NAG scientific software available at IPP - An introduction to algorithmic differentiation to compute exact and cheap derivatives

Viktor Mosenkis: The IPP site wide license covers the usage of the NAG Numerical Libraries and of the NAG Fortran Compiler on any Linux and on any Solaris system by all researchers and people working at IPP. This presentation will give detailed information about the technical benefits of the NAG scientific software and will explain how people at IPP can have access to it and easily use it. An update on the recent new products and features will also be given. .The NAG Numerical Libraries come with more than 1600 robust and performing mathematical and statistical algorithms to help quickly building high quality numerical models, simulations and applications for data analysis. NAG's algorithms can be used directly from most popular programming languages and development environments including Fortran, C/C++, MATLAB, Python, Java, R, C#/.NET, Excel/VB, Delphi, CUDA, etc. .The NAG Fortran Compiler is known to be the best in the world for error checking and for its high compliance with the Fortran standards so this makes it the compiler of choice to produce clean Fortran code. François Cassier: Understand the basics of Algorithmic Differentiation and how to use this technique to compute exact derivatives and speed up the computational methods of solving large problems. You will .See what constitutes Algorithmic Differentiation (AD) techniques .Understand key theoretical uses for sensitivities, optimization and other solvers .Learn why AD can help you to improve accuracy and performance of your derivative computation .See some of the toolsets that can help you access AD .Hear about some of the issues around using AD