[R-sig-dyn-mod] Turbidostat Model

[Thomas Petzoldt]

Hello Uta,

deSolve provides several mechanisms to approach this. I would suggest to 
use forcing functions:

signal <- approxfun(1:10, rep(c(1, 0.1), each=5))

growth <- function(t, x, parms) {
   korr <- signal(t) # interpolates linearly
   with(as.list(c(x, parms)), {
     dy <- r * korr * x
     list(dy)
   })
}

times <- seq(1, 10, 0.1)
parms <- c(r=1)
y <- 1
out <- ode(y, times, growth, parms)
plot(out)



This interpolates linearly between the values specified in the signal 
and you can, of course, also use other interpolation rules as well.

Alternatively, abrupt changes can also be modelled with events. Then, 
korr has to be an additional state variable with derivative zero (dkorr 
<-0) and events are used to manipulate it as needed.

Thomas Petzoldt


-- 
Dr. Thomas Petzoldt
Technische Universitaet Dresden
Fakulty of Forest, Geo and Hydro Sciences
Institute of Hydrobiology
01062 Dresden, Germany

http://tu-dresden.de/Members/thomas.petzoldt