Time-resolved GRB polarization with POLAR and GBM. Simultaneous spectral and polarization analysis withsynchrotron emission

Context. Simultaneous gamma-ray measurements of gamma-ray burst spectra and polarization o ff er a unique way to determine the underlying emission mechanism(s) in these objects, as well as probing the particle acceleration mechanism(s) that lead to the observed gamma-ray emission. Aims. We examine the jointly observed data from POLAR and Fermi-GBM of GRB 170114A to determine its spectral and polarization properties, and seek to understand the emission processes that generate these observations. We aim to develop an extensible and statistically sound framework for these types of measurements applicable to other instruments. Methods. We leveraged the existing 3ML analysis framework to develop a new analysis pipeline for simultaneously modeling the spectral and polarization data. We derived the proper Poisson likelihood for gamma-ray polarization measurements in the presence of background. The developed framework is publicly available for similar measurements with other gamma-ray polarimeters. The data are analyzed within a Bayesian probabilistic context and the spectral data from both instruments are simultaneously modeled with a physical, numerical synchrotron code. Results. The spectral modeling of the data is consistent with a synchrotron photon model as has been found in a majority of similarly analyzed single-pulse gamma-ray bursts. The polarization results reveal a slight trend of growing polarization in time reaching values of similar to 30% at the temporal peak of the emission. We also observed that the polarization angle evolves with time throughout the emission. These results suggest a synchrotron origin of the emission but further observations of many GRBs are required to verify these evolutionary trends. Furthermore, we encourage the development of time-resolved polarization models for the prompt emission of gamma-ray bursts as the current models are not predictive enough to enable a full modeling of our current data.