3 Dec
2008
3 Dec
'08
4:22 p.m.
2008/12/3 Peter Verswyvelen
Okay. It would be nice to experiment with higher order integration techniques too. Right now Reactive uses Euler integration but the user should be able to choose between Euler, Midpoint, Simpson (or whatever other quadrature rule). However, aren't we actually interested in approximating ordinary differential equations (or more exactly, initial value problems), instead of using integral directly? In which case we should choose between 1-th, 2nd, or 4th order Runge Kutta or something? I'm not an expert here, but it would be nice to think about this I guess.
I think a 4th order Runge Kutta is something we'd want to have, maybe as part of a FRP.Reactive.Physics package?