R help - Feb 2008 - Catching NaNs from pweibull()

Hello,

I am working with the nls() function and inserting a formula into it that
uses the pweibull function. However the pweibull function is annoyingly
producing NaNs, which nls() refuses to handle. I have put a sample of the
code below. Is there a way to prevent these NaNs from interfering, for
example a method to catch them? I get the following error when I try to run
the code: 

res.nls <- nls(cumsumcaught ~
Pcatch*Prepens.released*(1-pweibull(distances,k,l)),
+ start = list(k=1,l=12), trace=T, control = my.control)
Error in numericDeriv(form[[3]], names(ind), env) : 
        Missing value or an infinity produced when evaluating the model
In addition: Warning messages:
1: NaNs produced in: pweibull(q, shape, scale, lower.tail, log.p) 
2: NaNs produced in: pweibull(q, shape, scale, lower.tail, log.p) 

HERE IS THE CODE:

# Observed data
Prepens.caughtintraps = c(94,45,71,64,71,31,16,19,27,16,113)  

# Numbers of seeds released
Prepens.released = 1250  

distances = c(2,3,4,5,7.5,10,12.5,15,17.5,20,25)
totalseedscaught = sum(Prepens.caughtintraps)
Pcatch = totalseedscaught/Prepens.released

i=11
cumsumcaught = numeric(11)
while (i>0){
	cumsumcaught[i] <- sum(Prepens.caughtintraps[i:11])
	i=i-1
}

res.nls <- nls(cumsumcaught ~
Pcatch*Prepens.released*(1-pweibull(distances,k,l)),
start = list(k=1,l=12), trace=T)

Regards,
Rob



-- 
View this message in context:
http://www.nabble.com/Catching-NaNs-from-pweibull%28%29-tp15356718p15356718.html
Sent from the R help mailing list archive at Nabble.com.

That's odd, it works just fine for me
(R 2.6.1, i486-pc-linux-gnu)

  Can you send the results of sessionInfo() ?
 
# Observed data
Prepens.caughtintraps = c(94,45,71,64,71,31,16,19,27,16,113)  

# Numbers of seeds released
Prepens.released = 1250  

distances = c(2,3,4,5,7.5,10,12.5,15,17.5,20,25)
totalseedscaught = sum(Prepens.caughtintraps)
Pcatch = totalseedscaught/Prepens.released

## minor tweak: this is more efficient, gives same answer
cumsumcaught <- rev(cumsum(rev(Prepens.caughtintraps)))

res.nls <- nls(cumsumcaught ~
               Pcatch*Prepens.released*(1-pweibull(distances,k,l)),
               start = list(k=1,l=12), trace=TRUE)

## slightly more accurate -- use lower.tail instead of 1-pweibull()
## but makes no difference in this case
res.nls <- nls(cumsumcaught ~
               Pcatch*Prepens.released*
               pweibull(distances,k,l,lower.tail=FALSE),
               start = list(k=1,l=12), trace=TRUE)


13122.82 :   1 12 
11783.76 :   1.045207 12.668148 
11774.70 :   1.035480 12.676250 
11774.54 :   1.036564 12.679938 
11774.54 :   1.036398 12.679737 
11774.54 :   1.036420 12.679783 
11774.54 :   1.036417 12.679778 

Nonlinear regression model
  model:  cumsumcaught ~ Pcatch * Prepens.released * pweibull(distances,      k,
l, lower.tail = FALSE) 
   data:  parent.frame() 
     k      l 
 1.036 12.680 
 residual sum-of-squares: 11775

Number of iterations to convergence: 6 
Achieved convergence tolerance: 1.388e-06

Hello Ben,

Ok this is going to sound very bizare, but the code now works. I struggled
for a long time to see why it was returning NaN values. I'm sure the code is
exactly the same. But anyway it now seems to be working (for the time
being). Thanks for the input though.

Regards,
Rob




-- 
View this message in context:
http://www.nabble.com/Catching-NaNs-from-pweibull%28%29-tp15356718p15359644.html
Sent from the R help mailing list archive at Nabble.com.