| 1 |
\name{forcings} |
| 2 |
\alias{forcings} |
| 3 |
\title{ |
| 4 |
Passing Forcing Functions to Models Written in R or Compiled Code. |
| 5 |
} |
| 6 |
\description{ |
| 7 |
A \code{forcing function} is an external variable that is essential to the |
| 8 |
model, but not explicitly modeled. Rather, it is imposed as a time-series. |
| 9 |
Thus, if a model uses forcing variables, their value at each time point |
| 10 |
needs to be estimated by interpolation of a data series. |
| 11 |
} |
| 12 |
|
| 13 |
\details{ |
| 14 |
|
| 15 |
The \code{forcing functions} are imposed as a data series, that contains |
| 16 |
the values of the forcings at specified times. |
| 17 |
|
| 18 |
Models may be defined in compiled C or FORTRAN code, as well as in R. |
| 19 |
|
| 20 |
If the model is defined in \emph{R code}, it is most efficient to: |
| 21 |
|
| 22 |
1. define a function that performs the linear interpolation, |
| 23 |
using \R's \code{\link{approxfun}}. It is generally recommended to use |
| 24 |
\code{rule = 2}, such as to allow extrapolation outside of the time interval, |
| 25 |
especially when using the Livermore solvers, as these may exceed the last |
| 26 |
time point. |
| 27 |
|
| 28 |
2. call this function within the model's derivative function, |
| 29 |
to interpolate at the current timestep. |
| 30 |
|
| 31 |
See first example. |
| 32 |
|
| 33 |
If the models are defined in \emph{compiled C or FORTRAN code}, it is possible to |
| 34 |
use \code{deSolve}s forcing function update algorithm. This is the |
| 35 |
compiled-code equivalent of \code{approxfun} or \code{approx}. |
| 36 |
|
| 37 |
In this case:\cr |
| 38 |
1. the forcing function data series is provided by means |
| 39 |
of argument \code{forcings}, |
| 40 |
|
| 41 |
2. \code{initforc} is the name of the forcing function initialisation function, |
| 42 |
as provided in \file{dllname}, while |
| 43 |
|
| 44 |
3. \code{fcontrol} is a list used to finetune how the forcing update should |
| 45 |
be performed. |
| 46 |
|
| 47 |
The \bold{fcontrol} argument is a list that can supply any of the following |
| 48 |
components (conform the definitions in the \link[stats]{approxfun} function): |
| 49 |
\describe{ |
| 50 |
\item{method }{specifies the interpolation method to be used. |
| 51 |
Choices are \code{"linear"} or \code{"constant"},} |
| 52 |
\item{rule }{an integer describing how interpolation is to take place |
| 53 |
outside the interval [min(times), max(times)]. |
| 54 |
If \code{rule} is \code{1} then an error will be triggered and the |
| 55 |
calculation will stop if \code{times} extends the interval of the |
| 56 |
forcing function data set. If it is \code{2}, the \bold{default}, the |
| 57 |
value at the closest data extreme is used, a warning will be printed if |
| 58 |
\code{verbose} is \code{TRUE}, |
| 59 |
|
| 60 |
Note that the default differs from the \code{approx} default.} |
| 61 |
\item{f }{For \code{method = "constant"} a number between \code{0} and |
| 62 |
\code{1} inclusive, indicating a compromise between left- and |
| 63 |
right-continuous step functions. If \code{y0} and \code{y1} are the |
| 64 |
values to the left and right of the point then the value is |
| 65 |
\code{y0 * (1 - f) + y1 * f} so that \code{f = 0} is right-continuous and |
| 66 |
\code{f = 1} is left-continuous, |
| 67 |
} |
| 68 |
\item{ties }{Handling of tied \code{times} values. Either a function |
| 69 |
with a single vector argument returning a single number result or the |
| 70 |
string \code{"ordered"}. |
| 71 |
|
| 72 |
Note that the default is \code{"ordered"}, hence the existence of ties will |
| 73 |
NOT be investigated; in the \code{C} code this will mean that -if ties |
| 74 |
exist, the first value will be used; if the dataset is not ordered, |
| 75 |
then nonsense will be produced. |
| 76 |
|
| 77 |
Alternative values for \code{ties} are \code{mean}, \code{min} etc |
| 78 |
} |
| 79 |
} |
| 80 |
The defaults are: |
| 81 |
|
| 82 |
\code{fcontrol = list(method = "linear", rule = 2, f = 0, ties = "ordered")} |
| 83 |
|
| 84 |
Note that only ONE specification is allowed, even if there is more than |
| 85 |
one forcing function data set. |
| 86 |
|
| 87 |
More information about models defined in compiled code is in the package |
| 88 |
vignette ("compiledCode"). |
| 89 |
} |
| 90 |
\note{ |
| 91 |
How to write compiled code is described in package vignette |
| 92 |
\code{"compiledCode"}, which should be referred to for details. |
| 93 |
|
| 94 |
This vignette also contains examples on how to pass forcing functions. |
| 95 |
} |
| 96 |
\author{ |
| 97 |
Karline Soetaert, |
| 98 |
|
| 99 |
Thomas Petzoldt, |
| 100 |
|
| 101 |
R. Woodrow Setzer |
| 102 |
} |
| 103 |
\seealso{ |
| 104 |
\code{\link{approx}} or \code{\link{approxfun}}, the \R function, |
| 105 |
|
| 106 |
\code{\link{events}} for how to implement events. |
| 107 |
} |
| 108 |
\examples{ |
| 109 |
## ============================================================================= |
| 110 |
## FORCING FUNCTION: The sediment oxygen consumption example - R-code: |
| 111 |
## ============================================================================= |
| 112 |
|
| 113 |
## Forcing function data |
| 114 |
Flux <- matrix(ncol=2,byrow=TRUE,data=c( |
| 115 |
1, 0.654, 11, 0.167, 21, 0.060, 41, 0.070, 73,0.277, 83,0.186, |
| 116 |
93,0.140,103, 0.255, 113, 0.231,123, 0.309,133,1.127,143,1.923, |
| 117 |
153,1.091,163,1.001, 173, 1.691,183, 1.404,194,1.226,204,0.767, |
| 118 |
214, 0.893,224,0.737, 234,0.772,244, 0.726,254,0.624,264,0.439, |
| 119 |
274,0.168,284 ,0.280, 294,0.202,304, 0.193,315,0.286,325,0.599, |
| 120 |
335, 1.889,345, 0.996,355,0.681,365,1.135)) |
| 121 |
|
| 122 |
parms <- c(k=0.01) |
| 123 |
|
| 124 |
times <- 1:365 |
| 125 |
|
| 126 |
## the model |
| 127 |
sediment <- function( t, O2, k) |
| 128 |
list (c(Depo(t) - k * O2), depo = Depo(t)) |
| 129 |
|
| 130 |
# the forcing functions; rule = 2 avoids NaNs in interpolation |
| 131 |
Depo <- approxfun(x = Flux[,1], y = Flux[,2], method = "linear", rule = 2) |
| 132 |
|
| 133 |
Out <- ode(times = times, func = sediment, y = c(O2 = 63), parms = parms) |
| 134 |
|
| 135 |
## same forcing functions, now constant interpolation |
| 136 |
Depo <- approxfun(x = Flux[,1], y = Flux[,2], method = "constant", |
| 137 |
f = 0.5, rule = 2) |
| 138 |
|
| 139 |
Out2 <- ode(times = times, func = sediment, y = c(O2 = 63), parms = parms) |
| 140 |
|
| 141 |
mf <- par(mfrow = c(2, 1)) |
| 142 |
plot (Out, which = "depo", type = "l", lwd = 2, mfrow = NULL) |
| 143 |
lines(Out2[,"time"], Out2[,"depo"], col = "red", lwd = 2) |
| 144 |
|
| 145 |
plot (Out, which = "O2", type = "l", lwd = 2, mfrow = NULL) |
| 146 |
lines(Out2[,"time"], Out2[,"O2"], col = "red", lwd = 2) |
| 147 |
|
| 148 |
## ============================================================================= |
| 149 |
## SCOC is the same model, as implemented in FORTRAN |
| 150 |
## ============================================================================= |
| 151 |
|
| 152 |
out<- SCOC(times, parms = parms, Flux = Flux) |
| 153 |
|
| 154 |
plot(out[,"time"], out[,"Depo"], type = "l", col = "red") |
| 155 |
lines(out[,"time"], out[,"Mineralisation"], col = "blue") |
| 156 |
|
| 157 |
## Constant interpolation of forcing function - left side of interval |
| 158 |
fcontrol <- list(method = "constant") |
| 159 |
|
| 160 |
out2 <- SCOC(times, parms = parms, Flux = Flux, fcontrol = fcontrol) |
| 161 |
|
| 162 |
plot(out2[,"time"], out2[,"Depo"], type = "l", col = "red") |
| 163 |
lines(out2[,"time"], out2[,"Mineralisation"], col = "blue") |
| 164 |
|
| 165 |
|
| 166 |
\dontrun{ |
| 167 |
## ============================================================================= |
| 168 |
## show examples (see respective help pages for details) |
| 169 |
## ============================================================================= |
| 170 |
|
| 171 |
example(aquaphy) |
| 172 |
|
| 173 |
## show package vignette with tutorial about how to use compiled models |
| 174 |
## + source code of the vignette |
| 175 |
## + directory with C and FORTRAN sources |
| 176 |
vignette("compiledCode") |
| 177 |
edit(vignette("compiledCode")) |
| 178 |
browseURL(paste(system.file(package = "deSolve"), "/doc", sep = "")) |
| 179 |
} |
| 180 |
|
| 181 |
} |
| 182 |
|
| 183 |
\keyword{utilities} |