recent measurements of strain effects on the transition temperature (T$_c$) of Sr$_2$RuO$_4$, we study the strain response of a two-dimensional chiral p-wave supercoductor. We focus on the effects of inhomgeneous strain fields, which are always present in any such experiment, and which have been neglected in previous theoretical treatments. We show that the response of T$_c$ of a chiral superconductor to strain chages from being linear without inhomogeneity, to quadratic in the presence of inhomogeneity. We discuss our results in the context of the ongoing debate of superconductivity in Sr$_2$RuO$_4$.

cond-mat.supr-con↗

arXiv2019

Anomalous magnetic transport and extra quantum oscillation in semi-metallic photon-like fermion gas

In the absence of Lorentz symmetry, the pseudospin-1 counterpart of Weyl fermion (feroton) with linear dispersions and an exact flat band can emerge in condensed matter systems. The flat band branch of feroton is equivalent to the longitudinal photon in Maxwell theory, which is a redundant degree of freedom due to the emergent (fermionic) gauge symmetry. Upon coupling to an external magnetic field, the fermionic guage symmetry is broken and the flat band ferotons become gapless excitations characterized by Landau level indices ($n>1$). In the long wave length limit, these gapless modes are of the opposite chirality to the chiral anomaly related zero Landau level, which leads to much more plentiful magnetic transport properties. To further explore the novel properties of these gapless modes, we investigate the quantum oscillation through a generalized Lieb lattice model. We find an extra oscillating behavior which indicates the existence of these exotic gapless modes. We collect known {\it ab initio} calculation data from literature and discuss the possibility of realizing the semi-metallic feroton gas in real materials.

cond-mat.mes-hall↗

arXiv2018

Field Induced Oscillation of Two Majorana Modes in a Quantum Ring

We calculate the topological boundary modes found for quantum wire for a quantum ring. For the symmetric ring we find analytical solutions for two quasi-particles identifiable as the Majorana states and for asymmetric ring, we have also find approximate solution. By turning on a small field at time zero, we find a field induced oscillation of the Majorana states that was initially localized. The range of validity of our solution, the relation between the period of oscillation and the coin parameter and field strength, and the encoding of a message of the on/off of the field using the Majorana states are discussed. Our results suggest ways to reduce decoherence for small systems, thereby providing a good candidate for quantum memory and computing as the Majorana qubit is topologically protected.

quant-ph↗

arXiv2018

An Initial Step Towards Organ Transplantation Based on GitHub Repository

Organ transplantation, which is the utilization of codes directly related to some specific functionalities to complete ones own program, provides more convenience for developers than traditional component reuse. However, recent techniques are challenged with the lack of organs for transplantation. Hence, we conduct an empirical study on extracting organs from GitHub repository to explore transplantation based on large-scale dataset. We analyze statistics from 12 representative GitHub projects and get the conclusion that 1) there are abundant practical organs existing in commits with add as a key word in the comments; 2) organs in this repository mainly possess four kinds of contents; 3) approximately 70% of the organs are easy-to-transplant. Implementing our transplantation strategy for different kinds of organs, we manually extract 30 organs in three different programming languages, namely Java, Python, and C, and make unit tests for them utilizing four testing tools (two for Java, one for Python, and one for C). At last, we transplant three Java organs into a specific platform for a performance check to verify whether they can work well in the new system. All the 30 organs extracted by our strategy possess good performances in unit test with the highest passing rate reaching 97% and the lowest one still passing 80% and the three Java organs work well in the new system, providing three new functionalities for the host. All the results indicate the feasibility of organ transplantation based on open-source repository, bringing new idea for code reuse.

cs.SE↗

arXiv2018

Reference Map Technique for Incompressible Fluid-Structure Interaction

We present a general simulation approach for fluid-solid interactions based on the fully-Eulerian Reference Map Technique (RMT). The approach permits the modeling of one or more finitely-deformable continuum solid bodies interacting with a fluid and with each other. A key advantage of this approach is its ease of use, as the solid and fluid are discretized on the same fixed grid, which greatly simplifies the coupling between the phases. We use the method to study a number of illustrative examples involving an incompressible Navier-Stokes fluid interacting with multiple neo-Hookean solids. Our method has several useful features including the ability to model solids with sharp corners and the ability to model actuated solids. The latter permits the simulation of active media such as swimmers, which we demonstrate. The method is validated favorably in the flag-flapping geometry, for which a number of experimental, numerical, and analytical studies have been performed. We extend the flapping analysis beyond the thin-flag limit, revealing an additional destabilization mechanism to induce flapping.

physics.flu-dyn↗

arXiv2018

Residual spin susceptibility in the spin-triplet, orbital-singlet model

Nuclear magnetic resonance (NMR) and Knight shift measurements are critical tools in the identification of spin-triplet superconductors. We discuss the effects of spin orbit coupling on the Knight shift and susceptibilities for a variety of spin triplet multi-orbital gap functions with orbital-singlet character and compare their responses to "traditional" single band spin-triplet ($p_x+ip_y$) superconductors. We observe a non-negligible residual spin-susceptibility at low temperature.

cond-mat.supr-con↗

arXiv2018

Extended Coulomb liquid of paired hardcore boson model on a pyrochlore lattice

There is a growing interest in the $U(1)$ Coulomb liquid in both quantum materials in pyrochlore ice and cluster Mott insulators and cold atom systems. We explore a paired hardcore boson model on a pyrochlore lattice. This model is equivalent to the XYZ spin model that was proposed for rare-earth pyrochlores with "dipole-octupole" doublets. Since this model has no sign problem for quantum Monte Carlo (QMC) simulations in a large parameter regime, we carry out both analytical and QMC calculations. We find that the $U(1)$ Coulomb liquid is quite stable and spans a rather large portion of the phase diagram with boson pairing. Moreover, we numerically find thermodynamic evidence that the boson pairing could induce a possible $\mathbb{Z}_2$ liquid in the vicinity of the phase boundary between Coulomb liquid and $\mathbb{Z}_2$ symmetry-broken phase. Besides the materials' relevance with quantum spin ice, we point to quantum simulation with cold atoms on optical lattices.

cond-mat.str-el↗

arXiv2018

Bregman Parallel Direction Method of Multipliers for Distributed Optimization via Mirror Averaging

Distributed optimization aims to optimize a global objective formed by a sum of coupled local convex functions over a graph via only local computation and communication. In this paper, we propose the Bregman parallel direction method of multipliers (PDMM) based on a generalized averaging step named mirror averaging. We establish the global convergence and $O(1/T)$ convergence rate of the Bregman PDMM, along with its $O(n/\ln n)$ improvement over existing PDMM, where $T$ denotes the number of iterations and $n$ the dimension of solution variable. In addition, we can enhance its performance by optimizing the spectral gap of the averaging matrix. We demonstrate our results via a numerical example.

math.OC↗

arXiv2018

Aurora: Providing Trusted System Services for Enclaves On an Untrusted System

Intel SGX provisions shielded executions for security-sensitive computation, but lacks support for trusted system services (TSS), such as clock, network and filesystem. This makes \textit{enclaves} vulnerable to Iago attacks~\cite{DBLP:conf/asplos/CheckowayS13} in the face of a powerful malicious system. To mitigate this problem, we present Aurora, a novel architecture that provides TSSes via a secure channel between enclaves and devices on top of an untrusted system, and implement two types of TSSes, i.e. clock and end-to-end network. We evaluate our solution by porting SQLite and OpenSSL into Aurora, experimental results show that SQLite benefits from a \textit{microsecond} accuracy trusted clock and OpenSSL gains end-to-end secure network with about 1ms overhead.

cs.CR↗

arXiv2018

Optimal Cooperative Inference

Cooperative transmission of data fosters rapid accumulation of knowledge by efficiently combining experiences across learners. Although well studied in human learning and increasingly in machine learning, we lack formal frameworks through which we may reason about the benefits and limitations of cooperative inference. We present such a framework. We introduce novel indices for measuring the effectiveness of probabilistic and cooperative information transmission. We relate our indices to the well-known Teaching Dimension in deterministic settings. We prove conditions under which optimal cooperative inference can be achieved, including a representation theorem that constrains the form of inductive biases for learners optimized for cooperative inference. We conclude by demonstrating how these principles may inform the design of machine learning algorithms and discuss implications for human and machine learning.

cs.LG↗

arXiv2018

An asymptotically compatible meshfree quadrature rule for non-local problems with applications to peridynamics

We present a meshfree quadrature rule for compactly supported non-local integro-differential equations (IDEs) with radial kernels. We apply this rule to develop a strong-form meshfree discretization of a peridynamic solid mechanics model that requires no background mesh. Existing discretizations of peridynamic models have been shown to exhibit a lack of asymptotic compatibility to the corresponding linearly elastic local solution. By posing the quadrature rule as an equality constrained least squares problem, we obtain asymptotically compatible convergence via reproducability constraints. Our approach naturally handles traction-free conditions, surface effects, and damage modeling for both static and dynamic problems. We demonstrate high-order convergence to the local theory by comparing to manufactured solutions and to cases with crack singularities for which an analytic solution is available. Finally, we verify the applicability of the approach to realistic problems by reproducing high-velocity impact results from the Kalthoff-Winkler experiments.

math.NA↗

arXiv2017

Majorana fermions in three dimensions and realization in critical Weyl semimetals

We present two band models for free fermion with charge conjugation symmetry in three dimensions. Without time reversal symmetry (TRS), the weak pairing gapless $A$-phase is a Majorana fermion $p_x+ip_y$ wave FFLO state while the strong pairing gapped $B$-phase belongs to topologically trivial Class $D$. With TRS, there is a Majorana fermion $B$-phase belonging to Class $DIII$ with a non-zero Hopf invariant. The TRS $A$-phase is also a Majorana fermion FFLO state with TRS. The surface states of the TRS $B$-phase are either a valley-momentum locked Majorana-Dirac cone or a linear-quadratic mixed cone for a specific surface. The surface states of the $A$-phase on one surface are topologically nontrivial, either having $\mathbb{Z}$ or $\mathbb{Z}_2$ invariant depending on whether the system is TRS or not. The edge states of that surface are gapless Majorana modes. The Majorana fermion gapless FFLO states can be realized in critical Weyl semimetals (WSM) in which dual single Weyl nodes form dipoles and are nearly annihilated. The gapped $B$-phase emerges when Weyl node dipoles are about to be created. The WSM TaAs-family, a type-II WSM series Mo$_x$W$_{1-x}$Te$_2$-family, possible WSM La/LuBi$_{1-x}$ Sb$_x$Te$_3$ and topological crystalline insulators Sn$_{1-x}$Pb$_x$(Te,Se) are candidates to be manipulated into these critical states based on Majorana fermion models.

cond-mat.mes-hall↗