R help - Apr 2010 - Using optim function for logistic model simulation

Hello!

I'm a college undergrad desperately trying to finish up my thesis. I
have a dataset on the distribution of a grassland bird from the
Breeding Bird Survey. I have a very straightforward and simple version
of the logistic growth model to describe changes in this bird's
abundance over time. The variables are the natural birth rate,
mortality, and carrying capacity. The equation itself looks like this:
birds <- c*birds + b*birds*(1-birds/K) where c = mortality, b = birth
rate, and K = carrying capacity.

I'm trying to find the optimal value of K and I've been using the
optim function. The problem is that it doesn't really work. My code is
below:

b<-1.22
c<-0.55
bird<-bird.density[0]
eqn1<- function(K1, bird) {
	for (i in 1:8) {
		b<-1.22
		c<-0.55
		bird <- 0.55*bird + b*bird*1-b*bird*bird/K
	}
}

k1<-optim(c(0,10),eqn)

I realize this is a painfully amateurish question, but I really
appreciate any help! The other tricky part about this is that it looks
like there are 3 distinct time periods where the bird is increasing,
declining, and increasing again. I'm thinking that there are 3
different carrying capacity values, but my understanding of optim is
that it can only optimize one parameter variable at a time. So I was
going to take the brute force approach of artificially breaking up the
equation into 3 parts (i.e. make eqn1, eqn2, and eqn3, and optimize
each separately, with the input values for the later two functions
coming from the last number of birds from the previous loop).

Thank you all very much!

charlotte

Hi Charlotte ,

I can't reproduce your code, but skimming through it -
It would appear that:
1) in
eqn1<- function(K1, bird)
you didn't define "bird" (you did define it before the function,
so I'd
suggest just removing it from the function call like this:
eqn1<- function(K1)

2) you didn't "return" and value at the end of the function.

3) you use different name in optim then in the function you made.

4) I see no point for you to use the for loop.



Consider trying the following code:

b<-1.22
c<-0.55
bird<-bird.density[0] # I assume this exists
eqn<- function(K1, bird) {
               b<-1.22
               c<-0.55
               bird <- 0.55*bird + b*bird*1-b*bird*bird/K
            return(bird)
}

k1<-optim(c(0,10),eqn)



----------------Contact
Details:-------------------------------------------------------
Contact me: Tal.Galili@gmail.com |  972-52-7275845
Read me: www.talgalili.com (Hebrew) | www.biostatistics.co.il (Hebrew) |
www.r-statistics.com (English)
----------------------------------------------------------------------------------------------




On Mon, Apr 26, 2010 at 9:00 AM, Charlotte Chang
<c.h.w.chang@gmail.com>wrote:
> optim
	[[alternative HTML version deleted]]

Great! Thank you for your help!

-Charlotte
>>
>> On Mon, Apr 26, 2010 at 1:12 AM, Tal Galili <tal.galili at
gmail.com> wrote:
>>>
>>> mm...
>>> I also noticed the function you wrote didn't use?parenthesis,
mixed b and c and used different names for K.
>>> Your code is a great?exercise?in debugging ?(no offense intended :)
)
>>> Try using:
>>> bird<-bird.density[0] # I assume this exists
>>> eqn<- function(K1, b1 = 1.22, c1 = .55) {
>>> ?? ? ? ? ? ? ? b<-1.22
>>> ? ? ? ? ? ? ? ?c<-0.55
>>> ? ? ? ? ? ? ? ?bird <- c1*bird + b1*bird*(1-b1*bird)*bird/K1
>>> ?? ? ? ? ? ?return(bird)
>>> }
>>>
>>> k1<-optim(c(0,10),eqn)
>>>
>>>
>>>
>>> ----------------Contact
Details:-------------------------------------------------------
>>> Contact me: Tal.Galili at gmail.com | ?972-52-7275845
>>> Read me: www.talgalili.com (Hebrew) | www.biostatistics.co.il
(Hebrew) | www.r-statistics.com (English)
>>>
----------------------------------------------------------------------------------------------
>>>
>>>
>>>
>>>
>>> On Mon, Apr 26, 2010 at 11:08 AM, Tal Galili <tal.galili at
gmail.com> wrote:
>>>>
>>>> Hi?Charlotte ,
>>>> I can't reproduce your code, but skimming through it -
>>>> It would appear that:
>>>> 1) in
>>>> eqn1<- function(K1, bird)
>>>> you didn't define "bird" (you did define it
before the function, so I'd suggest just removing it from the function call
like this:
>>>> eqn1<- function(K1)
>>>> 2) you didn't "return" and value at the end of
the function.
>>>> 3) you use different name in optim then in the function you
made.
>>>> 4) I see no point for you to use the for loop.
>>>>
>>>>
>>>> Consider trying the following code:
>>>> b<-1.22
>>>> c<-0.55
>>>> bird<-bird.density[0] # I assume this exists
>>>> eqn<- function(K1, bird) {
>>>> ?? ? ? ? ? ? ? b<-1.22
>>>> ? ? ? ? ? ? ? ?c<-0.55
>>>> ? ? ? ? ? ? ? ?bird <- 0.55*bird + b*bird*1-b*bird*bird/K
>>>> ?? ? ? ? ? ?return(bird)
>>>> }
>>>>
>>>> k1<-optim(c(0,10),eqn)
>>>>
>>>>
>>>> ----------------Contact
Details:-------------------------------------------------------
>>>> Contact me: Tal.Galili at gmail.com | ?972-52-7275845
>>>> Read me: www.talgalili.com (Hebrew) | www.biostatistics.co.il
(Hebrew) | www.r-statistics.com (English)
>>>>
----------------------------------------------------------------------------------------------
>>>>
>>>>
>>>>
>>>>
>>>> On Mon, Apr 26, 2010 at 9:00 AM, Charlotte Chang
<c.h.w.chang at gmail.com> wrote:
>>>>>
>>>>> optim
>>>
>>
>