|
MATH768: Mathematical Modeling I
An exploration of the synthesis of continuum mechanics with stochastic machine learning
Time & Place: MWF 2:30-3:20PM, Phillips Hall Room 385, UNC, Chapel Hill
Instructor:
Sorin Mitran
Website: http://mitran-lab.amath.unc.edu/courses/MATH768/MATH768.xhtml
A two-track presentation of modeling of continua through:
Classical partial differential equation approaches with hypothesized constitutive relation closures
Application of deep learning to extract constitutive equations from data
Classical models (Cauchy elasticity, Navier-Stokes flow) are recovered through deep learning
Deep learning allows consideration of much more complicated behavior such as that encountered in biological media: cytoskeleton, tissue, polymeric flow
Hands-on introduction to TensorFlow, Mathematica neural networks
|
|