arXiv subjects

You Wu

Publications and source records attributed to You Wu.

88 records · Page 5Linked to original sources

arXiv2022

A Generalization of Array Codes with Local Properties and Efficient Encoding/Decoding

A maximum distance separable (MDS) array code is composed of $m\times (k+r)$ arrays such that any $k$ out of $k+r$ columns suffice to retrieve all the information symbols. Expanded-Blaum-Roth (EBR) codes and Expanded-Independent-Parity (EIP) codes are two classes of MDS array codes that can repair any one symbol in a column by locally accessing some other symbols within the column, where the number of symbols $m$ in a column is a prime number. By generalizing the constructions of EBR and EIP codes, we propose new MDS array codes, such that any one symbol can be locally recovered and the number of symbols in a column can be not only a prime number but also a power of an odd prime number. Also, we present an efficient encoding/decoding method for the proposed generalized EBR (GEBR) and generalized EIP (GEIP) codes based on the LU factorization of a Vandermonde matrix. We show that the proposed decoding method has less computational complexity than existing methods. Furthermore, we show that the proposed GEBR codes have both a larger minimum symbol distance and a larger recovery ability of erased lines for some parameters when compared to EBR codes. We show that EBR codes can recover any $r$ erased lines of a slope for any parameter $r$, which was an open problem in [2].

cs.IT↗

arXiv2022

RCP: A Low-overhead Reversible Coherence Protocol

This paper proposes RCP, a new reversible coherence protocol that ensures invisible speculative load execution (ISLE) with low overhead. RCP can be combined with processor mechanisms that eliminate the effects of speculative instructions on other instructions to achieve low overhead invisible speculative execution (ISE). ISE provides protection that is at least as strong as speculative privacy tracking (SPT) and stronger than speculative taint tracking (STT). RCP is designed by systematically extending the existing coherence protocol to incorporate speculative loads and states. The protocol is implemented in Gem5 and verified with Murphi. The results show that RCP based ISE incurs lower overhead than STT/SDO/SPT while providing similar/stronger protection.

cs.AR↗

arXiv2022

AssistQ: Affordance-centric Question-driven Task Completion for Egocentric Assistant

A long-standing goal of intelligent assistants such as AR glasses/robots has been to assist users in affordance-centric real-world scenarios, such as "how can I run the microwave for 1 minute?". However, there is still no clear task definition and suitable benchmarks. In this paper, we define a new task called Affordance-centric Question-driven Task Completion, where the AI assistant should learn from instructional videos to provide step-by-step help in the user's view. To support the task, we constructed AssistQ, a new dataset comprising 531 question-answer samples from 100 newly filmed instructional videos. We also developed a novel Question-to-Actions (Q2A) model to address the AQTC task and validate it on the AssistQ dataset. The results show that our model significantly outperforms several VQA-related baselines while still having large room for improvement. We expect our task and dataset to advance Egocentric AI Assistant's development. Our project page is available at: https://showlab.github.io/assistq/.

cs.CV↗

arXiv2022

Identification of new classical Be stars from the LAMOST MRS survey

Be stars are B-type main-sequence stars that display broad Balmer emission lines in their spectra. Identification of Be population is essential to further examine the formation and evolutionary models. We report the detection of classical Be (CBe) stars from observations with the Large sky Area Multi-Object fiber Spectroscopic Telescope Medium Resolution Survey of Date Release 7 (LAMOST MRS DR7). We used a deep convolutional neural network, the ResNet, with an 18-layer module to examine the morphology of the H alpha profile. We identified 1,162 candidate Be stars from the collection of 2,260,387 spectra for 789,918 stars in the database. The ResNet network achieves a Be star classification accuracy of 99.5%. Among the detections, 151 of these are prior known Be stars cross-matched from the literature. By applying a three-step test, we identified 183 new CBe stars. We find that 41 CBe stars are members of known open clusters. Based upon an investigation of the kinematics of the identified CBe stars from the Gaia EDR3 astrometric solutions, we identified 16 new runaways. These new identifications will provide a reference for future follow-ups to further investigate their physical properties.

astro-ph.SR↗

arXiv2021

Autofocusing Self-Imaging: The Symmetric Pearcey Talbot-like Effect

The Talbot like effect of symmetric Pearcey beams (SPBs) is presented numerically and experimentally in the free space. Owing to the Talbot like effect, the SPBs have the property of periodic and multiple autofocusing. Meanwhile, the focal positions and focal times of SPBs are controlled by the beam shift factor and the distribution factors. What is more, the beam shift factor can also affect the Talbot-like effect and the Talbot period. Therefore, several tiny optical bottles can be generated under the appropriate parameter setting. It is believed that the results can diversify the application of the Talbot effect.

physics.optics↗

arXiv2021

Airy-Gaussian vortex beams in the fractional nonlinear-Schrödinger medium

We address the propagation of vortex beams with the circular Airy-Gaussian shape in a (2+1)-dimensional optical waveguide modeled by the fractional nonlinear Schrodinger equation. Systematic analysis of autofocusing of the beams reveals a strongly non-monotonous dependence of the peak intensity in the focal plane on the corresponding Levy index, with a strong maximum at alpha =1.4. Effects of the nonlinearity strength, the ratio of widths of the Airy and Gaussian factors in the input, as well as the beam vorticity, on the autofocusing dynamics are explored. In particular, multiple autofocusing events occur if the nonlinearity is strong enough. Under the action of the azimuthal modulational instability, an axisymmetric beam may split into a set of separating bright spots. In the case of strong fractality (for alpha close to 1), the nonlinear beam self-traps, after the first instance of autofocusing, into a breathing vortical quasi-soliton. Radiation forces induced by the beam field are considered too, and a capture position for a probe nanoparticle is thus identified.

physics.optics↗

arXiv2021

ReasonBERT: Pre-trained to Reason with Distant Supervision

We present ReasonBert, a pre-training method that augments language models with the ability to reason over long-range relations and multiple, possibly hybrid contexts. Unlike existing pre-training methods that only harvest learning signals from local contexts of naturally occurring texts, we propose a generalized notion of distant supervision to automatically connect multiple pieces of text and tables to create pre-training examples that require long-range reasoning. Different types of reasoning are simulated, including intersecting multiple pieces of evidence, bridging from one piece of evidence to another, and detecting unanswerable cases. We conduct a comprehensive evaluation on a variety of extractive question answering datasets ranging from single-hop to multi-hop and from text-only to table-only to hybrid that require various reasoning capabilities and show that ReasonBert achieves remarkable improvement over an array of strong baselines. Few-shot experiments further demonstrate that our pre-training method substantially improves sample efficiency.

cs.CL↗

arXiv2020

TURL: Table Understanding through Representation Learning

Relational tables on the Web store a vast amount of knowledge. Owing to the wealth of such tables, there has been tremendous progress on a variety of tasks in the area of table understanding. However, existing work generally relies on heavily-engineered task-specific features and model architectures. In this paper, we present TURL, a novel framework that introduces the pre-training/fine-tuning paradigm to relational Web tables. During pre-training, our framework learns deep contextualized representations on relational tables in an unsupervised manner. Its universal model design with pre-trained representations can be applied to a wide range of tasks with minimal task-specific fine-tuning. Specifically, we propose a structure-aware Transformer encoder to model the row-column structure of relational tables, and present a new Masked Entity Recovery (MER) objective for pre-training to capture the semantics and knowledge in large-scale unlabeled data. We systematically evaluate TURL with a benchmark consisting of 6 different tasks for table understanding (e.g., relation extraction, cell filling). We show that TURL generalizes well to all tasks and substantially outperforms existing methods in almost all instances.

cs.IR↗

arXiv2020

Generating Representative Headlines for News Stories

Millions of news articles are published online every day, which can be overwhelming for readers to follow. Grouping articles that are reporting the same event into news stories is a common way of assisting readers in their news consumption. However, it remains a challenging research problem to efficiently and effectively generate a representative headline for each story. Automatic summarization of a document set has been studied for decades, while few studies have focused on generating representative headlines for a set of articles. Unlike summaries, which aim to capture most information with least redundancy, headlines aim to capture information jointly shared by the story articles in short length, and exclude information that is too specific to each individual article. In this work, we study the problem of generating representative headlines for news stories. We develop a distant supervision approach to train large-scale generation models without any human annotation. This approach centers on two technical components. First, we propose a multi-level pre-training framework that incorporates massive unlabeled corpus with different quality-vs.-quantity balance at different levels. We show that models trained within this framework outperform those trained with pure human curated corpus. Second, we propose a novel self-voting-based article attention layer to extract salient information shared by multiple articles. We show that models that incorporate this layer are robust to potential noises in news stories and outperform existing baselines with or without noises. We can further enhance our model by incorporating human labels, and we show our distant supervision approach significantly reduces the demand on labeled data.

cs.CL↗

arXiv2020

The merger history of primordial-black-hole binaries

As a candidate of dark matter, primordial black holes (PBHs) have attracted more and more attentions as they could be possible progenitors of the heavy binary black holes (BBHs) observed by LIGO/Virgo. Accurately estimating the merger rate of PBH binaries will be crucial to reconstruct the mass distribution of PBHs. It was pointed out the merger history of PBHs may shift the merger rate distribution depending on the mass function of PBHs. In this paper, we use 10 BBH events from LIGO/Virgo O1 and O2 observing runs to constrain the merger rate distribution of PBHs by accounting the effect of merger history. It is found that the second merger process makes subdominant contribution to the total merger rate, and hence the merger history effect can be safely neglected.

astro-ph.CO↗

arXiv2019

Hot subdwarf B stars with neutron star components II: Binary population synthesis

Context: Subdwarf B stars (sdBs) play a crucial role in stellar evolution, asteroseismology, and far-UV radiation of early-type galaxies, and have been intensively studied with observation and theory. It has theoretically been predicted that sdBs with neutron star (NS) companions exist in the Galaxy, but none have been discovered yet. This remains a puzzle in this field. In a previous study (hereafter Paper I), we have studied the formation channels of sdB+NS binaries from main-sequence (MS) stars plus NS binaries by establishing a model grid, but it is still unclear how these binaries consisting of MS stars and NS binaries came to be in the first place. Aims: We systematically study the formation of sdB+NS binaries from their original zero-age main-sequence progenitors. We bridge the gap left by our previous study in this way. We obtain the statistical population properties of sdB+NS binaries and provide some guidance for observational efforts. Methods: We first used Hurley's rapid binary evolution code BSE to evolve 10^7 primordial binaries to the point where the companions of NS+MS, NS+Hertzsprung gap (HG) star, and NS+Giant Branch (GB) star binaries have just filled their Roche lobes. Next, we injected these binaries into the model grid we developed in Paper I to obtain the properties of the sdB+NS populations. We adopted two prescriptions of NS natal kicks. Different values of common-envelope ejection efficiency were chosen to examine the effect of common-envelope evolution on the results. Conclusions: Most sdB+NS binaries are located in the Galactic disk with small RV semi-amplitudes. SdB+NS binaries with large RV semi-amplitudes are expected to be strong GWR sources, some of which could be detected by LISA in the future.

astro-ph.SR↗

arXiv2018

Formation of hot subdwarf B stars with neutron star components

Binary population synthesis predicts the existence of subdwarf B stars (sdBs) with neutron star (NS) or black hole (BH) companions. We systematically investigate the formation of sdB+NS binaries from binary evolution and aim to obtain some clues for a search for such systems. We started from a series of MS+NS systems and determined the parameter spaces for producing sdB+NS binaries from the stable Roche-lobe overflow (RLOF) channel and from the common envelope (CE) ejection channel. Various NS accretion efficiencies and NS masses were examined to investigate the effects they have. We show the characteristics of the produced sdB+NS systems, such as the mass of components, orbital period, the semi-amplitude of the radial velocity (K), and the spin of the NS component. In the stable RLOF channel, the orbital period of sdB+NS binaries produced in this way ranges from several days to more than 1000 days and moves toward the short-period (~ hr) side with increasing initial MS mass. the sdB+NS systems that result from CE ejection have very short orbital periods and then high values of K (up to 800km s^-1). Such systems are born in very young populations (younger than 0.3 Gyr) and are potential gravitational wave sources that might be resolved by the Laser Interferometer Space Antenna (LISA) in the future. Gravitational wave radiation may again bring them into contact on a timescale of only ~Myr. As a consequence, they are rare and hard to discover. The pulsar signal is likely a feature of sdB+NS systems caused by stable RLOF, and some NS components in sdB binaries may be millisecond pulsars.

astro-ph.SR↗

arXiv2015

$f(T)$ non-linear massive gravity and the cosmic acceleration

Inspired by the $f(R)$ non-linear massive gravity, we propose a new kind of modified gravity model, namely $f(T)$ non-linear massive gravity, by adding the dRGT mass term reformulated in the vierbein formalism, to the $f(T)$ theory. We then investigate the cosmological evolution of $f(T)$ massive gravity, and constrain it by using the latest observational data. We find that it slightly favors a crossing of the phantom divide line from the quintessence-like phase ($w_{de} > -1$) to the phantom-like one ($w_{de} < -1$) as redshift decreases.

gr-qc↗

arXiv2015

Post-Newtonian Approximation of Teleparallel Gravity Coupled with a Scalar Field

We use the parameterized post-Newtonian (PPN) formalism to explore the weak field approximation of teleparallel gravity non-minimally coupling to a scalar field $ϕ$, with arbitrary coupling function $ω(ϕ)$ and potential $V(ϕ)$. We find that all the PPN parameters are identical to general relativity (GR), which makes this class of theories compatible with the Solar System experiments. This feature also makes the theories quite different from the scalar-tensor theories, which might be subject to stringent constraints on the parameter space, or need some screening mechanisms to pass the Solar System experimental constraints.

gr-qc↗

arXiv2014

Vector Condensate and AdS Soliton Instability Induced by a Magnetic Field

We continue to study the holographic p-wave superconductor model in the Einstein-Maxwell-complex vector field theory with a non-minimal coupling between the complex vector field and the Maxwell field. In this paper we work in the AdS soliton background which describes a conformal field theory in the confined phase and focus on the probe approximation. We find that an applied magnetic field can lead to the condensate of the vector field and the AdS soliton instability. As a result, a vortex lattice structure forms in the spatial directions perpendicular to the applied magnetic field. As a comparison, we also discuss the vector condensate in the Einstein-SU(2) Yang-Mills theory and find that in the setup of the present paper, the Einstein-Maxwell-complex vector field model is a generalization of the SU(2) model in the sense that the vector field has a general mass and gyromagnetic ratio.

hep-th↗

arXiv2011

Karman constant and accurate mean flow prediction in a turbulent pipe

The Karman constant κ- widely used in atmospheric science and engineering turbulence modelling, and proposed by Prandtl in 1925 and von Karman in 1930 to describe the mean velocity of a turbulent wall-bounded flow - leads to a logarithmic profile in an overlap region near the wall. For over eighty years, its value was believed to be ~0.41. But more recently, many argue that it is not a constant, because of measured variations in different flows and at different Reynolds numbers (Re). Here, a multi-layer analytic theory is shown to lead to a re-interpretation of κas a global constant for both the overlap region and outer flow, and to yield a new method for its measurement. The newly determined value is 0.45 for both channel and pipe. It is shown that this new κ, together with other wall constants, yields a 99% accuracy in the prediction of mean velocity data at all points in high Re (up to 40 million) pipe flow. The theory also describes finite Re effect, and discovers a transition at the friction Re (i.e. Re_τ)=5000. An accurate model for the prediction of turbulent transport in canonical pipe and channel flows is achieved here, and we propose the model to be valid for a wide class of turbulent flows.

physics.flu-dyn↗