All Times ET
Keywords: approximate bayesian computation, simulation-based inference, astronomy
Observing the sky in X-rays and gamma-rays allows us to study phenomena not observable at any other wavelengths: massive stellar explosions in the form of Gamma-Ray Bursts, eruptions on highly magnetised neutron stars called magnetar Giant Flares, the fall of matter into stellar-sized and supermassive black holes. Because X-rays do not pass through the Earth’s atmosphere, we study the X-ray sky from space with missions like the Chandra X-ray Observatory, the XMM-Newton Observatory and the Nuclear Spectroscopic Telescope Array (NuSTAR). Many of our instruments are highly optimised for observing faint things, paradoxically making it very difficult to accurately infer physical knowledge from bright sources, which are often strongly affected by instrumental biases. While it is difficult to write down a statistical model for common instrumental biases like dead time and photon pile-up, writing a computer programme that acts as a simulator to generate realistic data is straightforward. Here, I will discuss recent work on employing simulation-based inference to mitigate common instrumental effects in X-ray detectors and show how we can make accurate physical inferences even for some of the brightest X-ray sources in the sky.