NAME
RungeKutta_stepConfined
SYNOPSIS
call RungeKutta_stepConfined (character (len=*), intent (in) :: method,
function, intent (in) :: f,
integer, intent (in) :: nc,
real, intent (in) :: x,
real, intent (in) :: y (:),
function, intent (in) :: ymin,
function, intent (in) :: ymax,
real, intent (in) :: eFrac,
real, intent (in) :: eBase (:),
real, intent (in) :: htry,
real, intent (out) :: hused,
real, intent (out) :: hnext,
real, intent (out) :: yout (:),
real, intent (out) :: eout (:))
DESCRIPTION
This routine performs a single confined Runge Kutta step. It copies the appropriate
Butcher tableau data from the repository and calls the corresponding RK stepper. The details
of the ODE function are contained in the passed array function 'f', which constitutes
the link of the general Runge Kutta module to the individual applications. Thus 'f' has
to be provided by an application outside of the Runge Kutta unit.
The error vector for the dependent variables is composed of two pieces (as suggested
by Numerical Recipies):
e (i) = eFrac * eBase (i)
In this error vector definition, eFrac denotes the fractional value each of the
individual eBase (i) values that define the total allowed maximum error for each
of the i-th dependent variables.
The code performs a confined RK step. The confinement boundaries are (potentially)
dependent on the current value of the dependent variables at the intermediate and final
RK points. This boundary dependency has to be transmitted via via 2 functions 'ymin' and
'ymax', which accept the current dependent variables array as argument input. The user
has to provide the external definition of these 2 functions.
Note that the layout of the dependent variable array can always be done such that the
confined variables come first. The user must therefore be careful to design its ODE
function in such a way that those confined dependent variables are at the beginning
of the dependent variable array.
ARGUMENTS
method : the type of RK method to be applied f : the ODE function containing details of the ODE system to be solved nc : the number of confined dependent variables (must be <= # dep variables) x : the initial value of the independent variable y : the initial values of the dependent variable(s) ymin : the lower bound confinement function (as function of dependent variables) ymax : the upper bound confinement function (as function of dependent variables) eFrac : the fractional value for each of the error base values eBase : the error base values for each of the dependent variables htry : the initial step size on the independent variable hused : the final step size applied on the independent variable hnext : the suggested next step size on the independent variable yout : the values of the dependent variable(s) after the accepted RK45 step eout : the returned error of all dependent variable(s)
NOTES
1) Contains an interfaced function in the declarations (see below).
2) The sizes of the arrays eBase, yout and eout can be larger than needed, i.e.,
they must not match exactly the number of dependent variables. Likewise with
the sizes of the boundary arrays ymin and ymax, which can be greater than
the number of confined dependent variables.