Apr 21, 2014

Robert Lindner, UW Madison

"Initial Conditions for Star Formation: Unbiased comparisons of ISM simulations and observations"

Comparing 21cm observations of the Interstellar Medium (ISM) to 

results from numerical simulations is important for constraining physical 

models of the neutral ISM and understanding star formation.  In 

analyzing 21cm HI emission/absorption spectra, the commonly used 

technique of "Gaussian components" provides a good description of 

the physical properties of individual ISM clouds.  However, it can 

be difficult to use the "Gaussian components" model to form unbiased 

conclusions about the ISM because of the significant amount of human 

interaction required in choosing the model's initial parameters.  We 

have produced a new algorithm, called Autonomous Gaussian Decomposition, 

which uses computer vision and machine learning for autonomously 

choosing these initial parameters-- allowing for truly unbiased 

comparisons between observations and simulations.  I will present 

the algorithm, discuss its performance, and display initial results 

in using it to compare recent high-sensitivity 21-SPONGE observations 

to high-resolution hydrodynamic ISM simulations.

Event Details

Apr 21, 2014



UW-Madison Astronomy Home