Subscribe to the gold package and get unlimited access to Shamra Academy
Register a new userRandom Forest Classification of Stars in the Galactic Centre
303
0
0.0
(
0
)
Authors
P. M. Plewa
Ask ChatGPT about the research
No Arabic abstract
Near-infrared high-angular resolution imaging observations of the Milky Ways nuclear star cluster have revealed all luminous members of the existing stellar population within the central parsec. Generally, these stars are either evolved late-type giants or massive young, early-type stars. We revisit the problem of stellar classification based on intermediate-band photometry in the K-band, with the primary aim of identifying faint early-type candidate stars in the extended vicinity of the central massive black hole. A random forest classifier, trained on a subsample of spectroscopically identified stars, performs similarly well as competitive methods (F1=0.85), without involving any model of stellar spectral energy distributions. Advantages of using such a machine-trained classifier are a minimum of required calibration effort, a predictive accuracy expected to improve as more training data becomes available, and the ease of application to future, larger data sets. By applying this classifier to archive data, we are also able to reproduce the results of previous studies of the spatial distribution and the K-band luminosity function of both the early- and late-type stars.
rate research
Read More
We present an evaluation of the performance of an automated classification of the Hipparcos periodic variable stars into 26 types. The sub-sample with the most reliable variability types available in the literature is used to train supervised algorithms to characterize the type dependencies on a number of attributes. The most useful attributes evaluated with the random forest methodology include, in decreasing order of importance, the period, the amplitude, the V-I colour index, the absolute magnitude, the residual around the folded light-curve model, the magnitude distribution skewness and the amplitude of the second harmonic of the Fourier series model relative to that of the fundamental frequency. Random forests and a multi-stage scheme involving Bayesian network and Gaussian mixture methods lead to statistically equivalent results. In standard 10-fold cross-validation experiments, the rate of correct classification is between 90 and 100%, depending on the variability type. The main mis-classification cases, up to a rate of about 10%, arise due to confusion between SPB and ACV blue variables and between eclipsing binaries, ellipsoidal variables and other variability types. Our training set and the predicted types for the other Hipparcos periodic stars are available online.
We propose an algorithm named best-scored random forest for binary classification problems. The terminology best-scored means to select the one with the best empirical performance out of a certain number of purely random tree candidates as each single tree in the forest. In this way, the resulting forest can be more accurate than the original purely random forest. From the theoretical perspective, within the framework of regularized empirical risk minimization penalized on the number of splits, we establish almost optimal convergence rates for the proposed best-scored random trees under certain conditions which can be extended to the best-scored random forest. In addition, we present a counterexample to illustrate that in order to ensure the consistency of the forest, every dimension must have the chance to be split. In the numerical experiments, for the sake of efficiency, we employ an adaptive random splitting criterion. Comparative experiments with other state-of-art classification methods demonstrate the accuracy of our best-scored random forest.
Two groups of astronomers used large telescopes Keck and VLT for decades to observe trajectories of bright stars near the Galactic Centre. Based on results of their observations astronomers concluded that trajectories of the stars are roughly elliptical and foci of the orbits are approximately coincide with the Galactic Centre position. It gives an opportunity to claim that the Newtonian potential of point like mass around $4.3times 10^6 M_odot$ is a good initial approximation for the gravitational potential near the Galactic Centre. In the last years, the astronomers found that gravitational redshift of S2 star near pericenter passage in May 2018 is in accordance with general relativity predictions. In 2020 the GRAVITY team found that the observed relativistic precession of S2 star orbit is also consistent with theoretical estimates calculated for a weak gravitational field approximation in a Schwarzschild black hole. In last years a a self-gravitating dark matter core--halo distribution suggested by Ruffini, Arguelles and Rueda (MNRAS, 2015) (RAR model) was proposed and recently Becerra-Vergara et al. (MNRAS, 2021) claimed that this model provides a better fit of trajectories of bright stars in comparison with the conventional model with the supermassive black hole. We confirm that in the case of this dark matter distribution model for a dense core trajectories of test bodies are elliptical but in this case centers (not foci) of these ellipses should coincide with the Galactic Centre and orbital periods do not depend on semi-major axis and it contradicts observational data and therefore, we concluded supermassive black hole is a preferable model in comparison with the a dense core--diluted halo density profile for the Galactic Centre.
Large fraction of studies of active galactic nuclei objects is based on performing follow-up observations using high-sensitivity instruments of high flux states observed by monitoring instruments (the so-called Target of Opportunity, ToO). Due to transient nature of such enhanced states it is essential to quickly evaluate if such a ToO event should be followed. We use a machine learning method to assess the possibility to predict the evolution of high flux states in gamma-ray band observed with Fermi-LAT in context of following such alerts with current and future Cherenkov telescopes. We probe flux and Test Statistic predictions using different training schemes and sample selections. We conclude that a partial prediction of the flux over a time scale of one day with an accuracy of ~35% is possible. The method provides accurate predictions of the raising/falling emission trend with 60 - 75% probability, however deeper investigations shows that this is likely based on typical properties of the source, rather than on the result of most recent measurements.
We present high-angular-resolution radio observations of the Arches cluster in the Galactic centre, one of the most massive young clusters in the Milky Way. The data were acquired in two epochs and at 6 and 10 GHz with the Karl G. Jansky Very Large Array (JVLA). The rms noise reached is three to four times better than during previous observations and we have almost doubled the number of known radio stars in the cluster. Nine of them have spectral indices consistent with thermal emission from ionised stellar winds, one is a confirmed colliding wind binary (CWB), and two sources are ambiguous cases. Regarding variability, the radio emission appears to be stable on timescales of a few to ten years. Finally, we show that the number of radio stars can be used as a tool for constraining the age and/or mass of a cluster and also its mass function.
Log in to be able to interact and post comments
comments
Fetching comments
Sign in to be able to follow your search criteria
