Gamma-ray bursts are transient, extremely bright flashes of gamma radiation. Their origin is attributed to a rapid fall-back of matter onto a newly formed black hole. Two classes of GRBs are firmly established: short GRBs, lasting up to 2 seconds, whose origin is a compact-object merger (a neutron star-neutron star or neutron star-black hole system), and long GRBs, with durations greater than 2 seconds, coming from the gravitational collapse of massive stars (of Wolf-Rayet or blue super giant types). Long GRBs are typically more luminous than short ones, have lower hardness ratio, are detected on greater redshifts and are hosted by star forming galaxies. Several subclasses have been also identified: X-ray flashes (when the X-ray radiation dominates over gamma radiation), ultra-long GRBS (lasting even 10,000 seconds), short GRBs with extended emission (having all properties characteristic of regular short GRBs except for an extended duration lasting up to 100 seconds). Nevertheless, there are two main physical mechanisms (i.e., mergers and collapsars) giving rise to a GRB.

However, several studies have claimed that a third class, intermediate in duration and softer than long GRBs, is present in the data gathered by several space telescopes (CGRO/BATSE, Swift/BAT, Fermi/GBM, Konus/Wind, RHESSI). Such a claim was put forward by fitting the empirical distributions of several observables with a mixture of Gaussian distributions (uni- and multivariate), and observing that at least three components are required to describe the data adequately. However, the underlying true distributions need not be Gaussian, but rather skewed (asymmetric) ones, coming from, e.g., an asymmetric distribution of envelope masses of the progenitors, introducing spurious components when modelled by Gaussian ones. I have indeed found that to be true, i.e. when skewed components are considered, only two of them are required to explain the observational data satisfactorily, making the existence of the presumed third class unnecessary and unlikely to be a real phenomenon.

The yet unanswered question is where does this asymmetry come from: is it a result of a real asymmetry present in the physical properties of the progenitors, or is it a statistical feature that could be explained by a composition of several factors? The research I am conducting is based on novel, rigorous analyses of available data, statistical and numerical simulations. One of the possible explanations could have been given by examining the impact of the redshift distribution of the sources on the temporal and energetic properties of GRBs. It turned out, though, that the cosmological dilation accounts for an insignificant portion of the observed skewness. Therefore, its source has to lie in the physical properties of the progenitors.

I also employ techniques stemming from nonlinear dynamics and time series analysis to aid the identification and classification of GRBs, e.g. the Hurst exponent or, the recently developed by myself, A—T plane.

Figure 1. (A) The distribution of durations of Fermi GRBs. Rows, from top to bottom: Gaussian distribution, skew normal distribution, sinh-arcsinh distribution, alpha-skew normal distribution. The best fit is obtained with a two-component mixture of skew-normal distribution [subpanel (c)]. (B) Bivariate fits in the duration-hardness ratio plane of Fermi GRBs. The best fit is attained by a two-component mixture of skewed Student-t distributions (2ST).

The research conducted on this topic in the years 2018-2021 was performed within a National Science Center (Narodowe Centrum Nauki, NCN) OPUS grant No. 2017/25/B/ST9/01208.