Sky Survey Project Research Articles

Search for Young Stellar Objects within 4XMM-DR13 Using CatBoost and SPE

Classifying and summarizing large data sets from different sky survey projects is essential for various subsequent scientific research. By combining data from 4XMM-DR13, Sloan Digital Sky Survey (SDSS) DR18, and CatWISE, we formed an XMM-WISE-SDSS sample that included information in the X-ray, optical, and infrared bands. By cross matching this sample with data sets from known spectral classifications from SDSS and LAMOST, we obtained a training data set containing stars, galaxies, quasars, and young stellar objects (YSOs). Two machine learning methods, CatBoost and Self-Paced Ensemble (SPE), were used to train and construct machine learning models through training sets to classify the XMM-WISE-SDSS sample. Notably, the SPE classifier showed excellent performance in YSO classification, identifying 1102 YSO candidates from 160,545 sources, including 258 known YSOs. Then we further verify whether these candidates are YSOs by the spectra in LAMOST and the identification in the SIMBAD and VizieR databases. Finally there are 412 unidentified YSO candidates. The discovery of these new YSOs is an important addition to existing YSO samples and will deepen our understanding of star formation and evolution. Moreover we provided a classification catalog for the whole XMM-WISE-SDSS sample.

Data mining techniques on astronomical spectra data – III. Association analysis

ABSTRACT Association analysis is an important task that aims to investigate correlations in astronomical spectral data and mine relationships between different data features. With the rapid development of various sky survey projects, multiple association analysis methods have been applied to efficiently investigate the correlation between spectral data. However, due to the different focuses of analysing the relationships within spectral data, the performance and applicability scenarios of association analysis methods vary. We present the third article in the series to provide a comprehensive review of algorithms for astronomical spectral association analysis. First, this paper outlines the ideas and applications of association analysis algorithms for astronomical spectra in the current literature. Secondly, experiments are conducted on a unified A-type stellar spectral data set constructed based on three different signal-to-noise ratios and data volumes to examine the performance of different algorithms in analysing the correlation between data features. The results indicate that association rule algorithms can more comprehensively and effectively uncover the correlations among different spectral features, while regression analysis algorithms offer a simpler and more intuitive approach to analysing relationships between features. The spectral data used in the experiments are obtained from the Large Sky Area Multi-Object Fiber Spectroscopic Telescope. Finally, the source code of association analysis algorithms and manuals for usage are provided on GitHub.

Classification of Astronomical Spectra Based on Multiscale Partial Convolution

The automated and efficient classification of astronomical spectra is an important research issue in the era of large sky surveys. Most current studies on automatic spectral classification primarily focus on specific data sets and demonstrate outstanding performance. However, the diversity in spectra poses formidable challenges for these classification models, as they exhibit limited capability to generalize across more comprehensive data sets. In response to these challenges, we pioneer a method called the multiscale partial convolution net (MSPC-Net), which amalgamates partial, large kernel, and grouped convolution to facilitate multilabel spectral classification. By harnessing the capabilities of partial convolution, MSPC-Net can effectively reduce the number of model parameters, accelerate the training process, and mitigate the overfitting issue. Integrating large kernel and grouped convolution empowers the model to capture local and global features simultaneously, enhancing its overall classification efficacy. To rigorously evaluate the model’s performance, we generate ten different data sets sourced from the Sloan Digital Sky Survey and Large Sky Area Multi-Object Spectroscopic Telescope. These data sets encompass stellar class, stellar subclass, and full classification, providing a comprehensive assessment across various application scenarios. The experimental results reveal that MSPC-Net consistently outperforms the other models across different data sets, especially demonstrating superior performance in the last two data sets with full classification. Consequently, MSPC-Net is poised to find extensive applications in the detailed classification for large-scale sky survey projects. This work not only addresses the challenges of generalization in spectral classification but also contributes significantly to the advancement of robust models for astronomical research.

A Model Estimator for Noisy Compact Emission Recovery in Radio Synthesis Imaging

Reconstruction of a noisy compact emission must consider not only the point-spread function but also the effect of noise. However, the traditional threshold method in widely-used CLEAN-based algorithms finds it difficult to effectively prevent noise in the model image during noisy compact-emission reconstruction. This significantly limits the performance in noisy compact-emission reconstruction, such as deep field imaging. There are two major difficulties in the accurate reconstruction of a Stokes-I image of compact emission: first, the threshold method that has been used in practice is difficult to use to separate compact emission and noise; and second, over-subtraction makes it difficult for the reconstructed Stokes-I model image to remain positive. Therefore, a filter-based denoizing mechanism is introduced in the search phase of the model components to separate signal and noise so that the signal can be effectively extracted. The relatively larger loop gain for positive components means that the reconstructed model is in line with astrophysics. This will reduce the errors between the true sky image and the model image. The new model estimator is tested on a simulated JVLA observation with realistic source distributions from the VLA Low-Frequency Sky Survey project and the SKADS/SCubed simulation. The experiments show that it is very effective when used to separate signal and noise to lower the noise in the model image. This work explores the use of existing common software CASA to achieve high dynamic range imaging, which is an important step toward square kilometer array data processing.

A novel approach for variable star classification based on imbalanced learning

Abstract The advent of time-domain sky surveys has generated a vast amount of light variation data, enabling astronomers to investigate variable stars with large-scale samples. However, this also poses new opportunities and challenges for the time-domain research. In this paper, we focus on the classification of variable stars from the Catalina Surveys Data Release 2 and propose an imbalanced learning classifier based on Self-paced Ensemble (SPE) method. Compared with the work of Hosenie et al. (2020), our approach significantly enhances the classification Recall of Blazhko RR Lyrae stars from 12% to 85%, mixed-mode RR Lyrae variables from 29% to 64%, detached binaries from 68% to 97%, and LPV from 87% to 99%. SPE demonstrates a rather good performance on most of the variable classes except RRab, RRc, and contact and semi-detached binary. Moreover, the results suggest that SPE tends to target the minority classes of objects, while Random Forest is more effective in finding the majority classes. To balance the overall classification accuracy, we construct a Voting Classifier that combines the strengths of SPE and Random Forest. The results show that the Voting Classifier can achieve a balanced performance across all classes with minimal loss of accuracy. In summary, the SPE algorithm and Voting Classifier are superior to traditional machine learning methods and can be well applied to classify the periodic variable stars. This paper contributes to the current research on imbalanced learning in astronomy and can also be extended to the time-domain data of other larger sky survey projects (LSST, etc.).

A deep learning based astronomical target detection framework for multi-colour photometry sky survey projects

Multi-colour photometry sky survey projects would obtain celestial object images of different colours with a wide field telescope and several different filters. Images of different colours could reveal different components of celestial objects. We would be able to investigate details of celestial objects and even discover many new celestial objects with these images. Nevertheless, celestial objects are randomly and sparsely distributed in observational images and we need to detect them from these images before we could carry out further study. According to properties of multi-colour images, we propose a novel deep learning based astronomical target detection framework. Our framework could adapt to images with any number of colours and directly detect celestial objects from these images. For extended targets, such as galaxies, our framework could detect and segment their images; for point like targets, such as stars or quasars, our framework could detect and locate their images. From practical application aspects, we have tested the performance of our framework by building a pipeline to process SDSS images with one, three and five colours. Results show that our framework could provide reliable detection results. Our framework could be used as a basic block to build pipelines for multi-colour photometry sky survey projects.

Predicting Supermassive Black Hole Mass with Machine Learning Methods

It is crucial to measure the mass of supermassive black holes (SMBHs) in understanding the co-evolution between the SMBHs and their host galaxies. Previous methods usually require spectral data which are expensive to obtain. We use the AGN catalog from the Sloan Digital Sky Survey project Data Release 7 (DR7) to investigate the correlations between SMBH mass and their host galaxy properties. We apply the machine learning algorithms, such as Lasso regression, to establish the correlation between the SMBH mass and various photometric properties of their host galaxies. We find an empirical formula that can predict the SMBH mass according to galaxy luminosity, colors, surface brightness, and concentration. The root-mean-square error is 0.5 dex, comparable to the intrinsic scatter in SMBH mass measurements. The 1σ scatter in the relation between the SMBH mass and the combined galaxy properties relation is 0.48 dex, smaller than the scatter in the SMBH mass versus galaxy stellar mass relation. This relation could be used to study the SMBH mass function and the AGN duty cycles in the future.

Searching for Gravitationally Lensed Gamma-Ray Bursts with Their Afterglows

Abstract Gamma-ray bursts (GRBs) at high redshifts are expected to be gravitationally lensed by objects of different mass scales. Other than a single recent claim, no lensed GRB has been detected so far by using gamma-ray data only. In this paper, we suggest that multiband afterglow data might be an efficient way to search for lensed GRB events. Using the standard afterglow model, we calculate the characteristics of the lensed afterglow lightcurves under the assumption of two popular analytic lens models: the point-mass and singular isothermal sphere models. In particular, when different lensed images cannot be resolved, their signals would be superimposed together with a given time delay. In this case, the X-ray afterglows are likely to contain several X-ray flares of similar width in linear scale and similar spectrum, and the optical afterglow lightcurve will show re-brightening signatures. Since the lightcurves from the image arriving later would be compressed and deformed in the logarithmic timescale, the larger time delay (i.e., the larger mass of the lens), the easier it is to identify the lensing effect. We analyzed the archival data of optical afterglows and found one potential candidate of the lensed GRB (130831A) with time delay ∼500 s; however, observations of this event in gamma-ray and X-ray bands seem not to support the lensing hypothesis. In the future, with the cooperation of the all-sky monitoring gamma-ray detectors and multiband sky survey projects, the method proposed in this paper would be more efficient in searching for strongly lensed GRBs.

The Milky Way Revealed by Variable Stars. I. Sample Selection of RR Lyrae Stars and Evidence for Merger History

Abstract In order to study the Milky Way, RR Lyrae (RRL) variable stars identified by Gaia, ASAS-SN, and ZTF sky survey projects have been analyzed as tracers in this work. Photometric and spectroscopic information of 3417 RRLs including proper motions, radial velocity, and metallcity are obtained from observational data of Gaia, LAMOST, GALAH, APOGEE, and RAVE. Precise distances of RRLs with typical uncertainties less than 3% are derived by using a recent comprehensive period–luminosity–metallicity relation. Our results from kinematical and chemical analysis provide important clues for the assembly history of the Milky Way, especially for the Gaia–Sausage ancient merger. The kinematical and chemical trends found in this work are consistent with those of recent simulations that indicated that the Gaia–Sausage merger had a dual origin in the Galactic thick disk and halo. As recent similar works have found, the halo RRL sample in this work contains a subset of radially biased orbits besides a more isotropic component. This higher orbital anisotropy component amounts to β ≃ 0.8, and it contributes between 42% and 83% of the halo RRLs at 4 < R( kpc) < 20.

Stellar Spectral Classification Based on Capsule Network

The rapid development of large-scale sky survey project has produced a large amount of stellar spectral data, which make the automatic classification of stellar spectral data a challenging task. In this paper, we have proposed a stellar spectral classification method based on a capsule network. At first, by using the one-dimensional convolutional network and short-time Fourier transform (STFT), the one-dimensional spectra of the F5, G5, and K5 types selected from the LAMOST Data Release 5 (DR5) are converted into the two-dimensional Fourier spectrum images. Then, the two-dimensional Fourier spectrum images are classified automatically by the capsule network. Because the capsule network can preserve the hierarchical pose relationships among the entities in the image, and it does not need any pooling layers, the experimental results show that the capsule network has a better classification performance, for the classifications of the F5, G5, and K5-type stellar spectra, its classification accuracy is superior to other classification methods.

Data analysis methods in astronomic objects classification (Sloan Digital Sky Survey DR14)

In the paper Sloan Digital Sky Survey DR14 dataset was investigated. It contains statistical information about many astronomical objects. The information was obtained within the framework of the Sloan Digital Sky Survey project. There are telescopes at the Earth surface, at the Earth orbit and in the Lagrange points of some systems (Earth–Moon, Sun–Earth). The telescopes gain information in different frequency ranges. The large quantity of statistical information leads to the demand for analytical algorithms and systems capable of making classification. Such information is marked up well enough to build machine learning classification systems. The paper presents the results of a number of classifiers. The handled data contains measures of three types of astronomical objects of the Sloan Digital Sky Survey DR14 dataset (star, quasar, galaxy). The CART decision tree, logistic regression, naïve Bayes classifiers and ensembles of classifiers (random forest, gradient boosting) were implemented. Conclusions about special features of each machine learning classifier trained to solve this task are made at the end of the paper. In some cases, classifiers’ structure can be explained physically. The accuracy of the classifiers built in this research is more than 90% (metrics F1, precision and recall are implemented, because the classes are unbalanced). Taking these values into account classification task is supposed to be successfully solved. At the same time, the structure of classifiers and importance of features can be used as a physical explanation of the solution.

Searching for RR Lyrae Stars by Imbalanced Learning

As standard candles, RR Lyrae stars are taken as distance markers and age indicators, which is helpful to estimate the distance and age of globular clusters as well as galactic and extragalactic (Local Group) locations. At the same time, as tracers, they reveal the structure, chemical and dynamic evolution of the galaxy, as well as the substructure of the halo. However, the number of RR stars is very small compared with other types of stars in nature. With the development and operation of large-scale sky survey projects, they have made a qualitative leap in number. Compared with other types of stars, RR Lyrae stars are still not enough in number. Therefore, separating RR Lyrae stars from stars belongs to an imbalanced learning problem. It may be solved by popular imbalanced learning methods.

Identification of temperature anomaly RR Lyrae stars in LAMOST survey, misclassification and binarities

RR Lyrae stars, a well-known type of pulsating variable stars, have been known about for more than a century. A large amount of photometric data on RR Lyrae stars has been accumulated by space- and ground-based sky surveys, but the spectral data are relatively poor. Fortunately, the LAMOST sky survey project provides an opportunity to view them from the point of view of spectra. We collect the atmospheric parameters of 1685 RR Lyrae stars provided by the LAMOST catalog, and carry out research by using the reliable Teff. We find that there is a clear correlation between their Teff and pulsation periods, which is consistent with the pulsation and evolution theories of RR Lyrae stars. In addition, we focus on those RR Lyrae stars with abnormal temperatures. After analyzing the data from several photometric surveys, we find that some of these temperature anomalies are misclassified variable stars (e.g., eclipsing binaries, pulsating stars on main sequence), and some are RR Lyrae binary candidates. For the latter, the temperatures of potential companions should be lower and their luminosities should not be neglected (e.g., red giant stars). We obtain that the ratio of temperature anomaly stars to all the sample stars is 4%, which means that the impact on the further analysis (e.g., kinematics analysis) is low. We also present the catalogs of spectral anomaly RR Lyrae stars.

3-D electronic density maps for enzyme reaction intermediates: big data and high performance computing services in a virtualized environment

Enzymatic reactions play key role in metabolism of all living organisms. Recent advances in computational methods allow detailed study of enzyme-catalyzed processes, providing data on all intermediate states along reaction path. Furthermore, experimental data on the structure of the reaction intermediates become more available due to advancing of ultrafast X-Ray spectroscopy. These data describes electronic density of intermediates, sometimes even in time series. Currently, prominent Protein Data Bank (PDB) keeps records on the raw crystallographic data on electron density, but usually describes only enzymes or enzymes-inhibitor complexes. A database service where the both experimental and simulated data can be stored and analyzed simultaneously will be a valuable scientific tool for bringing theory and experiment together.There are hundreds of enzymes known and one should expect large volumes of data on three-dimensional property like electronic density. At the same time, quantum chemistry simulations of large biomolecules require extensive computing resources. Thus, we turned into service model using Nautilus environment of Cognitive Hardware And Software Ecosystem Community Infrastructure (CHASE-CI) project as a blueprint. In that environment, each participating organization can donate their hardware resources, like storage or compute elements, while retaining flexible access control to running services and datasets. We have installed and tested a small-scale environment prototype using resources of LCC laboratory at Chemistry Department Lomonosov Moscow State University, Joint Institute of Nuclear Research (Dubna) and RSC Group company (Moscow). As it was efficiently implemented in Sloan Digital Sky Survey (SDSS) project, we identified basic search queries and structured database accordingly.

THE RE-PROCESSING RESULTS OF PHOTOGRAPHIC OBSERVATIONS OF ASTEROIDS WITH GAIA CATALOG AT THE MAO NAS OF UKRAINE

In total, about 4,500 asteroid po-sitions were determined from photographic observations conducted at the MAO NAS of Ukraine in 1952-1996. These are the early positions of the selected minor planets, mostly bright, the observations of which were initiated by the founders of the ORBITA observational program and were made in 1952-1986. The second later positions of asteroids obtained from digitized plates from photographic observations of the Northern Sky Survey project in 1981-1996. Using digital plate processing technologies, 2292 asteroid positions were obtained, the images of which were fixed during observations and identified during processing.The new star catalog Gaia DR2 was used to study changes in determining the accuracy of equatorial coordinates for various faint objects on plates. For several 8 plates with images of Pluto comets and faint asteroids of particular interest due to the lack of observations in the past, new complete processing of scans of these plates was performed. The Gaia DR2 was used as a reference catalog. The asteroid positions from early observations of 1952- 1986 were determined in accordance with classical linear reduction algorithms. These are 1651 positions of minor planets determined in the PPM catalog system and obtained from observations with the Double Wide-Angle Astrograph. The remaining 591 positions were determined in the systems of old reference catalogs of stars Yale, SAO, AGK3, and obtained from observations with the Double Long-Focus Astrograph. It was for these positions that the calibration with the Gaia catalog was performed. The data on the positions of minor planets digitized from publications and additional data on their reference systems deposited in previous years at VINITI (now VINITI Database RAS) were used. Comparison with JPL ephemeris was performed both for new determinations of asteroid coordinates with the Gaia catalog, and for past determinations obtained in systems of other reference catalogs.

ASTEROIDS SEARCH RESULTS IN DIGITIZED OBSERVATIONS OF THE NORTHERN SKY SURVEY PROJECT (KITAB PART)

Photographic observations of XX century contained numerous and varied information about all objects and events of the Universe fixed on the astronegatives. The original and interesting observations of small bodies of the Solar system in previous years can be selected and used for various scientific tasks. Existing databases and online services can help make such selection easily and quickly.The observations of chronologically earlier oppositions, photometric evaluation of brightness for long periods of time allow refining the orbits of asteroids and identifying various non-stationarities.Photographic observations of the Northern Sky Survey project were used for global search for small bodies of Solar system. About 2,000 photographic plates of Kitab part of the Northern Sky Survey project were made using Double Wide Angle Astrograph at the Kitab observatory, Uzbekistan during 1981-1989. Early, using that digitized observations the catalogue of equatorial coordinates and stellar magnitudes for 13 413 268 stars and galaxies up B≤17.5m was composed. At present, we analyzed all processing results for the search of asteroids and compilation of the catalogue of equatorial coordinates and stellar magnitudes of them.As a result more than 4,500 asteroids and comets with visual magnitude from 7m.7 to 17m.5 were identified now. All positions of asteroids were compared with ephemeris. A comparative analysis of O-C differences was carried out.New and interesting is that the moments of discovery of some identified asteroids much later than their moments of observation. In addition, some of them are the earliest observations of these asteroids in the world among all known observations. More than 915 observations of such asteroids have been found on the plates of Kitab part of the FON project.

Catalog of asteroids from processing of digitized Photographic Observations of Northern sky survey (FON) project

Catalog of asteroids from processing of digitized Photographic Observations of Northern sky survey (FON) project

CATALOGUE OF COORDINATES AND B-MAGNITUDES IN - 20є – +2є ZONE BASED ON THE ULUGH BEG ASTRONOMICAL INSTITUTE PART OF THE FON PROJECT

Catalogue of 13.4 million stars down to 17.5m was obtained in 2017 by using plates from the Kitab Observatory of Ulugh Beg Astronomical Institute (UBAI) of the Uzbek Academy of Sciences. Kitab's part of Photographic Sky Survey (Russian abbreviation is FON, in next contexts) project include more than 2600 photographic plates, exposed on the Double Astrograph of Zeiss (DAZ, D/F = 40/300, 69/mm) from 1981 to 1996. Digitizationof these plates was made by using Epson Expression 10000XL scanner with the 1200 dpi resolution. Catalogue includes objects from the 1963 plates in declination zone between -20° and +2° for middle epoch 1984.97. The equatorial coordinates of objects were determined in the Tycho-2 reference system and the B-magnitudes in the system of the photoelectric standards. Five participants from Uzbekistan, Germany and Ukraine have taken part in the processing of the digitized images. The average internal accuracy of the catalogue for one observation are 0.23 and 0.15m for the equatorial coordinates and B-magnitudes respectively. For the stars brighter than 14m the errors are 0.09 and 0.05m respectively. The analysis of the catalogue and its comparison with the several astrometric catalogues was done.

Catalog of positions and B magnitudes of stars of the part of Kitab northern sky survey (FON) project

Catalog of positions and B magnitudes of stars of the part of Kitab northern sky survey (FON) project

SEARCH SMALL BODIES IMAGES IN COLLECTIONS DIGITIZED PHOTOGRAPHIC OBSERVATIONS OF PREVIOUS YEARS

Photographic observations of XX century is an important source of information on small bodies of the Solar system. The observations of chronologically earlier oppositions, photometric evaluation of brightness for long periods of time allow refining the orbits of asteroids and identifying various non-stationaries. Unfortunately international databases were contained a very little observable data for accurate estimates of geometric and kinematic parameters of asteroids. We analyzed the results of digital processing of 338 photographic plates of Northern Sky Survey (FON) project. Modern approach to processing early observations using new digital technologies provides a sufficiently high accuracy of coordinates and magnitudes for all objects on astroplates. As a result 570 images of asteroids and 4 images of comets from 8-16 magnitudes were detected on these plates. Equatorial coordinates and stellar magnitudes of them were composed to catalog of asteroids. All positions were compared with ephemeris. Analysis of ephemeris calculations with known today about the several hundred small planets with observation data showed some interesting results.