Automatic Differentiation and its Applications

Speaker:
Paul Hovland
Argonne National LaboratoryAbstractAutomatic differentiation is a technique for computing the derivatives of a function defined by a computer subprogram. We provide an introduction to automatic differentiation tools and theory and describe the role of current and next generation automatic differentiation tools in large-scale optimization, sensitivity analysis, and the solution of nonlinear PDEs. We focus on the computational costs of automatic differentiation and the trend toward tools that are more robust, easier to use, and more powerful. We conclude with a short description of our research agenda, with an emphasis on the motivational role played by current and future applications.

Location: Kaprielian Hall (KAP) - rielian Hall, Room 203

Audiences: Everyone Is Invited

Contact: Evangeline Reyes