Matlab 2008b VC++ ODE45

I am trying to utilize the ODE45 functionality of Matlab in Visual C++ 2010. Searching the Internet, I see that there used to be a file "matlab.hpp" that could be included in a C++ project, but that in R2008b (the version of Matlab I am using), this is not an option. I did try searching Matlab Central for "Matlab 2008 C++ ODE45" and got no hits, so either that was too specific or there really is no solution already posted. If anyone knows how to do this (use ODE45 in C++), I would appreciate the help.

Thanks in advance.

[Paul Huter]

Further looking into things...still think utilizing the Matlab code in C++ will be the simplest solution. I have a numerical coding book that shows me how to implement RK4, and like I said, I've found multiple "answers" online. But the book I'm using as a guide walks me through the steps using ODE45 in Matlab for solving a system of equations. I kind of looked at the ODE45.m file, just not sure how to use it with C++ (is this even possible?).

Thanks.

Answer 1

[Jan]

Matlab's ODE45 is a stable and well tested implementation of explicit Runge-Kutta method of Dormand-Prince. It is a 4th order method using an embedded 5th order method for the stepsize control.

This method has been implemented in C directly very often. Look for the terms "RK45" and "Dormand-Prince" in the net, e.g. at www.netlib.org. The C-implementation can be 10 times faster than ODE45.m and I do not see advantages of using the Matlab integrator.

[Paul Huter]

Thanks Jan. I am currently using an implementation of RK4, and I know ODE45 uses RK4. I'm not 100% certain I'm doing things correctly though - I found multiple sites that provided different solutions for the same problem, which is similar to a problem I solved in a class (using Matlab for RK4), and all based on a book I am using to solve some things (the problem is integrating acceleration to find velocity and position). Moving further through this same book, it talks about using Matlab/Simulink and ODE45 to solve the integration, which was why I was hoping to implement in C++, since I thought it would be easier to follow an example in the book (which was written for solving the overall general problem I am working), rather than trying to figure out the best way to implement an RK4 based on the numerous possible implementations on the Internet.

However, if you know of a good solution for integrating acceleration (which is partially a function of velocity) to solve for velocity, and also to integrate velocity (which will have been converted to a different coordinate system) to solve for velocity, I would appreciate the help. Like I said, I know I've implemented RK4 before, and I can follow the examples on (the many) websites, but they don't exactly match up to my particular problem, so I'm looking for a more general solution. If you don't have anything, that's fine, I'll keep searching.

Thanks again for the quick response.

Paul.