Solving Ordinary Differential Equations in R

Both Runge-Kutta and linear multistep methods are available to solve initial value problems for ordinary differential equations in the R packages deSolve and deTestSet. Nearly all of these solvers use adaptive step size control, some also control the order of the formula adaptively, or switch between different types of methods, depending on the local properties of the equations to be solved. We show how to trigger the various methods using a variety of applications pointing, where necessary, to problems that may arise. For instance, many practical applications involve discontinuities. As the integration routines assume that a solution is sufficiently differentiable over a time step, handing such discontinuities requires special consideration. We give examples of how we can implement a nonsmooth forcing term, switching behavior, and problems that include sudden jumps in the dependent variables. Since much computational efficiency can be gained by using the correct method for a particular problem, we end this chapter by providing a few guidelines as to how the most efficient solution method for a particular problem can be found.