15

More specifically, i'm interested in 8th order Dormand-Prince embedded method, it's based on Runge-Kutta, and stiff equations.

I use Numerical Recipes 3 but i often have trouble compiling their libraries. I'd like to know of alternatives.

kirill_igum
  • 3,953
  • 5
  • 47
  • 73

2 Answers2

17

You can also try odeint. It has the classical Runge-Kutta solvers, Rosenbrock4 for stiff systems and some multi-step method. It is header-only, but you need the boost libraries.

headmyshoulder
  • 6,240
  • 2
  • 20
  • 27
  • 4
    Update: Odeint is now part of boost: http://www.boost.org/doc/libs/master/libs/numeric/odeint/doc/html/index.html – Timtro Oct 09 '15 at 15:45
7

The GNU Scientific Library has several differential equation solvers. They have one that uses Prince-Dormand. It's written in C so you shouldn't have trouble compiling it.

David Nehme
  • 21,379
  • 8
  • 78
  • 117
  • 3
    A small note: the DP stepper from GSL is explicit, and probably not that well suited for stiff problems with very different time scales. You may consider testing with an implicit method (GSL has some). – Alexandre C. Oct 01 '11 at 21:15