Gravitational waves are ripples in spacetime caused by cosmic collisions like those between black holes. Since their discovery in 2015, they've opened a new window to our universe. However, analyzing just one gravitational wave signal using traditional methods takes days if not weeks. This PhD Thesis develops smart computer algorithms that dramatically speed up this analysis. By combining artificial intelligence with existing inference techniques, we can now determine where gravitational waves come from and what properties their sources have in minutes rather than weeks. Our new methods can even untangle overlapping gravitational waves and work well even for smaller black holes, which was previously challenging. As detectors become more sensitive, thousands of gravitational waves will be detected each year. Thanks to this research, scientists can analyze this flood of cosmic signals quickly and reliably, leading to new astronomical discoveries.
Alex Kolmus pursued his academic journey at uu77, where he completed degrees in Chemistry (2014) and Physics (2017) before earning his Master's in Physics (2019). His doctoral research (2019-2024) focused on applying machine learning to gravitational wave astronomy, developing innovative methods that dramatically reduce analysis time from days to minutes. Beyond his academic work, Alex has demonstrated his technical expertise in various contexts. In 2017, he was part of a team that won the Jury Prize at the Volkswagen Deep Learning and Robotics Challenge. He also co-founded GraphKite (2017-2020), where he developed machine learning solutions for the insurance industry and supported startups in implementing data-driven approaches. Alex's research combines astrophysics with cutting-edge machine learning techniques, creating new tools that help scientists interpret the cosmic signals from colliding black holes and neutron stars more efficiently.