Using machine learning to explore the long-term evolution of IGR J17091-3624

Project Description

Black hole X-ray binaries are systems in which a normal star transfers matter onto a stellar-mass black hole, producing bright X-ray emission that changes over time. Most systems show relatively stable spectral states during outbursts, but IGR J17091-3624 is exceptional because it displays an unusually rich variety of X-ray variability patterns not seen in many sources. Previous studies classified these patterns into different variability classes based on X-ray count rates and colour properties, relying partly on human judgment. In this project, the student will analyze the archival data of IGR J17091-3624 and characterize its dynamic behavior using machine learning. This project will introduce the student to X-ray astronomy, modern time-series analysis, and machine learning, while exploring whether the known variability classes can be understood as combinations of broader modes of behaviour.