Combining Computer Models for Stellar Evolution: From Sun-Like Stars to White Dwarf Cinders

• Activity Number: 377
• Type: Invited
• Date/Time: Tuesday, July 31, 2012 : 2:00 PM to 3:50 PM
• Sponsor: Section on Physical and Engineering Sciences
• Author(s): David A. van Dyk*+ and Nathan Stein and Ted von Hippel and Steven DeGennaro and Elizabeth J Jeffery and William H Jefferys
• Companies: Imperial College London and Harvard University and Embry-Riddle Aeronautical University and The University of Texas at Austin and Space Telescope Science Institute and The University of Texas at Austin/The University of Vermont
• Keywords: Computer Models ; Astronomy ; Model Building ; MCMC ; Parallel Computing

Abstract

Sophisticated computer models are used to describe the complex physics involved in stellar evolution. Like a likelihood, these models predict observed quantities given unknown parameters. Including them as components of a multi-level model, however, leads to significant modeling, inferential, and computational challenges. Although understanding the mass loss that stars experience as they age is essential for realistic physical models of stellar evolution, there remains substantial uncertainty in the relationship between the initial mass of a sun-like star and its final mass as a white dwarf. We address this by formulating the so-called initial-final mass relation (IFMR) which we use as a parametric link between the computer models for sun-like stars and white dwarfs. We use a combination of sophisticated MCMC and simple emulators of the computer models to tackle the computational challenges. This strategy allows us to apply the full force of powerful statistical tools to build, fit, check, and improve the statistical models and their computer model components. In contrast to traditional methods, we can estimate the uncertainty in our fit using principled statistical techniques.