星族合成方法的研究及应用

发布时间：2018-04-10 07:07

本文选题：星系　切入点：演化-方法　出处：《中国科学技术大学》2013年博士论文

【摘要】：星系的光谱中包含了大量和其物理性质相关的信息,如何快速有效的分析星系的观测光谱并获得大样本星系的相关参数,对我们研究星系的形成和演化至关重要。随着现代巡天的迅速发展,星系的光谱数据到达了前所未有的数量,尤其是斯隆数字巡天更是带来了数百万的高质量星系光谱。在这些高质量数据增加的同时,也需要我们能发展出快速有效的数据自动处理方法。这篇论文的主要内容包括利用盲源分离方法,实现对大样本光谱数据的快速拟合分析,以及利用我们的光谱分析方法,研究极亮红外星系的演化。论文首先介绍了几种不同的盲源分离方法,包括主成分分析(PCA),独立成分分析(ICA)和基于贝叶斯理论的概率独立成分分析(PICA)。我们详细描述PICA中朴素平均场算法(EL-ICA)和高级平均场(MF-ICA)算法。相比于其他几种盲源分离方法,MF-ICA方法更准确,计算过程更快速,因此我们选择MF-ICA方法用于星系光谱分析。论文第三章介绍了演化的星族合成模型,总结了模型的主要参数：恒星形成历史,初始质量函数,恒星演化轨迹和恒星光谱库。演化的星族合成模型可以生成完备的简单星族(SSPs)光谱库,从库中挑选出若干个SSP作为模板,结合矩阵求逆的方法,可以实现对星系观测光谱的拟合和星族性质相关参数的推断。本文第四章把MF-ICA方法应用于星系的光谱分析中,实现了对星系观测光谱的快速拟合和星族成分的分析,并完成了对DR8所有光谱数据的分析。我们利用MF-ICA方法将星族合成模型的SSP光谱库压缩成极少量的特征成分,利用这些特征成分,完成对星系光谱的重构。相比于将星系光谱分解为若干个SSP的传统矩阵求逆方法,MF-ICA的应用极大的提高了光谱拟合效率和相关参数的确定。我们采用了三种不同的方法来检验我们的光谱分析方法的准确性和可靠性：1)拟合模型星系；2)对比前人基于不同方法得到的拟合结果；3)拟合SDSS光谱数据,分析估算出的参数之间的关系。结果表明,我们的方法不仅能很好的拟合出星系的光谱,同时也能准确的估算出星系的相关参数,比如恒星成分、恒星形成历史、星光红化以及恒星速度弥散。考虑到光谱分辨率和信噪比对拟合效果的影响,我们也对DEEP2光谱数据进行初步拟合,结果表明我们的方法对信噪比较低的DEEP2光谱仍然适用。最后,我们利用基于MF-ICA的光谱分析方法,完成了对极亮红外星系(ULIRGs)的星族分析,并对其演化进行了研究。通过WISE数据和SDSS光谱数据的交叉匹配,我们最终确定了419个窄线ULIRGs样本。根据测量的发射线性质,将这些ULIRGs分类为HⅡ-like ULIRGs, Seyfert2ULIRGs和Composite ULIRGSo我们利用基于MF-ICA的光谱分析方法,对这些ULIRGs的光谱进行了拟合,并分析这三种不同类型ULIRGs的星族性质,包括星族比例、恒星年龄和恒星质量,同时研究了它们之间的演化关系。结果表明,对于有无AGN成分的ULIRGs,它们之间的星族成分没有明显的区别。我们的分析结果认为这三种不同类型的ULIRGs之间可能不存在演化关系。
[Abstract]:The spectra of galaxies contain a large number of related and the physical nature of the information, how to fast and effective observation of spectral analysis of galaxies and the relevant parameters of Galaxy sample, study the formation and evolution of galaxies is very important to us. With the rapid development of modern survey, spectral data of galaxies to hitherto unknown quantities, especially Sloan digital the sky is high quality spectra of galaxies brought millions. The increase in these high quality data at the same time, we also need to develop a rapid and effective method for automatic data processing. The main contents of this thesis include the method of blind source separation to achieve fast fitting of the large sample spectral data analysis, and the use of our spectrum the analysis method, the evolution of ultraluminous infrared galaxies.
This paper firstly introduces several different BSS methods, including principal component analysis (PCA), independent component analysis (ICA) and probabilistic independent component analysis based on Bayesian theory (PICA). We have a detailed description of PICA in the naive mean field algorithm (EL-ICA) and mean field (MF-ICA) compared to several other algorithms. Blind source separation method, MF-ICA method is more accurate, the calculation process more quickly, so we choose the MF-ICA method for the analysis of spectra of galaxies.
The third chapter introduces the evolution of the stellar population synthesis model, summarizes the main parameters of the model: the star formation history, initial mass function, stellar evolution tracks and stellar spectral libraries. Population synthesis model evolution of the simple stellar population generating complete (SSPs) spectral library, from the library to pick out a number of SSP as a template method the inverse matrix, and can realize the fitting parameters on the properties of stellar Galaxy observation spectrum extrapolation.
In the fourth chapter, the spectral analysis method is applied to the MF-ICA galaxy, realize the rapid analysis and fitting of observed spectra of stellar populations of galaxies, and completed the analysis of all DR8 spectral data. We use the MF-ICA method to the SSP spectral library of stellar population synthesis models compressed into characteristic components in very small amounts, using these feature components and reconstruction of Galaxy spectra. Compared with the traditional matrix decomposition of spectra of galaxies for several methods of SSP inverse, the application of MF-ICA can greatly improve the efficiency and to determine the spectral fitting parameters. The accuracy and reliability we used three different methods to test the spectral analysis of our method: 1) fitting model of galaxy; 2) compared with previous fitting results of different methods based on SDSS; 3) fitting the spectral data, the relationship between the parameters of the estimated. The results show that our party Not only can fit the spectra of galaxies is very good, but also can accurately estimate the parameters of galaxies, such as stellar component, star formation history, star reddening and stellar velocity dispersion. Considering the effects of spectral resolution and signal-to-noise ratio of the fitting effect, we also carried out a preliminary fitting of the DEEP2 spectra data, results we show that the DEEP2 spectral method for low SNR are still applicable.
Finally, we use the MF-ICA analysis method based on spectrum, the completion of the ultraluminous infrared Galaxy (ULIRGs) of the population analysis, and its evolution is studied. Through the cross WISE data and SDSS spectral data, we finally identified 419 narrow line ULIRGs sample. According to the nature of the emission line measurements, these ULIRGs classification H II -like ULIRGs Seyfert2ULIRGs and Composite ULIRGSo, we use MF-ICA analysis method based on spectrum, spectrum of these ULIRGs fitting, and the analysis of stellar population properties of these three different types of ULIRGs, including the population proportion, stellar age and stellar mass, and study the evolution of the relationship between them. The results showed that for non AGN components ULIRGs, stellar populations have no obvious difference between them. Our results show that evolution may not exist between these three different types of ULIRGs Relationship.

【学位授予单位】：中国科学技术大学
【学位级别】：博士
【学位授予年份】：2013
【分类号】：P15

【相似文献】