Automated Separation of Stars and Normal Galaxies Based on Statistical Mixture Modeling with RBF Neural Networks

For LAMOST, the largest sky survey program in China, the solution of the problem of automatic discrimination of stars from galaxies by spectra has shown that the results of the PSF test can be significantly refined. However, the problem is made worse when the redshifts of galaxies are not available. We present a new automatic method of star/(normal) galaxy separation, which is based on Statistical Mixture Modeling with Radial Basis Function Neural Networks (SMM-RBFNN). This work is a continuation of our previous one, where active and non-active celestial objects were successfully segregated. By combining the method in this paper and the previous one, stars can now be effectively separated from galaxies and AGNs by their spectra-a major goal of LAMOST, and an indispensable step in any automatic spectrum classification system. In our work, the training set includes standard stellar spectra from Jacoby's spectrum library and simulated galaxy spectra of EO, SO, Sa, Sb types with redshift ranging from 0 to 1     

The kpc-scale radio source population

We are conducting a multi-wavelength (radio, optical, and X-ray) observational campaign to classify, morphologically and physically, a sample of 55 flat-spectrum radio sources dominated by structure on kpc-scales. This sample contains 22 compact-/medium-sized symmetric object candidates, a class of objects thought to be in the early stages of the evolution of radio galaxies. The vast majority of the remaining objects have core-plus-one-sided-jet structures, half of which show sharply bent jets, probably due to strong interactions with the interstellar medium of the host galaxies. Once the observational campaign is completed, we will constrain evolutionary theories of radio galaxies at their intermediate stages and possibly understand the physics of the hypothesised narrow-line region in active galactic nuclei, given our advantageous statistical position.     

F stars, metallicity and the ages of red galaxies at z > 1

We explore whether the rest-frame near-ultraviolet spectral region, observable in high-redshift galaxies via optical spectroscopy, contains sufficient information to allow the degeneracy between age and metallicity to be lifted. We do this by first testing the ability of evolutionary synthesis models to reclaim the correct metallicity when fitted to the near-ultraviolet spectra of F stars of known (subsolar and supersolar) metallicity. F stars are of particular interest because the rest-frame near-ultraviolet spectra of the oldest known elliptical galaxies at   z > 1  appear to be dominated by F stars near to the main-sequence turn-off.
We find that, in the case of the F stars, where the Hubble Space Telescope ultraviolet spectra have a high signal-to-noise ratio, fitting models in which the metallicity is allowed to vary as a free parameter is rather successful at deriving the correct metallicity. As a result, the estimated turn-off ages of these stars yielded by model-fitting are well constrained. Encouraged by this we have fitted these same variable-metallicity models to the deep, optical spectra of the   z ≃ 1.5 mJy  radio galaxies 53W091 and 53W069 obtained with the Keck telescope. While the age and metallicity are not so easily constrained for these galaxies, we find that even when metallicity is allowed as a free parameter, the best estimates of their ages are still ≥3 Gyr, with ages younger than 2 Gyr now strongly excluded. Furthermore, we find that a search of the entire parameter space of metallicity and star formation history using MOPED leads to the same conclusion. Our results therefore continue to argue strongly against an Einstein–de Sitter universe, and favour a Λ-dominated universe in which star formation in at least these particular elliptical galaxies was completed somewhere in the redshift range   z = 3–5  .     

The X-ray spectra of optically selected Seyfert 2 galaxies: are there any Seyfert 2 galaxies with no absorption?

We present an X-ray spectral analysis of a sample of eight bona fide Seyfert 2 galaxies, selected on the basis of their high [O  iii ] λ 5007 flux, from the Ho et al. spectroscopic sample of nearby galaxies. We find that, in general, the X-ray spectra of our Seyfert 2 galaxies are complex, with some of our objects having spectra different from the 'typical' spectrum of X-ray selected Seyfert 2 galaxies. Two (NGC 3147 and 4698) show no evidence for intrinsic absorption. We suggest that this is a result of the fact that when the torus suppresses the intrinsic medium and hard energy flux, underlying emission from the host galaxy, originating in circumnuclear starbursts, and scattering from warm absorbers contributes in these energy bands more significantly. Our ASCA data alone cannot discriminate whether low-absorption objects are Compton-thick active galactic nuclei (AGNs) with a strong scattered component or lack an obscuring torus. The most striking example of our low absorption Seyfert 2 is NGC 4698. Its spectrum could be explained by either a dusty warm absorber or a lack of broad-line clouds so that its appearance as a Seyfert 2 is intrinsic and not a result of absorption.     

Deep Hubble Space Telescope imaging surveys and the formation of spheroidal galaxies

We have extended our previous analysis of morphologically selected elliptical and S0 galaxies in the Hubble Deep Field (HDF) North to include Hubble Space Telescope ( HST ) data in the HDF South and the HDFS–NICMOS areas. Our final sample amounts to 69 E/S0 galaxies with K <20.15 over an area of 11 arcmin². Although a moderately small number over a modest sky area, this sample benefits from the best imaging and photometric data available on high-redshift galaxies. Multi-waveband photometry allows us to estimate with good accuracy the redshifts for the majority of these galaxies, which lack a spectroscopic measure. We confirm our previous findings that massive E/S0s tend to disappear from flux-limited samples at z >1.4. This adds to the evidence that the rest-frame colours and spectral energy distributions (SEDs) of the numerous objects found at 0.8< z <1.2 are inconsistent with a very high redshift of formation for the bulk of stars, while they are more consistent with protracted (either continuous or episodic) star formation down to z ≤1. These results based on high-quality imaging on a small field can be complemented with data from colour-selected extremely red objects (EROs) on much larger sky areas: our claimed demise of E/S0s going from z =1 to z =1.5 is paralleled by a similarly fast decrease in the areal density of EROs when the colour limit is changed from ( R − K )=5 to ( R − K )=6 (corresponding to z ≃1 and z ≃1.3 respectively). Altogether, the redshift interval from 1 to 2 seems to correspond to a very active phase for the assembly of massive E/S0 galaxies in the field, and also probably one where a substantial fraction of their stars are formed.     

A survey of hard spectrum ROSAT sources – II. Optical identification of hard sources

We have surveyed 188 ROSAT Position Sensitive Proportional Counter (PSPC) fields for X-ray sources with hard spectra ( α <0.5); such sources must be major contributors to the X-ray background at faint fluxes. In this paper we present optical identifications for 62 of these sources: 28 active galactic nuclei (AGN) which show broad lines in their optical spectra (BLAGN), 13 narrow emission line galaxies (NELGs), five galaxies with no visible emission lines, eight clusters and eight Galactic stars.
The BLAGN, NELGs and galaxies have similar distributions of X-ray flux and spectra. Their ROSAT spectra are consistent with their being AGN obscured by columns of 20.5< log( N _H/cm⁻²)<23 . The hard spectrum BLAGN have a distribution of X-ray to optical ratios which is similar to that found for AGN from soft X-ray surveys (1< α _OX<2) . However, a relatively large proportion (15 per cent) of the BLAGN, NELGs and galaxies are radio loud. This could be because the radio jets in these objects produce intrinsically hard X-ray emission, or if their hardness is caused by absorption, it could be because radio-loud objects are more X-ray luminous than radio-quiet objects. The eight hard sources identified as clusters of galaxies are the brightest, and softest group of sources and hence clusters are unlikely to be an important component of the hard, faint population.
We propose that BLAGN are likely to constitute a significant fraction of the faint, hard, 0.5–2 keV population and could be important to reproducing the shape of the X-ray background, because they are the most numerous type of object in our sample (comprising almost half the identified sources), and because all our high redshift ( z >1) identified hard sources have broad lines.     

The relationship between X-ray variability and the central black hole mass

We assembled a sample of Seyfert 1 galaxies, quasi-stellar objects (QSOs) and low-luminosity active galactic nuclei (LLAGNs) observed by ASCA , the central black hole masses of which have been measured. We found that the X-ray variability (which is quantified by the 'excess variance' σ _rms²) is significantly anti-correlated with the central black hole mass, and it is likely that a linear relationship of σ _rms²∝ M _bh⁻¹ exists. It can be interpreted that the short time-scale X-ray variability is caused by some global coherent variations in the X-ray emission region, which is scaled by the size of the central black hole. Hence the central black hole mass is the driving parameter of the previously established relation between X-ray variability and luminosity. Our findings favour the hypothesis that the narrow-line Seyfert 1 galaxies and QSOs harbour smaller black holes than the broad-line objects, and can also easily explain the observational fact that high-redshift QSOs have greater variability than local AGNs at a given luminosity. Further investigations are needed to confirm our findings, and a large sample X-ray variability investigation can give constraints on the physical mechanisms and evolution of AGNs.     

A Comparison of BBN,ADTree and MLP in separating Quasars from Large Survey Catalogues

We compare the performance of Bayesian Belief Networks (BBN), Multilayer Perception (MLP) networks and Alternating Decision Trees (ADtree) on separating quasars from stars with the database from the 2MASS and FIRST survey catalogs. Having a train- ing sample of sources of known object types, the classifiers are trained to separate quasars from stars. By the statistical properties of the sample, the features important for classifica- tion are selected. We compare the classification results with and without feature selection. Experiments show that the results with feature selection are better than those without feature selection. From the high accuracy found, it is concluded that these automated methods are robust and effective for classifying point sources. They may all be applied to large survey projects (e.g. selecting input catalogs) and for other astronomical issues, such as the parame- ter measurement of stars and the redshift estimation of galaxies and quasars.     

Studying young stellar populations in G345.5+1.5 molecular cloud

Using different sky surveys, we studied IRAS sources and embedded clusters located in the G345.5+1.5 region. Our analysis comprises multi-wavelength (optical to radio) data, exploited using various methods: photometry, near-infrared spectroscopy for ten stars in the embedded cluster DBS-114, and astrometry. We estimated the main parameters of the embedded stellar populations in the G345.5+1.5 molecular cloud, such as their extent, reddening, age, and mass. We also found a consistent distance value using different approaches. For each studied population, we classified several point objects as early main sequence stars, stars with infrared excess, and class I/II YSOs. For the particular case of DBS 114, our spectral classification revealed four B-type stars, and we used astrometric information from GAIA EDR3. The combination of optical and infrared information revealed an abnormal reddening law in some embedded clusters. Our analysis favored a scenario with a Lyman continuum emission excess at some of the studied stellar populations.     

A catalogue of potential adaptive optics survey fields from the UKIRT archive

We present a multicolour catalogue of faint galaxies situated close to bright stars,   V ≲ 15  , with the aim of identifying high-redshift galaxies suitable for study with adaptive optics-equipped near-infrared imagers and spectrographs. The catalogue is constructed from archival calibration observations of the United Kingdom Infrared Telescope (UKIRT) Faint Standard stars with the UKIRT Fast Track Imager (UFTI) camera on UKIRT. We have analysed the deepest 16 fields from the archive to provide a catalogue of galaxies brighter than   K ∼ 20.3  lying between 3 and 25 arcsec of the guide stars. We identify 111 objects in a total survey area of  8.7 arcmin²  . Of these, 87 are classified as galaxies based on their light profiles in our ∼0.5 arcsec median seeing K -band images. 12 of the galaxies have  ( J − K ) ≥ 2.0  consistent with them lying at high redshifts,   z ≳ 2  . These 12 very red galaxies have K -band magnitudes of   K = 18.1–20.1  and separations from the guide stars of 4–20 arcsec and hence are very well suited to adaptive optics studies to investigate their morphologies and spectral properties on sub-kpc scales. We provide coordinates and JHK photometry for all catalogued objects.     

Rotation Periods of Nine ROSAT Selected Solar-Type Stars

We monitored 16 X-ray selected young solar-type stars for light variation and found appreciable periodic light variability with amplitudes of a few hundredths of a magni-tude in nine of the objects. Using the method of Phase Dispersion Minimization (PDM) and Fourier analysis (software PERIOD04), the rotation periods of these stars were determined from the photometric data. The rotation periods of all nine stars are shorter than about 3days. It is suggested that, as with the Pleiades cluster, small amplitude light variations are quite common among young solar-type stars with rotation periods around 3 days or less. This gives further evidence for the spin up of solar-type stars predicted by models of angular momentum evolution of pre-main sequence stars.     

The active nuclei of z < 1.0 3CRR radio sources

We combine Chandra and XMM–Newton X-ray data from our previous papers with new X-ray observations and with Spitzer mid-infrared (mid-IR) data in order to study the nature of the nuclei of radio galaxies and radio-loud quasars with   z < 1.0  from the 3CRR sample. The significant increase in sample size over our previous work, the reduction of bias in the sample as a result of new observations and the availability of more mid-IR data allow us to show conclusively that almost all objects classed as low-excitation radio galaxies in optical spectroscopic studies lack a radiatively efficient active nucleus. We show that the distribution of absorbing columns in the narrow-line radio galaxies differs from the population of X-ray-selected radio-quiet type 2 quasars and from that in local Seyfert 2s. We comment on the current evidence for the nature of the soft X-ray component in radio-galaxy nuclear spectra, concluding that a jet origin for this component is very hard to evade. Finally, we discuss the recently discovered 'fundamental plane' of black hole activity, showing that care must be taken when placing radio-loud active galactic nucleus (AGN) on such diagnostic diagrams.     

Color-Redshift Relations and Photometric Redshift Estimations of Quasars in Large Sky Surveys

With a recently constructed composite quasar spectrum and the X2 minimization technique, we describe a general method for estimating the photometric redshifts of a large sample of quasars by deriving theoretical color-redshift relations and comparing the theoretical colors with the observed ones. We estimated the photometric redshifts from the 5-band SDSS photometric data of 18678 quasars in the first major data release of SDSS and compared them with their spectroscopic redshifts. The difference is less than 0.1 for 47% of the quasars and less than 0.2 for 68%. Based on the calculation of the theoretical color-color diagrams of stars, galaxies and quasars both on the SDSS system and on the BATC system, we expect that we would be able to select candidates of high redshift quasars more efficaciously with the latter than with the former, provided the BATC survey can detect objects with magnitudes fainter than 21.     

Template-based classification of SDSS-<Emphasis Type="BoldItalic">GALEX</Emphasis> point sources

We have classified a sample of 37,492 objects from SDSS into QSOs, galaxies and stars using photometric data over five wave bands (u, g, r, i and z) and UV GALEX data over two wave bands (near-UV and far-UV) based on a template fitting method. The advantage of this method of classification is that it does not require any spectroscopic data and hence the objects for which spectroscopic data is not available can also be studied using this technique. In this study, we have found that our method is consistent by spectroscopic methods given that their UV information is available. Our study shows that the UV colours are especially important for separating quasars and stars, as well as spiral and starburst galaxies. Thus it is evident that the UV bands play a crucial role in the classification and characterization of astronomical objects that emit over a wide range of wavelengths, but especially for those that are bright at UV. We have achieved the efficiency of 89% for the QSOs, 63% for the galaxies and 84% for the stars. This classification is also found to be in agreement with the emission line diagnostic diagrams.     

Multiwavelength study of X-ray selected star-forming galaxies within the Chandra Deep Field-South

We have combined multiwavelength observations of a selected sample of star-forming galaxies with galaxy evolution models in order to compare the results obtained for different star formation rate (SFR) tracers and to study the effect that the evolution of the star-forming regions has on them. We also aimed at obtaining a better understanding of the corrections due to extinction and nuclear activity on the derivation of the SFR. We selected the sample from Chandra data for the well studied region Chandra Deep Field -South (CDFS) and chose the objects that also have ultraviolet (UV) and infrared (IR) data from Galaxy Evolution Explorer ( GALEX ) and Great Observatories Origins Deep Survey (GOODS) Spitzer , respectively.
Our main finding is that there is good agreement between the extinction corrected SFR(UV) and the SFR(X), and we confirm the use of X-ray luminosities as a trustful tracer of recent star formation activity. Nevertheless, at SFR(UV) larger than about  5 M_⊙ yr⁻¹  there are several galaxies with an excess of SFR(X) suggesting the presence of an obscured active galactic nucleus (AGN) not detected in the optical spectra. We conclude that the IR luminosity is driven by recent star formation even in those galaxies where the SFR(X) is an order of magnitude higher than the SFR(UV) and therefore may harbour an AGN. One object shows SFR(X) much lower than expected based on the SFR(UV); this SFR(X) 'deficit' may be due to an early transient phase before most of the massive X-ray binaries were formed. An X-ray deficit could be used to select extremely young bursts in an early phase just after the explosion of the first supernovae associated with massive stars and before the onset of massive X-ray binaries.     

X-ray spectroscopy of stars

Non-degenerate stars of essentially all spectral classes are soft X-ray sources. Their X-ray spectra have been important in constraining physical processes that heat plasma in stellar environments to temperatures exceeding one million degrees. Low-mass stars on the cooler part of the main sequence and their pre-main sequence predecessors define the dominant stellar population in the galaxy by number. Their X-ray spectra are reminiscent, in the broadest sense, of X-ray spectra from the solar corona. The Sun itself as a typical example of a main-sequence cool star has been a pivotal testbed for physical models to be applied to cool stars. X-ray emission from cool stars is indeed ascribed to magnetically trapped hot gas analogous to the solar coronal plasma, although plasma parameters such as temperature, density, and element abundances vary widely. Coronal structure, its thermal stratification and geometric extent can also be interpreted based on various spectral diagnostics. New features have been identified in pre-main sequence stars; some of these may be related to accretion shocks on the stellar surface, fluorescence on circumstellar disks due to X-ray irradiation, or shock heating in stellar outflows. Massive, hot stars clearly dominate the interaction with the galactic interstellar medium: they are the main sources of ionizing radiation, mechanical energy and chemical enrichment in galaxies. High-energy emission permits to probe some of the most important processes at work in these stars, and put constraints on their most peculiar feature: the stellar wind. Medium and high- resolution spectroscopy have shed new light on these objects as well. Here, we review recent advances in our understanding of cool and hot stars through the study of X-ray spectra, in particular high-resolution spectra now available from XMM-Newton and Chandra. We address issues related to coronal structure, flares, the composition of coronal plasma, X-ray production in accretion streams and outflows, X-rays from single OB-type stars, massive binaries, magnetic hot objects and evolved WR stars.     

Neighbourhoods of isolated star forming dwarf galaxies

We report the discovery of small groups of uncatalogued, compact, star forming (SF) dwarf galaxies (DGs) by Hα mapping of the neighbourhoods of apparently isolated, catalogued, SF DGs. Our sample consists of dwarf  ( M ≥−18 mag)  galaxies at least 2 Mpc away from any other catalogued galaxy. The galaxies were selected to exhibit Hα emission of any intensity, i.e. not selecting only strong starbursts, as an indicator of recent or on-going star formation with the goal of understanding why are they presently forming stars. We identified possible neighbours by imaging the galaxies and their surroundings through Hα filters centred at or near the redshift of the galaxy, and searching for localized Hα emission with the characteristics of the line emission from the sample galaxies.
We identified 20 possible SF neighbour galaxies, 17 of them not previously catalogued, in three of the five search fields where we had good quality data, and present here their positions and, images and morphology, as well as some indications of binarity. The relatively large number of possible neighbour candidates, combined with their relative faintness, argue that it would be virtually impossible to identify truly isolated galaxies. It seems that the objects we selected as extremely isolated are probably the brightest members of sparse groups of galaxies, where the other members are also DGs that are presently forming stars. In order to enhance the confidence of this statement regular redshifts are required for our candidate neighbours.     

A multiwavelength study of the early evolution of the classical nova LMC 1988 1

We study star-formation-inducing mechanisms in galaxies through multiwavelength measurements of a sample of dwarf galaxies in the Virgo cluster described in Paper I. Our main goal is to test how star-formation-inducing mechanisms depend on several parameters of the galaxies, such as morphological type and hydrogen content. We derive the star formation rate and star formation histories of the galaxies, and check their dependence on other parameters.   Comparison of the sample galaxies with population synthesis models shows that these objects have significantly lower metallicity than the solar value. The colours can generally be explained as a combination of two different stellar populations: a young (3–20 Myr) metal-poor population which represents the stars currently forming presumably in a starburst, and an older (0.1–1 Gyr) population of previous stellar generations. There is evidence that the older stellar population was also formed in a starburst. This is consistent with the explanation that star formation in this type of objects takes place in short bursts followed by long quiescent periods.   No significant correlation is found between the star formation properties of the sample galaxies and their hydrogen content. Apparently, when star formation occurs in bursts, other parameters influence the star formation properties more significantly than the amount of atomic hydrogen. No correlation is found between the projected Virgocentric distance and the rate of star formation in the galaxies, suggesting that tidal interactions are not significant in triggering star formation in cluster dwarf galaxies.     

The most metal-poor galaxies

Summary. Metallicity is a key parameter that controls many aspects in the formation and evolution of stars and galaxies. In this review we focus on the metal deficient galaxies, in particular the most metal-poor ones, because they play a crucial r?le in the cosmic scenery. We first set the stage by discussing the difficult problem of defining a global metallicity and how this quantity can be measured for a given galaxy. The mechanisms that control the metallicity in a galaxy are reviewed in detail and involve many aspects of modern astrophysics: galaxy formation and evolution, massive star formation, stellar winds, chemical yields, outflows and inflows etc. Because metallicity roughly scales as the galactic mass, it is among the dwarfs that the most metal-poor galaxies are found. The core of our paper reviews the considerable progress made in our understanding of the properties and the physical processes that are at work in these objects. The question on how they are related and may evolve from one class of objects to another is discussed. While discussing metal-poor galaxies in general, we present a more detailed discussion of a few very metal-poor blue compact dwarf galaxies like IZw18. Although most of what is known relates to our local universe, we show that it pertains to our quest for primeval galaxies and is connected to the question of the origin of structure in the universe. We discuss what do QSO absorption lines and known distant galaxies tell us already? We illustrate the importance of star-forming metal-poor galaxies for the determination of the primordial helium abundance, their use as distance indicator and discuss the possibility to detect nearly metal-free galaxies at high redshift from Ly emission. Received 19 August 1999 / Published online: 15 February 2000     

Stellar X-ray Sources in the Rosette Nebula

We present optical photometric and spectroscopic studies of ROSAT X-ray stellar sources in the Rosette Nebula star-forming region. The brightest X-ray sources are either massive stars or active T Tauri stars associated with the open cluster NGC 2244, or are foreground stars. Some of the spectra of the young stars newly identified in the region are presented.