The following technical report is available from http://aib.informatik.rwth-aachen.de:
dco/c++ User Guide Uwe Naumann, Klaus Leppkes, and Johannes Lotz AIB 2014-03
dco/c++ is a highly flexible and efficient implementation of first- and higher-order tangent-linear and adjoint Algorithmic Differentiation (AD) by operator overloading in C++. It combines a cache-optimized internal representation based on C++ expression templates with an intuitive and powerful application programmer interface (API). Support for externally differentiated functions and for optimized data flow reversal through checkpointing is provided. dco/c++ has been applied successfully to a number of numerical simulations in the context of, for example, large-scale parameter estimation and shape optimization. Starting with an introduction to the fundamentals of AD this user guide describes the various modes of dco/c++ in terms of their APIs and with the help of very simple examples. dco/c++ is actively developed resulting in a steady evolution of the associated user guide.