A formal method for identifying distinct states of variability in time-varying sources: SgrA* as an example

Abstract

Continuously time variable sources are often characterized by their power spectral density and flux distribution. These quantities can undergo dramatic changes over time if the underlying physical processes change. However, some changes can be subtle and not distinguishable using standard statistical approaches. Here, we report a methodology that aims to identify distinct but similar states of time variability. We apply this method to the Galactic supermassive black hole, where 2.2 um flux is observed from a source associated with SgrA*, and where two distinct states have recently been suggested. Our approach is taken from mathematical finance and works with conditional flux density distributions that depend on the previous flux value. The discrete, unobserved (hidden) state variable is modeled as a stochastic process and the transition probabilities are inferred from the flux density time series. Using the most comprehensive data set to date, in which all Keck and a majority of the publicly available VLT data have been merged, we show that SgrA* is sufficiently described by a single intrinsic state. However the observed flux densities exhibit two states: a noise-dominated and a source-dominated one. Our methodology reported here will prove extremely useful to assess the effects of the putative gas cloud G2 that is on its way toward the black hole and might create a new state of variability.