We have spent more than a century building physical models of stars. However, like in many physical models, which include (for example) the models of QCD, cosmological structure formation, and the Earth's atmosphere, the physical models disagree with the data in structured and repeatable ways, and we have enormous amounts of data to discover, test, and measure these disagreements. In many cases, we can combine good physical knowledge with good data to build a predictive model that is more powerful than any model built with either used on its own. Here I demonstrate these general points with the specific example of red-giant stars, where our data-driven model is now delivering more precise measurements of detailed stellar chemical compositions (the products of stellar nucleosynthesis) than any purely physical model.
Physics Colloquium: "Data-driven Models of Stars" David Hogg (NYU)
January 17, 2017
4:00PM - 5:00PM
PRB 1080
Add to Calendar
2017-01-17 17:00:00
2017-01-17 18:00:00
Physics Colloquium: "Data-driven Models of Stars" David Hogg (NYU)
We have spent more than a century building physical models of stars. However, like in many physical models, which include (for example) the models of QCD, cosmological structure formation, and the Earth's atmosphere, the physical models disagree with the data in structured and repeatable ways, and we have enormous amounts of data to discover, test, and measure these disagreements. In many cases, we can combine good physical knowledge with good data to build a predictive model that is more powerful than any model built with either used on its own. Here I demonstrate these general points with the specific example of red-giant stars, where our data-driven model is now delivering more precise measurements of detailed stellar chemical compositions (the products of stellar nucleosynthesis) than any purely physical model.
PRB 1080
OSU ASC Drupal 8
ascwebservices@osu.edu
America/New_York
public
Date Range
Add to Calendar
2017-01-17 16:00:00
2017-01-17 17:00:00
Physics Colloquium: "Data-driven Models of Stars" David Hogg (NYU)
We have spent more than a century building physical models of stars. However, like in many physical models, which include (for example) the models of QCD, cosmological structure formation, and the Earth's atmosphere, the physical models disagree with the data in structured and repeatable ways, and we have enormous amounts of data to discover, test, and measure these disagreements. In many cases, we can combine good physical knowledge with good data to build a predictive model that is more powerful than any model built with either used on its own. Here I demonstrate these general points with the specific example of red-giant stars, where our data-driven model is now delivering more precise measurements of detailed stellar chemical compositions (the products of stellar nucleosynthesis) than any purely physical model.
PRB 1080
Center for Cosmology and AstroParticle Physics (CCAPP)
ccapp@osu.edu
America/New_York
public