arXiv Science⌕ Search

arXiv subjects

Chao Liu

Publications and source records attributed to Chao Liu.

At least 361 records · Page 20Linked to original sources

On the lower bounds of Davenport constant

Let $G = C_{n_1} \oplus \cdots \oplus C_{n_r}$ with $1 < n_1 | \cdots | n_r$ be a finite abelian group. The Davenport constant $\mathsf D(G)$ is the smallest integer $t$ such that every sequence $S$ over $G$ of length $|S|\ge t$ has a non-empty zero-sum subsequence. It is a starting point of zero-sum theory but only has a trivial lower bound $\mathsf D^*(G) = n_1 + \cdots + n_r - r + 1$, which equals $\mathsf D(G)$ over $p$-groups. We investigate the non-dispersive sequences over group $C_n^r$, thereby revealing the growth of $\mathsf D(G)-\mathsf D^*(G)$ over non-$p$-groups $G = C_n^r \oplus C_{kn}$ with $n,k \ne 1$. We give a general lower bound of $\mathsf D(G)$ over non-$p$-groups and show that, let $G$ be abelian groups with $\exp(G)=m$ and rank $r$, fix $m>0$ a non-prime-power, then for each $N>0$ there exists an $\varepsilon>0$ such that if $|G|/m^r<\varepsilon $, then $\mathsf D(G)-\mathsf D^*(G)>N$.

math.CO↗

Deriving the stellar labels of LAMOST spectra with Stellar LAbel Machine (SLAM)

The LAMOST survey has provided 9 million spectra in its Data Release 5 (DR5) at R$\sim$1800. Extracting precise stellar labels is crucial for such a large sample. In this paper, we report the implementation of the Stellar LAbel Machine (SLAM), which is a data-driven method based on Support Vector Regression (SVR), a robust non-linear regression technique. Thanks to the capability to model highly non-linear problems with SVR, SLAM generally can derive stellar labels over a wide range of spectral types. This gives it a unique capability compared to other popular data-driven methods. To illustrate this capability, we test the performance of SLAM on stars ranging from Teff$\sim$4000 to $\sim$8000 K trained on LAMOST spectra and stellar labels. At g-band signal-to-noise ratio (SNRg) higher than 100, the random uncertainties of Teff, logg and [Fe/H] are 50 K, 0.09 dex, and 0.07 dex, respectively. We then set up another SLAM model trained by APOGEE and LAMOST common stars to demonstrate its capability of dealing with high dimensional problems. The spectra are from LAMOST DR5 and the stellar labels of the training set are from APOGEE DR15, including Teff, logg, [M/H],[$α$/M], [C/M], and [N/M]. The cross-validated scatters at SNRg$\sim$100 are 49 K, 0.10 dex, 0.037 dex,0.026 dex, 0.058 dex, and 0.106 dex for these parameters, respectively. This performance is at the same level as other up-to-date data-driven models. As a byproduct, we also provide the latest catalog of $\sim$1 million LAMOST DR5 K giant stars with SLAM-predicted stellar labels in this work.

astro-ph.SR↗

MIOpen: An Open Source Library For Deep Learning Primitives

Deep Learning has established itself to be a common occurrence in the business lexicon. The unprecedented success of deep learning in recent years can be attributed to: abundance of data, availability of gargantuan compute capabilities offered by GPUs, and adoption of open-source philosophy by the researchers and industry. Deep neural networks can be decomposed into a series of different operators. MIOpen, AMD's open-source deep learning primitives library for GPUs, provides highly optimized implementations of such operators, shielding researchers from internal implementation details and hence, accelerating the time to discovery. This paper introduces MIOpen and provides details about the internal workings of the library and supported features. MIOpen innovates on several fronts, such as implementing fusion to optimize for memory bandwidth and GPU launch overheads, providing an auto-tuning infrastructure to overcome the large design space of problem configurations, and implementing different algorithms to optimize convolutions for different filter and input sizes. MIOpen is one of the first libraries to publicly support the bfloat16 data-type for convolutions, allowing efficient training at lower precision without the loss of accuracy.

cs.LG↗

Salient Instance Segmentation via Subitizing and Clustering

The goal of salient region detection is to identify the regions of an image that attract the most attention. Many methods have achieved state-of-the-art performance levels on this task. Recently, salient instance segmentation has become an even more challenging task than traditional salient region detection; however, few of the existing methods have concentrated on this underexplored problem. Unlike the existing methods, which usually employ object proposals to roughly count and locate object instances, our method applies salient objects subitizing to predict an accurate number of instances for salient instance segmentation. In this paper, we propose a multitask densely connected neural network (MDNN) to segment salient instances in an image. In contrast to existing approaches, our framework is proposal-free and category-independent. The MDNN contains two parallel branches: the first is a densely connected subitizing network (DSN) used for subitizing prediction; the second is a densely connected fully convolutional network (DFCN) used for salient region detection. The MDNN simultaneously outputs saliency maps and salient object subitizing. Then, an adaptive deep feature-based spectral clustering operation segments the salient regions into instances based on the subitizing and saliency maps. The experimental results on both salient region detection and salient instance segmentation datasets demonstrate the satisfactory performance of our framework. Notably, its APr@0.5 and Apr@0.7 reaches 73.46% and 60.14% in the salient instance dataset, substantially higher than the results achieved by the state-of-the-art algorithm.

cs.CV↗

Tracing Kinematic and Chemical Properties of Sagittarius Stream by K-Giants, M-Giants, and BHB stars

We characterize the kinematic and chemical properties of $\sim$3,000 Sagittarius (Sgr) stream stars, including K-giants, M-giants, and BHBs, select from SEGUE-2, LAMOST, and SDSS separately in Integrals-of-Motion space. The orbit of Sgr stream is quite clear from the velocity vector in $X$-$Z$ plane. Stars traced by K-giants and M-giants present the apogalacticon of trailing steam is $\sim$ 100 kpc. The metallicity distributions of Sgr K-, M-giants, and BHBs present that the M-giants are on average the most metal-rich population, followed by K-giants and BHBs. All of the K-, M-giants, and BHBs indicate that the trailing arm is on average more metal-rich than leading arm, and the K-giants show that the Sgr debris is the most metal-poor part. The $α$-abundance of Sgr stars exhibits a similar trend with the Galactic halo stars at lower metallicity ([Fe/H] $<\sim$ $-$1.0 dex), and then evolve down to lower [$α$/Fe] than disk stars at higher metallicity, which is close to the evolution pattern of $α$-element of Milky Way dwarf galaxies. We find $V_Y$ and metallicity of K-giants have gradients along the direction of line-of-sight from the Galactic center in $X$-$Z$ plane, and the K-giants show that $V_Y$ increases with metallicity at [Fe/H] $>\sim-$1.5 dex. After dividing the Sgr stream into bright and faint stream according to their locations in equatorial coordinate, the K-giants and BHBs show that the bright and faint stream present different $V_Y$ and metallicities, the bright stream is on average higher in $V_Y$ and metallicity than the faint stream.

astro-ph.GA↗

High-resolution Infrared Spectrograph for Exoplanet Characterization with the Keck and Thirty Meter Telescopes

HISPEC (High-resolution Infrared Spectrograph for Exoplanet Characterization) is a proposed diffraction-limited spectrograph for the W.M. Keck Observatory, and a pathfinder for the MODHIS facility project (Multi-Object Diffraction-limited High-resolution Infrared Spectrograph) on the Thirty Meter Telescope. HISPEC/MODHIS builds on diffraction-limited spectrograph designs which rely on adaptively corrected single-mode fiber feeds. Seeing-limited high-resolution spectrographs, by virtue of the conservation of beam etendue, grow in volume following a D^3 power law (D is the telescope diameter), and are subject to daunting challenges associated with their large size. Diffraction-limited spectrographs fed by single mode fibers are decoupled from the telescope input, and are orders of magnitude more compact and have intrinsically stable line spread functions. Their efficiency is directly proportional to the performance of the adaptive optics (AO) system. AO technologies have matured rapidly over the past two decades and are baselined for future extremely large telescopes. HISPEC/MODHIS will take R>100,000 spectra of a few objects in a 10" field-of-view sampled at the diffraction limit (~10-50 mas), simultaneously from 0.95 to 2.4 microns (y-K). The scientific scope ranges from exoplanet infrared precision radial velocities, spectroscopy of transiting, close-in, and directly imaged exoplanets (atmospheric composition and dynamics, RM effect, spin measurements, Doppler imaging), brown dwarf characterization, stellar physics/chemistry, proto-planetary disk kinematics/composition, Solar system, extragalactic science, and cosmology. HISPEC/MODHIS features a compact, cost-effective design optimized to fully exploit the existing Keck-AO and future TMT-NFIRAOS infrastructures and boost the scientific reach of Keck Observatory and TMT soon after first light.

astro-ph.IM↗

Smoking Gun of the Dynamical Processing of the Solar-type Field Binary Stars

We investigate the binarity properties in field stars using more than 50\,000 main-sequence stars with stellar mass from 0.4 to 0.85\,$M_\odot$ observed by LAMOST and {\emph Gaia} in the solar neighborhood. By adopting a power-law shape of the mass-ratio distribution with power index of $γ$, we conduct a hierarchical Bayesian model to derive the binary fraction ($f_{b}$) and $γ$ for stellar populations with different metallicities and primary masses ($m_1$). We find that $f_b$ is tightly anti-correlated with $γ$, i.e. the populations with smaller binary fraction contains more binaries with larger mass-ratio and vice versa.The high-$γ$ populations with $γ>1.2$ have lower stellar mass and higher metallicity, while the low-$γ$ populations with $γ<1.2$ have larger mass or lower metallicity. The $f_b$ of the high-$γ$ group is anti-correlated with [Fe/H] but flat with $m_1$.Meanwhile, the $f_b$ of the low-$γ$ group displays clear correlation with $m_1$ but quite flat with [Fe/H]. The substantial differences are likely due to the dynamical processing when the binaries were in the embedded star clusters in their early days. The dynamical processing tends to destroy binaries with smaller primary mass, smaller mass-ratio, and wider separation. Consequently, the high-$γ$ group containing smaller $m_1$ is more effectively influenced and hence contains less binaries, many of which have larger mass-ratio and shorter period. However, the low-$γ$ group is less affected by the dynamical processing due to their larger $m_1$. These are evident that the dynamical processing does effectively work and significantly reshape the present-day binary properties of field stars.

astro-ph.SR↗

DELTA: A DEep learning based Language Technology plAtform

In this paper we present DELTA, a deep learning based language technology platform. DELTA is an end-to-end platform designed to solve industry level natural language and speech processing problems. It integrates most popular neural network models for training as well as comprehensive deployment tools for production. DELTA aims to provide easy and fast experiences for using, deploying, and developing natural language processing and speech models for both academia and industry use cases. We demonstrate the reliable performance with DELTA on several natural language processing and speech tasks, including text classification, named entity recognition, natural language inference, speech recognition, speaker verification, etc. DELTA has been used for developing several state-of-the-art algorithms for publications and delivering real production to serve millions of users.

cs.CL↗

Revealing the Complicated Story of the Cetus Stream with StarGO

We use a novel cluster identification tool StarGO to explore the metal poor ([Fe/H] $<$ -1.5) outer stellar halo (d $>$ 15 kpc) of the Milky Way using data from Gaia, LAMOST and SDSS. Our method is built using an unsupervised learning algorithm, a self-organizing map, which trains a 2-D neural network to learn the topological structures of a data set from an n-D input space. Using a 4-D space of angular momentum and orbital energy, we identify three distinct groups corresponding to the Sagittarius, Orphan, and Cetus Streams. For the first time we are able to discover a northern counterpart to the Cetus stream. We test the robustness of this new detection using mock data and find that the significance is more than 5-sigma. We also find that the existing southern counterpart bifurcates into two clumps with different radial velocities. By exploiting the visualization power of StarGO, we attach MW globular clusters to the same trained neural network. The Sagittarius stream is found to have five related clusters, confirming recent literature studies, and the Cetus stream has one associated cluster, NGC 5824. This latter association has previously been postulated, but can only now be truly confirmed thanks to the high-precision Gaia proper motions and large numbers of stellar spectra from LAMOST. The large metallicity dispersion of the stream indicates that the progenitor cannot be a globular cluster. Given the mean metallicity of the stream, we propose that the stream is the result of a merger of a low-mass dwarf galaxy that hosted a large nuclear star cluster (NGC 5824).

astro-ph.GA↗

EASYFLOW: Keep Ethereum Away From Overflow

While Ethereum smart contracts enabled a wide range of blockchain applications, they are extremely vulnerable to different forms of security attacks. Due to the fact that transactions to smart contracts commonly involve cryptocurrency transfer, any successful attacks can lead to money loss or even financial disorder. In this paper, we focus on the overflow attacks in Ethereum , mainly because they widely rooted in many smart contracts and comparatively easy to exploit. We have developed EASYFLOW , an overflow detector at Ethereum Virtual Machine level. The key insight behind EASYFLOW is a taint analysis based tracking technique to analyze the propagation of involved taints. Specifically, EASYFLOW can not only divide smart contracts into safe contracts, manifested overflows, well-protected overflows and potential overflows, but also automatically generate transactions to trigger potential overflows. In our preliminary evaluation, EASYFLOW managed to find potentially vulnerable Ethereum contracts with little runtime overhead.

cs.SE↗

Identifying Galactic Halo Substructure in 6D Phase-space Using $\sim$13,000 LAMOST K Giants

We construct a large halo K-giant sample by combining the positions, distances, radial velocities, and metallicities of over 13,000 LAMOST DR5 halo K giants with the Gaia DR2 proper motions, which covers a Galactocentric distance range of 5-120 kpc. Using a position-velocity clustering estimator (the 6Distance), we statistically quantify the presence of position-velocity substructure at high significance: K giants have more close pairs in position-velocity space than a smooth stellar halo. We find that the amount of substructure in the halo increases with increasing distance and metallicity. With a percolation algorithm named friends-of-friends (FoF) to identify groups, we identify members belonging to Sagittarius (Sgr) Streams, Monoceros Ring, Virgo overdensity, Hercules-Aquila Cloud, Orphan Streams and other unknown substructures and find that the Sgr streams account for a large part of grouped stars beyond 20 kpc and enhance the increase of substructure with distance and metallicity. For the first time, we identify spectroscopic members of Monoceros Ring in the south and north Galactic hemisphere, which presents a rotation of about 185 km s^{-1} and mean metallicity is -0.66 dex.

astro-ph.GA↗

Detecting the Sagittarius stream with LAMOST DR4 M giants \& Gaia DR2

We use LAMOST DR4 M giants combined with Gaia DR2 proper motions and ALLWISE photometry to obtain an extremely pure sample of Sagittarius (Sgr) stream stars. Using TiO5 and CaH spectral indices as an indicator, we selected out a large sample of M giant stars from M dwarf stars in LAMOST DR4 spectra. Considering the position, distance, proper motion and the angular momentum distribution, we obtained 164 pure Sgr stream stars. We find that the trailing arm has higher energy than the leading arm in same angular momentum. The trailing arm we detected extends to a heliocentric distance of $\sim 130$ kpc at $\tildeΛ_{\odot}\sim 170^{\circ}$, which is consistent with the feature found in RR Lyrae in \citet{sgr2017}. Both of these detections of Sgr, in M giants and in RR Lyrae, imply that the Sgr stream may contain multiple stellar populations with a broad metallicity range.

astro-ph.GA↗

SDSS-IV MaNGA: Environmental dependence of the Mgb/ -sigma_* relation for nearby galaxies

We use a sample of ~3000 galaxies from the MaNGA MPL-7 internal data release to study the alpha abundance distribution within low-redshift galaxies. We use the Lick index ratio Mgb/ as an alpha abundance indicator to study relationships between the alpha abundance distribution and galaxy properties such as effective stellar velocity dispersion within 0.3 effective radii (sigma_*), galaxy environment, and dark matter halo formation time (z_f). We find that (1) all galaxies show a tight correlation between Mgb/ and sigma_*; (2) `old' (H_beta < 3) low-sigma_* galaxies in high local density environment and inner regions within galaxy groups are enhanced in Mgb/ , while `young' (H_beta>3) galaxies and high-mass galaxies show no or less environmental dependence; (3) `old' galaxies with high-z_f show enhanced Mgb/ over low- and medium-z_f; (4) Mgb/ gradients are close to zero and show dependence on sigma_* but no obvious dependence on the environment or z_f. Our study indicates that stellar velocity dispersion or galaxy mass is the main parameter driving the Mgb/ enhancement, although environments appear to have modest effects, particularly for low- and medium-mass galaxies.

astro-ph.GA↗

A Catalogue of OB Stars from LAMOST Spectroscopic Survey

We present 22\,901 OB spectra of 16\,032 stars identified from LAMOST DR5 dataset. A larger sample of OB candidates are firstly selected from the distributions in the spectral line indices space. Then all 22\,901 OB spectra are identified by manual inspection. Based on a sub-sample validation, we find that the completeness of the OB spectra reaches about $89\pm22$\% for the stars with spectral type earlier than B7, while around $57\pm16$\% B8--B9 stars are identified. The smaller completeness for late B stars is lead to the difficulty to discriminate them from A0--A1 type stars. The sub-classes of the OB samples are determined using the software package MKCLASS. With a careful validation using 646 sub-samples, we find that MKCLASS can give fairly reliable sub-types and luminosity class for most of the OB stars. The uncertainty of the spectral sub-type is around 1 sub-type and the uncertainty of the luminosity class is around 1 level. However, about 40\% of the OB stars are failed to be assigned to any class by MKCLASS and a few spectra are significantly misclassified by MKCLASS. This is likely because that the template spectra of MKCLASS are selected from nearby stars in the solar neighborhood, while the OB stars in this work are mostly located in the outer disk and may have lower metallicity. The rotation of the OB stars may also be responsible for the mis-classifications. Moreover, we find that the spectral and luminosity classes of the OB stars located in the Galactic latitude larger than 20$^\circ$ are substantially different with those located in latitude smaller than 20$^\circ$, which may either due to the observational selection effect or hint a different origin of the high Galactic latitude OB stars.

astro-ph.SR↗

An intuitive 3D map of the Galactic warp's precession traced by classical Cepheids

The Milky Way's neutral hydrogen (HI) disk is warped and flared. However, a dearth of accurate HI-based distances has thus far prevented the development of an accurate Galactic disk model. Moreover, the extent to which our Galaxy's stellar and gas disk morphologies are mutually consistent is also unclear. Classical Cepheids, primary distance indicators with distance accuracies of 3-5%, offer a unique opportunity to develop an intuitive and accurate three-dimensional picture. Here, we establish a robust Galactic disk model based on 1339 classical Cepheids. We provide strong evidence that the warp's line of nodes is not oriented in the Galactic Center-Sun direction. Instead, it subtends a mean angle of 17.5 \pm 1 (formal) \pm 3 (systematic) and exhibits a leading spiral pattern. Our Galaxy thus follows Briggs' rule for spiral galaxies, which suggests that the origin of the warp is associated with torques forced by the massive inner disk. The stellar disk traced by Cepheids follows the gas disk in terms of their amplitudes; the stellar disk extends to at least 20 kpc. This morphology provides a crucial, updated map for studies of the kinematics and archaeology of the Galactic disk.

astro-ph.GA↗

Ripple patterns in in-plane velocities of OB stars from LAMOST and Gaia

With about 12 000 OB type stars selected from the LAMOST and Gaia survey, we study their 3 dimensional velocity distribution over the range of galactocentric radius from 6 to 15 kpc in the Galactic disk plane. A clear ripple pattern in the radial velocity ($V_R$) map is shown. The median $V_R$ reaches $-8$ km s$^{-1}$ at $R\sim9$ kpc, then increases to $\sim0$ km s$^{-1}$ at $R\sim12$ kpc, and later declines to below $-10$ km s$^{-1}$ beyond $R\sim13$ kpc. The median azimuthal velocity ($V_ϕ$) map shows a similar pattern but has roughly $1/4$ phase difference with the radial velocity. Although the ripple of negative $V_R$ at $\sim9$ kpc extends to about 40$^\circ$ in the azimuth angle, it does not align with either the Local or the Perseus spiral arm. Moreover, the farther ripple beyond 13 kpc does not match the Outer spiral arm either. This indicates that the non-axisymmetric kinematic features are not induced by perturbations of known spiral structures. The central rotating bar can not lead to such patterns in the outer disk either. External perturbation of a dwarf galaxy or a dark matter sub-halo can induce such patterns but requires more evidence from both observations and simulations. The $V_ϕ$ map in the $Z$--$V_Z$ plane of the OB stars is also investigated. Despite asymmetry to some degree, no spiral pattern is found. This is reasonable since most of the OB stars have ages much younger than 100 Myrs, which is smaller than one orbital period around the Galactic center.

astro-ph.GA↗

Frequent flare events on the short-period M-type eclipsing binary BX Tri

We present long-term, multi-color photometric monitoring and spectroscopic observations of the short-period M-type eclipsing binary BX Tri. Six flare events were recorded in 4 years from 2014 to 2017. Three of them were detected on one night within an orbital cycle. The strongest one was identified on December 23, 2014. With the amplitudes of $Δ$B=0.48 mag, $Δ$V=0.28 mag, $Δ$R=0.10 mag, $Δ$I=0.02 mag, the total energy due to the flare event was measured to be 4.08 ($\pm$0.24) $\times$$10^{34}$ erg, exceeding the superflare level ($10^{34}$). Based on the observations, the evolutionary status of the binary system as well as the long-term orbital period changes were analyzed. It reveals that BX Tri is probably a semi-detached system with the primary component filling its Roche lobe. The extremely high occurrence of flare events of the binary could be related to the rapid mass transfer between components.

astro-ph.SR↗

Neural RGB->D Sensing: Depth and Uncertainty from a Video Camera

Depth sensing is crucial for 3D reconstruction and scene understanding. Active depth sensors provide dense metric measurements, but often suffer from limitations such as restricted operating ranges, low spatial resolution, sensor interference, and high power consumption. In this paper, we propose a deep learning (DL) method to estimate per-pixel depth and its uncertainty continuously from a monocular video stream, with the goal of effectively turning an RGB camera into an RGB-D camera. Unlike prior DL-based methods, we estimate a depth probability distribution for each pixel rather than a single depth value, leading to an estimate of a 3D depth probability volume for each input frame. These depth probability volumes are accumulated over time under a Bayesian filtering framework as more incoming frames are processed sequentially, which effectively reduces depth uncertainty and improves accuracy, robustness, and temporal stability. Compared to prior work, the proposed approach achieves more accurate and stable results, and generalizes better to new datasets. Experimental results also show the output of our approach can be directly fed into classical RGB-D based 3D scanning methods for 3D scene reconstruction.

cs.CV↗