基于深度学习技术的恒星大气物理参数自动估计

深度学习是当前机器学习、模式识别和人工智能领域中的一项热点研究技术,非常适用于处理复杂的大规模数据.基于深度学习理论构建了一个5层的栈式自编码深度神经网络,对恒星大气物理参数进行自动估计,网络各层的节点数分别为3821-500-100-50-1.使用美国大型巡天项目Sloan发布的Sloan Digital Sky Survey(SDSS)实测光谱以及由Kurucz的New Opacity Distribution Function(NEWODF)模型得到的理论光谱进行了实验验证,对有效温度(Teff)、表面重力加速度(lg g)和金属丰度([Fe/H])3个物理参数进行了自动估计.结果表明,栈式自编码深度神经网络的估计精度较好,其中在SDSS数据上的平均绝对误差分别为:79.95(Teff/K),0.0058(lg(Teff/K)),0.1706(lg(g/(cm·s~(-2)))),0.1294 dex([Fe/H]);在理论数据上的平均绝对误差分别是:15.34(Teff/K),0.0011(lg(Teff/K)),0.0214(lg(g/(cm·s~(-2)))),0.0121 dex([Fe/H]).     

多任务Lasso回归法在恒星光谱物理参量估计中的应用

多任务学习方法在机器学习、计算机视觉、人工智能领域已得到广泛关注,利用任务间的相关性,将多个任务同时学习的效果优于每个任务单独学习的情况.采用多任务Lasso回归法(Multi-task Lasso Regression)用于恒星光谱物理参量的估计,不仅可以获取不同物理参量间的共同的特征信息,而且也可以很好地保留不同物理参量的特有的补充信息.使用恒星大气模拟模型合成光谱库ELODIE中的光谱数据和美国大型巡天项目Sloan发布的SDSS实测光谱数据进行实验,模型估算精度优于相关文献中的方法,特别是对重力加速度(lg g)和化学丰度([Fe/H])的估计.实验中通过改变光谱的分辨率,施加不同信噪比(SNR)的噪声,来说明模型的稳定性强.结果表明,模型精度受光谱分辨率和噪声的影响,但噪声对其影响更大,可见,多任务Lasso回归法不仅操作简便,稳定性强,而且也提高了模型的整体预测精度.     

低分辨率恒星光谱的[α/Fe]估计方法研究

研究了低分辨率恒星光谱的[α/Fe]估计问题.所给方案包括以下3个步骤:首先,使用Haar小波对原始光谱进行四级分解,去除高频成分,以抑制高频噪声干扰;然后,基于光谱数据成分与[α/Fe]的相关性和LASSO(Least Absolute Shrinkage and Selection Operator)算法选择光谱特征;最后,基于MARCS恒星光谱库和多元线性回归方法对[α/Fe]进行测量.并使用ELODIE、SDSS(Sloan Digital Sky Survey)、LAMOST(Large Sky Area Multi-Object Fibre Spectroscopic Telescope)和4个星团的低分辨率恒星光谱数据验证了方法的有效性.在317个ELODIE光谱样本、412个SDSS光谱样本和1276个LAMOST光谱样本(信噪比SNRG大于20)上的(系统偏差,精度)分别为(0.04dex,0.064 dex)、(0.16 dex,0.065 dex)和(0.05 dex,0.062 dex).在球状星团M13、M15以及疏散星团NGC2420、M67上的估计结果与文献值基本一致.     

Automated estimation of stellar fundamental parameters from low resolution spectra： the PLS method

PLS （Partial Least Squares regression） is introduced into an automatic estimation of fundamental stellar spectral parameters. It extracts the most correlative spectral component to the parameters （Teff, log g and [Fe/H]）, and sets up a linear regression function from spectra to the corresponding parameters. Considering the properties of stellar spectra and the PLS algorithm, we present a piecewise PLS regression method for estimation of stellar parameters, which is composed of one PLS model for Teff, and seven PLS models for log g and [Fe/H] estimation. Its performance is investigated by large experiments on flux calibrated spectra and continuum normalized spectra at different signal-to-noise ratios （SNRs） and resolutions. The results show that the piecewise PLS method is robust for spectra at the medium resolution of 0.23 nm. For low resolution 0.5 nm and 1 nm spectra, it achieves competitive results at higher SNR. Experiments using ELODIE spectra of 0.23 nm resolution illustrate that our piecewise PLS models trained with MILES spectra are efficient for O ～ G stars： for flux calibrated spectra, the systematic offsets are 3.8%, 0.14 dex, and -0.09 dex for Teff, log g and [Fe/H], with error scatters of 5.2%, 0.44 dex and 0.38 dex, respectively; for continuum normalized spectra, the systematic offsets are 3.8%, 0.12dex, and -0.13 dex for Teff, log g and [Fe/H], with error scatters of 5.2%, 0.49 dex and 0.41 dex, respectively. The PLS method is rapid, easy to use and does not rely as strongly on the tightness of a parameter grid of templates to reach high precision as Artificial Neural Networks or minimum distance methods do.     

基于郭守敬望远镜(LAMOST)试观测数据新发现的贫金属星候选体

贫金属星因其内在蕴涵着银河系早期化学元素形成和演化等信息,所以他们对银河系及宇宙早期形成历史的研究具有重要意义。目前对贫金属星的搜寻与研究已成为天文学的一大国际研究热点。随着郭守敬望远镜(LAMOST)试运行的开展及数据的积累,我国已具备自主开展此领域研究工作的有力观测设备及数据资源。通过采用有效的的测量方法,对郭守敬望远镜试观期获得的低分辨率(R≈2 000)恒星光谱数据进行恒星大气物理参数测量,新发现了8颗[Fe/H]-1.00 dex的贫金属星候选体,其中1颗[Fe/H]=-2.73dex属于很贫的贫金属星(VMP)。工作首次显示并证明了郭守敬望远镜能够有效地开展银河系内搜寻大量贫金属星的工作。     

A comparison of stellar atmospheric parameters from the LAMOST and APOGEE datasets

We have compared stellar parameters, including temperature, gravity and metallicity, for common stars in the LAMOST DR2 and SDSS DR12/APOGEE datasets. It is found that the LAMOST dataset provides a more well-defined red clump feature than the APOGEE dataset in the Teff versus log g diagram. With this advantage, we have separated red clump stars from red giant stars, and attempt to establish calibrations between the two datasets for the two groups of stars. The results show that there is a good consistency in temperature with a calibration close to the one-to-one line, and we can establish a satisfactory metallicity calibration of[Fe/H]APOGEE= 1.18[Fe/H]LAMOST + 0.11 with a scatter of ～ 0.08 dex for both the red clump and red giant branch samples. For gravity, there is no correlation for red clump stars between the two datasets, and scatters around the calibrations of red giant stars are substantial. We found two main sources of scatter in log g for red giant stars. One is a group of stars with 0.00253 × Teff- 8.67 log g 2.6 located in the forbidden region, and the other is the contaminated red clump stars, which could be picked out from the unmatched region where stellar metallicity is not consistent with position in the Teff versus log g diagram. After excluding stars in these two regions,we have established two calibrations for red giant stars, log g APOGEE = 0.000615 ×Teff,LAMOST+ 0.697 × log g LAMOST- 2.208(σ = 0.150) for [Fe/H] -1 and log gAPOGEE= 0.000874×Teff,LAMOST+0.588×log g LAMOST-3.117(σ = 0.167)for [Fe/H] -1. The calibrations are valid for stars with Teff = 3800- 5400 K and log g = 0- 3.8 dex, and are useful in work aiming to combine the LAMOST and APOGEE datasets in a future study. In addition, we find that an SVM method based on asteroseismic log g is a good way to greatly improve the accuracy of gravity for these two regions, at least in the LAMOST dataset.     

贫金属星的重元素丰度Ⅱ.细致的光谱分析

本文确定了贫金属矮星和亚巨星的重元素丰度.基于高分辨率高信噪比光谱得到27颗样本星的锶、钇、锆、钡、镧、铈和铕丰度.主要结果为:(1)第一峰区元素锶-钇-锆显示了奇偶效应,在晕族星中钇相对于锶和锆过贫。(2)对[Fe/H]>-2.5的恒星钇丰度为常数([Y/Fe]=-0.06).(3)对于[Fe/H]<-1.7的样本星,随金属丰度减小钡过贫增加·(4)镧和铈相对于铁过贫.(5)在金属丰度-1.0>[Fe/H]>- 2.0 区间,铕表现过富.(6)对[Y/Fe],[Zr/Fe]和[Ba/Fe]存在约0.1dex的本质弥散.     

Seismic diagnostics of solar-like oscillating stars

High precision and long-lasting Kepler data enabled us to estimate stellar properties with asteroseismology as an accurate tool. We performed asteroseismic analysis on six solar-like stars observed by the Kepler mission: KIC 6064910, KIC 6766513, KIC 7107778, KIC 10079226, KIC 10147635 and KIC 12069127. The extraction of seismic information includes two parts. First, we obtained two global asteroseismic parameters, mean large separation ?_ν and frequency of maximum power ν_(max),with autocorrelation function and collapsed autocorrelation function. Second, we extracted individual oscillation modes ν_(nl) with low-l degree using a least-squares fit. Stellar grid models were built with Yale Rotating Stellar Evolution Code(YREC) to analyze stellar properties. They covered the range of M = 0.8 ~ 1.8 M_⊙with a step of 0.02 M_⊙ and [Fe/H] =-0.3 ~ 0.4 dex with a step of 0.1 dex.We used a Bayesian approach to estimate stellar fundamental parameters of the six stars, under the constraints of asteroseismic parameters(?_ν, ν_(max)) and non-asteroseismic parameters(Teff, [Fe/H]). We discover that the six targets include five sub-giant stars with 1.2 ~1.5 M_⊙ and one main-sequence star with 1.08 M_⊙, and with ages in the range of 3 ~5 Gyr.     

LASPM:the LAMOST stellar parameter pipeline for M-type stars and application to the sixth and seventh data release(DR6 and DR7)

The molecular-rich atmospheres of M type stars complicate our understanding to their atmospheric properties.Recently,great progress has been made in atmospheric modeling of M-type stars,and we take advantage of the updated BT-Settl model grid to develop a pipeline LAS PM to measure atmospheric parameters(T_(eff),log g,[M/H]) of M-type stars from low-resolution spectra.The pipeline was applied to the sixth and seventh data release(DR6 DR7) of Large Sky Area Multi-Object Fiber Spectroscopic Telescope(LAMOST),which released atmospheric parameters for 610419 and 680 185 Mtype spectra,respectively.The key algorithm is to find the best-matching for templates in the synthetic spectral library for an observed spectrum,and then minimizing χ~2 through a linear combination of five best-matching templates.The intrinsic precisions of the parameters were estimated by using the multiple epoch observations for the same stars,which are 118 K,0.20 dex,0.29 dex for T_(eff),log g,and [M/H]respectively.The Teff and log g are consistent with the spectral and luminosity classifications by LAMOST 1D pipeline,and the loci of giants and dwarfs both on spectral index and color-magnitude diagrams show the validity.The metallicities of LASPM are also checked with the selected members of four open clusters(NGC 2632,Melotte 22,ASCC_16,and ASCC_19),which are consistent without any bias.Comparing the results between LASPM and the APOGEE Stellar Parameter and Chemical Abundance Pipeline(ASPCAP),there is a scatter of 73 K,0.22 dex,0.21 dex for T_(eff),log g,and [M/H],respectively.     

发射线星系[OⅡ]λ3727/Hα谱线流量比的研究

利用从斯隆数字巡天(Sloan Digital Sky Survey,简称SDSS)第4次释放的光谱数据中选取的10~5个发射线星系样本,研究了[O_Ⅱ]λ3727/Hα流量比与星系尘埃消光、气体电离态和金属丰度的关系.发现尘埃消光改正对[O_Ⅱ]λ3727/Hα谱线流量比影响显著,消光改正前、后的[O_Ⅱ]λ3727/Hα谱线流量比的中值分别为0.48和0.89;尘埃消光改正后,F([O_Ⅱ]λ3727)-F(Hα)的弥散显著减小.贫金属星系的[O_Ⅱ]λ3727/Hα谱线流量比随星系气体的电离度增高而减小,而富金属星系不存在这种关系.另外,[O_Ⅱ]λ3727/Hα流量比与星系金属丰度相关.当12+lg(O/H)8.5时,星系[O_Ⅱ]λ3727/Hα流量比随金属丰度增加而下降;12+lg(O/H)8.5的星系,谱线流量比与金属丰度正相关.最后,利用气体电离度参数和星系的金属丰度,给出了计算不同类型星系[O_Ⅱ]λ3727/Hα流量比的公式.LAMOST望远镜将观测到大量红移z0.4的星系光谱,利用该公式可以给出星系的[O_Ⅱ]λ3727/Hα流量比,从而可以利用[O_Ⅱ]λ3727谱线流量计算z0.4星系的恒星形成率.     

巨早型星系的近红外谱合成

在8000至8670的波长范围内,利用一个由144颗恒星光谱组成的光谱厍,我们对57个巨椭圆和S0星系(M_B<-21)的光谱和它们的平均光谱进行了光谱合成.这个光谐库包括了光谱型为G,K和M的巨星和矮星,它具有的金属丰度[Fe/H]覆盖了-0.1到-0.5的范围,表面重力log g为1.0至5.0.光谱合成的结果表明:巨椭圆和S0星系的金属丰度约为太阳的一倍半;有效表面重力分布在3.2—4.1的范围内;矮星在8400附近光的贡献可与巨星相比较.     

Estimating stellar atmospheric parameters based on Lasso features

With the rapid development of large scale sky surveys like the Sloan Digital Sky Survey （SDSS）, GAIA and LAMOST （Guoshoujing telescope）, stellar spectra can be obtained on an ever-increasing scale. Therefore, it is necessary to estimate stel- lar atmospheric parameters such as Teff, log g and [Fe/H] automatically to achieve the scientific goals and make full use of the potential value of these observations. Feature selection plays a key role in the automatic measurement of atmospheric parameters. We propose to use the least absolute shrinkage selection operator （Lasso） algorithm to select features from stellar spectra. Feature selection can reduce redundancy in spectra, alleviate the influence of noise, improve calculation speed and enhance the robustness of the estimation system. Based on the extracted features, stellar atmospheric param- eters are estimated by the support vector regression model. Three typical schemes are evaluated on spectral data from both the ELODIE library and SDSS. Experimental results show the potential performance to a certain degree. In addition, results show that our method is stable when applied to different spectra.     

SAGE巡天介绍Ⅰ——测光系统和数据处理

SAGE(Stellar Abundance and Galactic Evolution)巡天采用一套对恒星大气参数敏感的独特测光系统,包括uSC,vSAGE,g,r,i,Hα_n,Hα_w和λ_(DDO51)(David Dunlap Observatory,DDO)共8个波段。该系统对恒星大气参数(有效温度T_(eff)、表面重力lg g和金属丰度[Fe/H])和星际消光的测量比传统的uvbyβ系统(Str?mgren-Crawford,SC)更敏感。由于g,r,i波段带宽较宽,因而观测效率较高,对恒星大气参数的测量精度和观测效率均超过uvbyβ系统和宽带测光系统。利用美国亚利桑那大学Steward天文台2.3 m Bok望远镜、新疆天文台南山1 m望远镜和乌兹别克斯坦MAO(Maidanak Astronomical Observatory)1 m望远镜已展开巡天观测。项目计划在约4年时间内完成所有的巡天观测,北天天区覆盖面积约12000(~?)~2,高精度测光(信噪比为100σ)深度约15 mag(V波段);5σ探测极限可深至约20 mag(V波段)。最终将探测约5×108颗恒星,并得出其恒星大气参数以及星际消光。该样本比之前的u,v,b,y,β测光巡天,如GCS(Geneva-Copenhagen Survey)巡天和HM(Hauck and Mermilliod 1998)巡天,扩展到更深的区域,即深度达到7～8 mag(V波段)。利用观测得到的大样本恒星测光星表,可以开展如下课题的研究:(1)得到银河系约5×10~8颗恒星大样本的金属丰度分布,并根据恒星演化模型得到其年龄分布;(2)预期搜寻到一批贫金属星候选体,并利用其他望远镜进行下一步观测;(3)结合高质量的uSC波段数据,预期可以找到一批白矮星;(4)结合其他星表,预期找到一批长周期变源以及移动天体;(5)结合uSC波段数据和其他星表,预期可以找到一批类星体。     

GD-1星流金属丰度与运动特性分析

星流在星系形成与演化过程中扮演了重要的角色,对银河系中星流的研究将有助于进一步探究银河系的合并历史.将LAMOST(Large Sky Area Multi-Object Fiber Spectroscopic Telescope)DR6光谱数据以及SDSS(Sloan Digital Sky Survey)DR12光谱数据分别与Gaia(Global Astrometric Interferometer for Astrophysics)DR2天体测量数据交叉匹配,获得恒星自行等数据.对GD-1星流在速度空间、几何空间和金属丰度上进行限制,从LAMOST DR6和SDSS DR12数据中共获得了157颗星流成员星.GD-1星流的平均金属丰度为[Fe/H]=-2.16±0.10 dex,延伸长度超过80°.收集前人给出的GD-1星流高概率成员星,组成较大的成员星样本进行对比分析,发现GD-1星流的金属丰度分布呈现内低外高的特点,沿着星流方向径向速度分布特点是两端大、中间小,?₁=-20°(?₁为GD-1星流坐标系横坐标)和?₁=-60°附近的间隙是因为成员星运动差异形成的.根据成员星分布及其速度分布特性,推测GD-1星流起源位置是在?₁=-40°附近.     

Exploring the spectral information content in the LAMOST medium-resolution survey(MRS)

Although high-resolution stellar spectra allow us to derive precise stellar labels(effective temperature, metallicity, surface gravity, elemental abundances, etc.) based on resolved atomic lines and molecular bands, low-resolution spectra have been proved to be competitive in determining many stellar labels at comparable precision. It is useful to consider the spectral information content when assessing the capability of a stellar spectrum in deriving precise stellar labels. In this work, we quantify the information content brought by the LAMOST-II medium-resolution spectroscopic survey(MRS) using the gradient spectra as well as the coefficients-of-dependence(CODs). In general, the wavelength coverage of the MRS well constrains the stellar labels but the sensitivities of different stellar labels vary with spectral types and metallicity of the stars of interest. Consequently, this affects the performance of the stellar label determination from the MRS spectra. By applying the SLAM method to the synthetic spectra which mimic the MRS data, we find that the precision of the fundamental stellar parameters Teff, log g and [M/H] are better when combining both the blue and red bands of the MRS. This is especially important for warm stars because the Hα line located in the red part plays a more important role in determining the effective temperature for warm stars. With blue and red parts together, we are able to reach similar performance to the low-resolution spectra except for warm stars. However, at [M/H] ~-2.0 dex, the uncertainties of fundamental stellar labels estimated from MRS are substantially larger than that from low-resolution spectra. We also tested the uncertainties of T_(eff), log g and [M/H] from MRS data induced from the radial velocity mismatch and find that a mismatch of about 1 km s~(-1), which is typical for LAMOST MRS data, would not significantly affect the stellar label estimates. Finally, reference precision limits are calculated using synthetic gradient spectra, according to which we expect abundances of at least 17 elements to be measured precisely from MRS spectra.     

Investigation of the shortest period Am type eclipsing binary TYC 6408–989–1

The first BV bands photometric observations and the low-resolution spectrum of the shortest period Am type eclipsing binary TYC 6408-989-1 have been obtained.The stellar atmospheric parameters of the primary star were obtained through the spectral fitting as follows:Teff=6990±117 K,log g=4.25±0.26 cm s^-2,[Fe/H]=-0.45±0.03 dex.The original spectra obtained by European Southern Observatory(ESO)were processed with an IRAF package by us.Based on the ESO blue-violet spectra,TYC 6408-989-1 was concluded as a marginal Am(Am:)star with a spectral type of kA3 hF1 mA5Ⅳ-Ⅴidentified through the MKCLASS program.The observed light curves were analyzed through the WilsonDevinney code.The final photometric solutions show that TYC 6408-989-1 is a marginal contact binary with a low mass ratio(q=0.27).The temperature of the secondary component derived through the light curve analysis is significantly higher than main sequence stars.In addition,TYC 6408-989-1 is a poor thermal contact binary.The temperature differences between the two components is about 1800 K.TYC6408-989-1 should be located in the oscillation stage predicted by the thermal relaxation oscillations theory(TRO)and will evolve into the shallow contact stage eventually.The very short period(less than one day),marginal Am peculiarity and quite large rotational velocity(v sin i■160 km s^-1)make TYC 6408-989-1 become a challenge to the cut-off of rotation velocities and periods of Am stars.We have collected the well known eclipsing Am binaries with absolute parameters from the literature.