Distinguishing short and longFermigamma-ray bursts

Abstract

Two classes of gamma-ray bursts (GRBs), short and long, have been determined without any doubts, and are usually ascribed to different progenitors, yet these classes overlap for a variety of descriptive parameters. A subsample of 46 long and 22 short Fermi GRBs with estimated Hurst Exponents (HEs), complemented by minimum variability time-scales (MVTS) and durations (T 90 ) is used to perform a supervised Machine Learning (ML) and Monte Carlo (MC) simulation using a Support Vector Machine (SVM) algorithm. It is found that while T 90 itself performs very well in distinguishing short and long GRBs, the overall success ratio is higher when the training set is complemented by MVTS and HE. These results may allow to introduce a new (non-linear) parameter that might provide less ambiguous classification of GRBs.

On the relationship between the Hurst exponent, the ratio of the mean square successive difference to the variance, and the number of turning points

“…Among many existing computational algorithms for estimating H (Rescaled Range Analysis (R/S) [5,11,12,20], Detrended Fluctuation Analysis (DFA) [21,22,23,24], wavelet approach [8,9,25,26], Detrended Moving Average (DMA) [27,28,29], to mention only a few), herein their extraction was performed by the wavelet approach with the Haar wavelet as a basis. The Haar wavelet was chosen due to its simplicity, most compact support and equivalence to the Allan variance [30].…”

Graph-based clustering of gamma-ray bursts

“…BATSE) do not rule out the possibility of three being present. This observation that bigger samples are more liberal when it comes to accepting the possibility of a third class was also observed when using mixture models (Tarnopolski 2019b). The second major result is that there are no hints of more than three groups (despite some works implying even five or seven).…”

“…The pictures painted by the remaining four data sets (Fermi, Konus, Swift, and RHESSI) reveal a gradual transition of the domination of two over three groups, with the cases of Fermi and Konus resembling the results for BATSE, while Swift and RHESSI are more similar to Suzaku. This seems to be an effect of the sample size (Tarnopolski 2019b), that is, in less numerous data sets there are simply not enough points to highlight more than two groups. Overall, the results point at either two or three clusters, hence do not conform with the works that find more GRB classes, for example five or seven (Acuner & Ryde 2018;Ruffini et al 2018).…”

“…For each instrument, fluences in different energy bands are available, hence the definitions of H 32 are as follows: H 32 = F 50−100 keV F 25−50 keV for Swift; H 32 = F 200−750 keV F 50−200 keV for Konus; H 32 = F 120−1500 keV F 25−120 keV for RHESSI; and H 32 = F 240−520 keV F 110−240 keV for Suzaku. These were examined with mixture models as well (Tarnopolski 2019b); compared to this previous work, herein two duplicates in the Suzaku sample were removed, one duplicate in the case of RHESSI, and three outliers (with log H 32 < −0.3) and two duplicates from Swift.…”

“…Investigating the two-dimensional realm of hardness ratios and durations has led to analogous conclusions: Gaussian mixtures have often pointed at three groups (Horváth et al 2006;Řípa et al 2009;Horváth et al 2010;Veres et al 2010;Horváth et al 2018), but several works have indicated that only two are required ( Řípa et al 2012;Yang et al 2016;Tarnopolski 2019b;von Kienlin et al 2020). Moreover, considering skewed distributions, two-component mixtures have also been indicated (Tarnopolski 2019b,a).…”