I am trying to fit a non-linear model using nls function. The model has a conditioning and works by groups therefore I try to implement a loop which covers over 300 combinations of starting values. I am using tryCatch but to my surprise the loop crashes, I guess because of the starting values or lower and upper bounds.
tryCatch(
for(r in 1:nrow(st4)){
halfLE3[[r]] <- nls(
inty ~ I(time < (position/velocity) + dinitial) * Inty_S0 +
(time >= (position/velocity) + dinitial) *
I(Intyf[probe] + (Inty_S0[probe] - Intyf) *
(exp(-Decay * (time - (position/velocity) + dinitial)))),
data = Data4,
algorithm = "port",
control = list(warnOnly = TRUE),
start = list(Decay=st4[r,3], dinitial=st4[r,2],
Intyf=rep(st4[r,4], length(levels(Data4$probe))),
Inty_S0=rep(st4[r,5], length(levels(Data4$probe))),
velocity=st4[r,1]),
lower = list(Decay= .1, dinitial=0.1, velocity=10, Intyf=0.1, inty_S0=.5),
upper = list(Decay= 1, dinitial=10, velocity=8000, Intyf=1, inty_S0=1.5)
)
}
,error = function(e) {e}
)
st4 is a dataframe with 300 combinations of starting values. probe refers to group.
Do you have any idea why the loop crashes even using the tryCatch? Do you have any idea what I can improve? I used the nls.control changing the maxiter and other parameters but works same as the control I use here above.
I would be very grateful for any suggestion.
