QuoteOriginally posted by: jawabeanwhy would there be a book on MC and C++? what's added value in "C++" part in this case? what's special about C++ implementations?basically what's the premium over a book "Monte Carlo"?We wanted a MC++ It should be obvious.C++ is the most natural language for this kind of problem in our opinion. Performance and extendibility. And it needs to be linked up with _legacy_ systems (backwards compatibity). It's the closest shot to a language that is suitable for scientific computation. You can use OOP, GP and modular models. An important issue was that we provide code so that the reader does not have to code his own stuff based on algorithmic or pseudocode input.The alternatives are C#, Matlab and Java. Although C# is OK for PDEs I would not like it for MC one reason being that you have to do everything from scratch. And it's about 20% slower than C++.Neither Joerg nor myself have been approached to do it in Java. I have not done Java since 1996 so I cannot say if it suitable. Other options might be Erlang, F# and Ocaml but they are bit exclusive. C? well no.Most people still want C++ as far as we can see. C++ is really flexible (templates and later all the maths stuff in boost).Finally, we wanted to write a book about the theory, numerics _and_ code so that people can run the programs themselves with their own parameters. The rationale is in the bespoke Preface.That's all I suppose
Last edited by Cuchulainn
on May 31st, 2009, 10:00 pm, edited 1 time in total.