Talk by Paul Hovland, ANL - Wednesday, 24 Jan.

[Margery Ishmael]

Wednesday, 24 January at 2:30 pm in Ryerson (annex) 276

"Foundations of Automatic Differentiation"

                  by Paul Hovland
              Argonne National Laboratory

Abstract: We describe some of the theoretical elements that underpin automatic
differentiation.  Automatic differentiation (AD) is a technique for
augmenting computer subprograms with instructions for the computation of
derivatives.  Effective implementation and use of AD depends upon graph
algorithms, compiler technology, and parallel algorithms in addition to
numerical analysis.  We discuss the implementation and use of automatic
differentiation, drawing upon our experience in the development of the
ADIFOR and ADIC tools.

The talk will be followed by refreshments in Ryerson 255.