On Mon, 2 Jul 2012, Mario Mulansky wrote:
after getting some good amount of positive feedback on our talk at boostcon we are now asking for a formal review process of the odeint library, currently hosted on github: https://github.com/headmyshoulder/odeint-v2
I found your page comparing performance with other Runge-Kutta implementations. http://headmyshoulder.github.com/odeint-v2/performance.html The documentation indicates you have support for various types of stepper. http://headmyshoulder.github.com/odeint-v2/doc/boost_numeric_odeint/odeint_i... Have you considered whether the odeint API could act as an interface to CVODE or CVODES in Sundials? https://computation.llnl.gov/casc/sundials/main.html If that works, then it validates the odeint API. If there are gaps or inefficiencies, then they may indicate improvements for odeint. Thanks, Daniel