arXiv subjects

Jincheng Yu

Publications and source records attributed to Jincheng Yu.

64 records · Page 4Linked to original sources

arXiv2020

LAMOST Medium-Resolution Spectroscopic Survey (LAMOST-MRS): Scientific goals and survey plan

Since September 2018, LAMOST starts a new 5-year medium-resolution spectroscopic survey (MRS) using bright/gray nights. We present the scientific goals of LAMOST-MRS and propose a near optimistic strategy of the survey. A complete footprint is also provided. Not only the regular medium-resolution survey, but also a time-domain spectroscopic survey is being conducted since 2018 and will be end in 2023. According to the detailed survey plan, we expect that LAMOST-MRS can observe about 2 million stellar spectra with ~7500 and limiting magnitude of around G=15 mag. Moreover, it will also provide about 200 thousand stars with averagely 60-epoch observations and limiting magnitude of G~14 mag. These high quality spectra will give around 20 elemental abundances, rotational velocities, emission line profiles as well as precise radial velocity with uncertainty less than 1 km/s. With these data, we expect that LAMOST can effectively leverage sciences on stellar physics, e.g. exotic binary stars, detailed observation of many types of variable stars etc., planet host stars, emission nebulae, open clusters, young pre-main-sequence stars etc.

astro-ph.SR↗

arXiv2020

On the Chemical and Kinematic Consistency Between N-rich Metal-poor Field Stars and Enriched Populations in Globular Clusters

Interesting chemically peculiar field stars may reflect their stellar evolution history and their possible origin in a different environment from where they are found now, which is one of the most important research fields in Galactic archaeology. To explore this further, we have used the CN-CH bands around 4000 A to identify N-rich metal-poor field stars in LAMOST DR3. Here we expand our N-rich metal-poor field star sample to ~100 stars in LAMOST DR5, where 53 of them are newly found in this work. We investigate light elements of the common stars between our sample and APOGEE DR14. While Mg, Al, and Si abundances generally agree with the hypothesis that N-rich metal-poor field stars come from enriched populations in globular clusters, it is still inconclusive for C, N, and O. After integrating the orbits of our N-rich field stars and a control sample of normal metal-poor field stars, we find that N-rich field stars have different orbital parameter distributions compared to the control sample, specifically, apocentric distances, maximum vertical amplitude (Zmax), orbital energy, and z direction angular momentum (Lz). The orbital parameters of N-rich field stars indicate that most of them are inner-halo stars. The kinematics of N-rich field stars support their possible GC origin. The spatial and velocity distributions of our bona fide N-rich field star sample are important observational evidence to constrain simulations of the origin of these interesting objects.

astro-ph.GA↗

arXiv2018

Hu-Fu: Hardware and Software Collaborative Attack Framework against Neural Networks

Recently, Deep Learning (DL), especially Convolutional Neural Network (CNN), develops rapidly and is applied to many tasks, such as image classification, face recognition, image segmentation, and human detection. Due to its superior performance, DL-based models have a wide range of application in many areas, some of which are extremely safety-critical, e.g. intelligent surveillance and autonomous driving. Due to the latency and privacy problem of cloud computing, embedded accelerators are popular in these safety-critical areas. However, the robustness of the embedded DL system might be harmed by inserting hardware/software Trojans into the accelerator and the neural network model, since the accelerator and deploy tool (or neural network model) are usually provided by third-party companies. Fortunately, inserting hardware Trojans can only achieve inflexible attack, which means that hardware Trojans can easily break down the whole system or exchange two outputs, but can't make CNN recognize unknown pictures as targets. Though inserting software Trojans has more freedom of attack, it often requires tampering input images, which is not easy for attackers. So, in this paper, we propose a hardware-software collaborative attack framework to inject hidden neural network Trojans, which works as a back-door without requiring manipulating input images and is flexible for different scenarios. We test our attack framework for image classification and face recognition tasks, and get attack success rate of 92.6% and 100% on CIFAR10 and YouTube Faces, respectively, while keeping almost the same accuracy as the unattacked model in the normal mode. In addition, we show a specific attack scenario in which a face recognition system is attacked and gives a specific wrong answer.

cs.CR↗

arXiv2018

A Survey of FPGA-Based Neural Network Accelerator

Recent researches on neural network have shown significant advantage in machine learning over traditional algorithms based on handcrafted features and models. Neural network is now widely adopted in regions like image, speech and video recognition. But the high computation and storage complexity of neural network inference poses great difficulty on its application. CPU platforms are hard to offer enough computation capacity. GPU platforms are the first choice for neural network process because of its high computation capacity and easy to use development frameworks. On the other hand, FPGA-based neural network inference accelerator is becoming a research topic. With specifically designed hardware, FPGA is the next possible solution to surpass GPU in speed and energy efficiency. Various FPGA-based accelerator designs have been proposed with software and hardware optimization techniques to achieve high speed and energy efficiency. In this paper, we give an overview of previous work on neural network inference accelerators based on FPGA and summarize the main techniques used. An investigation from software to hardware, from circuit level to system level is carried out to complete analysis of FPGA-based neural network inference accelerator design and serves as a guide to future work.

cs.AR↗

arXiv2018

Stellar Population Properties of Ultracompact Dwarfs in M87: a Mass-metallicity Correlation Connecting Low-metallicity Globular Clusters and Compact Ellipticals

We derive stellar population parameters for a representative sample of ultracompact dwarfs (UCDs) and a large sample of massive globular clusters (GCs) with stellar masses $\gtrsim$ 10$^{6}$ $M_{\odot}$ in the central galaxy M87 of the Virgo galaxy cluster, based on model fitting to the Lick-index measurements from both the literature and new observations. After necessary spectral stacking of the relatively faint objects in our initial sample of 40 UCDs and 118 GCs, we obtain 30 sets of Lick-index measurements for UCDs and 80 for GCs. The M87 UCDs have ages $\gtrsim$ 8 Gyr and [$α$/Fe] $\simeq$ 0.4 dex, in agreement with previous studies based on smaller samples. The literature UCDs, located in lower-density environments than M87, extend to younger ages and smaller [$α$/Fe] (at given metallicities) than M87 UCDs, resembling the environmental dependence of the Virgo dE nuclei. The UCDs exhibit a positive mass-metallicity relation (MZR), which flattens and connects compact ellipticals at stellar masses $\gtrsim$ 10$^{8}$ $M_{\odot}$. The Virgo dE nuclei largely follow the average MZR of UCDs, whereas most of the M87 GCs are offset towards higher metallicities for given stellar masses. The difference between the mass-metallicity distributions of UCDs and GCs may be qualitatively understood as a result of their different physical sizes at birth in a self-enrichment scenario or of galactic nuclear cluster star formation efficiency being relatively low in a tidal stripping scenario for UCD formation. The existing observations provide the necessary but not sufficient evidence for tidally stripped dE nuclei being the dominant contributors to the M87 UCDs.

astro-ph.GA↗

arXiv2017

The age-velocity dispersion relation of the Galactic discs from LAMOST-Gaia data

We present the age-velocity dispersion relation (AVR) in three dimensions in the solar neighbourhood using 3,564 commonly observed sub-giant/red-giant branch stars selected from LAMOST, which gives the age and radial velocity, and \emph{Gaia}, which measures the distance and proper motion. The stars are separated into metal-poor (${\rm [Fe/H]<-0.2}$\,dex and metal-rich (${\rm [Fe/H]>-0.2}$\,dex) groups, so that the metal-rich stars are mostly $α$-poor, while the metal-poor group are mostly contributed by $α$-enhanced stars. Thus, the old and metal-poor stars likely belong to the chemically defined thick disc population, while the metal-rich sample is dominated by the thin disc. The AVR for the metal-poor sample shows an abrupt increase at $\gtrsim7$\,Gyr, which is contributed by the thick disc component. On the other hand, most of the thin disc stars with ${\rm [Fe/H]>-0.2}$\,dex display a power-law like AVR with indices of about 0.3--0.4 and 0.5 for the in-plane and vertical dispersions, respectively. This is consistent with the scenario that the disc is gradually heated by the spiral arms and/or the giant molecular clouds. Moreover, the older thin disc stars ($>7$\,Gyr) have a rounder velocity ellipsoid, i.e. $σ_ϕ/σ_{\rm z}$ is close to 1.0, probably due to the more efficient heating in vertical direction. Particularly for the old metal-poor sample located with $|z|>270$\,pc, the vertical dispersion is even larger than its azimuthal counterpart. Finally, the vertex deviations and the tilt angles are plausibly around zero with large uncertainties.

astro-ph.GA↗

arXiv2017

Simulations of Fractal Star Cluster Formation: I. New Insights for Measuring Mass Segregation of Star Clusters with SubStructure

We compare the existent methods including the minimum spanning tree based method and the local stellar density based method, in measuring mass segregation of star clusters. We find that the minimum spanning tree method reflects more the compactness, which represents the global spatial distribution of massive stars, while the local stellar density method reflects more the crowdedness, which provides the local gravitational potential information. It is suggested to measure the local and the global mass segregation simultaneously. We also develop a hybrid method that takes both aspects into account. This hybrid method balances the local and the global mass segregation in the sense that the predominant one is either caused either by dynamical evolution or purely accidental, especially when such information is unknown a priori. In addition, we test our prescriptions with numerical models and show the impact of binaries in estimating the mass segregation value. As an application, we use these methods on the Orion Nebula Cluster (ONC) observations and the Taurus cluster. We find that the ONC is significantly mass segregated down to the 20th most massive stars. In contrast, the massive stars of the Taurus cluster are sparsely distributed in many different subclusters, showing a low degree of compactness. The massive stars of Taurus are also found to be distributed in the high-density region of the subclusters, showing significant mass segregation at subcluster scales. Meanwhile, we also apply these methods to discuss the possible mechanisms of the dynamical evolution of the simulated substructured star clusters.

astro-ph.GA↗

arXiv2012

The selection of LEGUE disk targets for LAMOST's pilot survey

We describe the target selection algorithm for the low latitude disk portion of the LAMOST Pilot Survey, which aims to test systems in preparation for the LAMOST spectroscopic survey. We use the PPMXL (Roeser et al. 2010) astrometric catalog, which provides positions, proper motions, B/R/I magnitudes (mostly) from USNO-B (Monet et al. 2003) and J/H/Ks from The Two Micron All Sky Survey (2MASS, see Skrutskie et al. 2006) as well. We chose 8 plates along the Galactic plane, in the region $0^\circ<α<67^\circ$ and $42^\circ<δ<59^\circ$, that cover 22 known open clusters with a range of ages. Adjacent plates may have small overlapping. Each plate covers an area $2.5^\circ$ in radius,with central star (for Shack-Hartmann guider) brighter than $\sim8^{\rm th}$ magnitude. For each plate, we create an input catalog in the magnitude range $11.3<Imag<16.3$ and $Bmag$ available from PPMXL. The stars are selected to satisfy the requirements of the fiber positioning system and have a uniform distribution in the $I$ vs. $B-I$ color-magnitude diagram. Our final input catalog consists of 12,000 objects on each of 8 plates that are observable during the winter observing season in Xinglong Station of the National Astronomical Observatory of China.

astro-ph.GA↗

arXiv2012

A Test for Radial Mixing Using Local Star Samples

We use samples of local main-sequence stars to show that the radial gradient of [Fe/H] in the thin disk of the Milky Way decreases with mean effective stellar temperature. Many of these stars are visiting the solar neighborhood from the inner and outer Galaxy. We use the angular momentum of each star about the Galactic center to determine the guiding center radius and to eliminate the effects of epicyclic motion, which would otherwise blur the estimated gradients. We interpret the effective temperature as a proxy for mean age, and conclude that the decreasing gradient is consistent with the predictions of radial mixing due to transient spiral patterns. We find some evidence that the trend of decreasing gradient with increasing mean age breaks to a constant gradient for samples of stars whose main-sequence life-times exceed the likely age of the thin disk.

astro-ph.GA↗

arXiv2011

Rapid dynamical mass segregation and properties of fractal star clusters

We investigate the evolution of young star clusters using N-body simulations. We confirm that subvirial and fractal-structured clusters will dynamically mass segregate on a short timescale (within 0.5 Myr). We adopt a modified minimum-spanning-tree (MST) method to measure the degree of mass segregation, demonstrating that the stars escaping from a cluster's potential are important for the temporal dependence of mass segregation in the cluster. The form of the initial velocity distribution will also affect the degree of mass segregation. If it depends on radius, the outer parts of the cluster would expand without undergoing collapse. In velocity space, we find 'inverse mass segregation,' which indicates that massive stars have higher velocity dispersions than their lower-mass counterparts.

astro-ph.GA↗