Yong-Heng Zhao

Comparison of decision tree methods for finding active objects

Abstract The automated classification of objects from large catalogs or survey projects is an important task in many astronomical surveys. Faced with various classification algorithms, astronomers should select the method according to their requirements. Here we describe several kinds of decision trees for finding active objects by multi-wavelength data, such as REPTree, Random Tree, Decision Stump, Random Forest, J48, NBTree, AdTree. All decision tree approaches investigated are in the WEKA package. The classification performance of the methods is presented. In the process of classification by decision tree methods, the classification rules are easily obtained, moreover these rules are clear and easy to understand for astronomers. As a result, astronomers are inclined to prefer and apply them, thus know which attributes are important to discriminate celestial objects. The experimental results show that when various decision trees are applied in discriminating active objects (quasars, BL Lac objects and active galaxies) from non-active objects (stars and galaxies), ADTree is the best only in terms of accuracy, Decision Stump is the best only considering speed, J48 is the optimal choice considering both accuracy and speed.

The Large Sky Area Multi-Object Fiber Spectroscopic Telescope (LAMOST)

The Large Sky Area Multi-Object Fiber Spectroscopic Telescope (LAMOST, also called the Guo Shou Jing Telescope) is a special reflecting Schmidt telescope. LAMOST’s special design allows both a large aperture (effective aperture of 3.6 m–4.9 m) and a wide field of view (FOV) (5 ° ). It has an innovative active reflecting Schmidt configuration which continuously changes the mirror’s surface that adjusts during the observation process and combines thin deformable mirror active optics with segmented active optics. Its primary mirror (6.67 m×6.05 m) and active Schmidt mirror (5.74 m×4.40 m) are both segmented, and composed of 37 and 24 hexagonal sub-mirrors respectively. By using a parallel controllable fiber positioning technique, the focal surface of 1.75 m in diameter can accommodate 4000 optical fibers. Also, LAMOST has 16 spectrographs with 32 CCD cameras. LAMOST will be the telescope with the highest rate of spectral acquisition. As a national large scientific project, the LAMOST project was formally proposed in 1996, and approved by the Chinese government in 1997. The construction started in 2001, was completed in 2008 and passed the official acceptance in June 2009. The LAMOST pilot survey was started in October 2011 and the spectroscopic survey will launch in September 2012. Up to now, LAMOST has released more than 480 000 spectra of objects. LAMOST will make an important contribution to the study of the large-scale structure of the Universe, structure and evolution of the Galaxy, and cross-identification of multiwaveband properties in celestial objects.

LAMOST spectral survey--An overview

LAMOST (Large sky Area Multi-Object fiber Spectroscopic Telescope) is a Chinese national scientific research facility operated by National Astronomical Observatories, Chinese Academy of Sciences (NAOC). After two years of commissioning beginning in 2009, the telescope, instruments, software systems and operations are nearly ready to begin the main science survey. Through a spectral survey of millions of objects in much of the northern sky, LAMOST will enable research in a number of contemporary cutting edge topics in astrophysics, such as discovery of the first generation stars in the Galaxy, pinning down the formation and evolution history of galaxies — especially the Milky Way and its central massive black hole, and looking for signatures of the distribution of dark matter and possible sub-structures in the Milky Way halo. To maximize the scientific potential of the facility, wide national participation and international collaboration have been emphasized. The survey has two major components: the LAMOST ExtraGAlactic Survey (LEGAS) and the LAMOST Experiment for Galactic Understanding and Exploration (LEGUE). Until LAMOST reaches its full capability, the LEGUE portion of the survey will use the available observing time, starting in 2012. An overview of the LAMOST project and the survey that will be carried out in the next five to six years is presented in this paper. The science plan for the whole LEGUE survey, instrumental specifications, site conditions, and the descriptions of the current on-going pilot survey, including its footprints and target selection algorithm, will be presented as separate papers in this volume.

Automated clustering algorithms for classification of astronomical objects

Data mining is an important and challenging problem for the efficient analysis of large astronomical databases and will become even more important with the development of the Global Virtual Observatory. In this study, learning vector quantization (LVQ), single-layer perceptron (SLP) and support vector machines (SVM) were used for multi-wavelength data classification. A feature selection technique was used to evaluate the significance of the considered features for the results of classification. We conclude that in the situation of fewer features, LVQ and SLP show better performance. In contrast, SVM shows better performance when considering more features. The focus of the automatic classification is on the development of an efficient feature-based classifier. The classifiers trained by these methods can be used to preselect AGN candidates.

Automatic determination of stellar atmospheric parameters and construction of stellar spectral templates of the Guoshoujing telescope (LAMOST)

A number of spectroscopic surveys have been carried out or are planned to study the origin of the Milky Way. Their exploitation requires reliable automated methods and softwares to measure the fundamental parameters of the stars. Adopting the ULySS package, we have tested the effect of different resolutions and signal-to-noise ratios (SNR) on the measurement of the stellar atmospheric parameters (effective temperature Teff, surface gravity log g, and metallicity [Fe/H]). We show that ULySS is reliable for determining these parameters with medium-resolution spectra (R ~ 2000). Then, we applied the method to measure the parameters of 771 stars selected in the commissioning database of the Guoshoujing Telescope (LAMOST). The results were compared with the SDSS/SEGUE Stellar Parameter Pipeline (SSPP), and we derived precisions of 167 K, 0.34 dex, and 0.16 dex for Teff, log g and [Fe/H] respectively. Furthermore, 120 of these stars are selected to construct the primary stellar spectral template library (Version 1.0) of LAMOST, and will be deployed as basic ingredients for the LAMOST automated parametrization pipeline.

The correlations between the spin frequencies and kHz QPOs of neutron stars in LMXBs

Aims. We studied the correlations between spin frequencies and kilohertz quasi-periodic oscillations ( kHz QPOs) in neutron star low-mass X-ray binaries. Methods. The updated data on kHz QPOs and spin frequencies are statistically analyzed. Results. We find that when two simultaneous kHz QPOs are present in the power spectrum, the minimum frequency of upper kHz QPO is at least 1.3 times higher than the spin frequency, i. e. nu(s) = -( 0.19 +/- 0.05)nu(s) + ( 389.40 +/- 21.67) Hz. If we shift this correlation in the direction of the peak separation by a factor of 1.5, this correlation matches the data points of the two accretionpowered millisecond X- ray pulsars, SAX J1808.4- 3658 and XTE J1807-294.

Classification in Multidimensional Parameter Space: Methods and Examples

We put forward two classification algorithms, support vector machines (SVM) and learning vector quantization (LVQ), to study the distribution of various astronomical sources in the multidimensional parameter space. By positional cross-correlation, the multiwavelength data of 1656 active galactic nuclei (AGNs), 3718 stars, and 173 galaxies are obtained from optical (USNO-A2.0), X-ray (ROSAT), and infrared (Two Micron All-Sky Survey) bands. We have applied principal component analysis (PCA) to the sample, unveiling correlations between parameters and reducing the dimensionality of the input parameter space. Then, taking the preprocessed data of PCA as input, we apply SVM and LVQ to classify stars, AGNs, and normal galaxies and compare their performances. From the classified results, we conclude that PCA+LVQ and PCA+SVM are effective methods to classify sources with multiwavelength data; moreover, the two methods gave comparable results in a number of situations. Generally, PCA+SVM gave better results; however, PCA+LVQ was considerably faster in terms of computation time. What is more, the classifiers derived by these methods can be used to preselect candidates for large surveys, reducing time and energy wasted. Therefore, the efficiency of high-cost telescopes will be improved. In addition, these methods will be useful to develop the toolkits of the International Virtual Observatory.

THE DIFFERENT NATURE OF SEYFERT 2 GALAXIES WITH AND WITHOUT HIDDEN BROAD-LINE REGIONS

We compile a large sample of 120 Seyfert 2 galaxies (Sy2s) which contains 49 hidden broad-line region (HBLR) Sy2s and 71 non-HBLR Sy2s. From the difference in the power sources between two groups, we test whether HBLR Sy2s are dominated by active galactic nuclei (AGNs) and whether non-HBLR Sy2s are dominated by starbursts. We show that (1) HBLR Sy2s have larger accretion rates than non-HBLR Sy2s; (2) HBLR Sy2s have larger [Ne V] λ14.32/[Ne II] λ12.81 and [O IV] λ25.89/[Ne II] λ12.81 line ratios than non-HBLR Sy2s; and (3) HBLR Sy2s have smaller IRAS f 60/f 25 flux ratios, which show the relative strength of the host galaxy and nuclear emission, than non-HBLR Sy2s. Consequently, we suggest that HBLR Sy2s and non-HBLR Sy2s are AGN dominated and starburst dominated, respectively. In addition, non-HBLR Sy2s can be classified into luminous (L [O III]>1041 erg s–1) and less luminous (L [O III] < 1041 erg s–1) samples, when considering only their obscuration. We suggest that (1) the invisibility of polarized broad lines (PBLs) in the luminous non-HBLR Sy2s depends on the obscuration and (2) the invisibility of PBLs in the less luminous non-HBLR Sy2s depends on the very low Eddington ratio rather than the obscuration.

Random forest algorithm for classification of multiwavelength data

We introduced a decision tree method called Random Forests for multi-wavelength data classification. The data were adopted from different databases, including the Sloan Digital Sky Survey (SDSS) Data Release five, USNO, FIRST and ROSAT. We then studied the discrimination of quasars from stars and the classification of quasars, stars and galaxies with the sample from optical and radio bands and with that from optical and X-ray bands. Moreover, feature selection and feature weighting based on Random Forests were investigated. The performances based on different input patterns were compared. The experimental results show that the random forest method is an effective method for astronomical object classification and can be applied to other classification problems faced in astronomy. In addition, Random Forests will show its superiorities due to its own merits, e.g. classification, feature selection, feature weighting as well as outlier detection.

Forecasting the dark energy measurement with baryon acoustic oscillations: prospects for the LAMOST surveys

The Large Area Multi-Object Fibre Spectroscopic Telescope (LAMOST) is a dedicated spectroscopic survey telescope being built in China, with an effective aperture of 4 m and equipped with 4000 fibres. Using the LAMOST telescope, one could make redshift survey of the large-scale structure (LSS). The baryon acoustic oscillation features in the LSS power spectrum provide standard rulers for measuring dark energy and other cosmological parameters. In this paper, we investigate the measurement precision achievable for a few possible surveys: (1) a magnitude-limited survey of all galaxies, (2) a survey of colour-selected luminous red galaxies (LRG) and (3) a magnitude-limited, high-density survey of z < 2 quasars. For each survey, we use the halo model to estimate the bias of the sample, and calculate the effective volume. We then use the Fisher matrix method to forecast the error on the dark energy equation of state and other cosmological parameters for different survey parameters. In a few cases, we also use the Markov Chain Monte Carlo method to make the same forecast as a comparison. The fibre time required for each of these surveys is also estimated. These results would be useful in designing the surveys for LAMOST.