R help - Feb 2010 - Trouble with optim function

Hi all,

I'm trying to make a little script to determine an "unknown" rate
for a
number of known exponential trials.



My Code:
#Set Trials and generate number
trials=100
rand<-runif(1,0,1)
vector=0

#Generate vector of 100 random exponentials and sum them
for (i in 1:100) {
vector<-rexp(trials,rate=rand)
}
sumvect=sum(vector)

#Create the log likelihood function
LL<-function(x) {(trials*log(x))-(x*sumvect)}

optim(1,LL,method="BFGS")



The "rand" variable should be between 0 and 1 and is what I am trying
to
approximate. However, as it is I generally get a value in the tens of 
thousands. I'm sure it's something simple but I can't find my error.

Jon

You're minimizing the log likelihood, but you want to minimize the
*negative*
log likelihood.  Also, optimize() is better than optim() if you have a
function of only one argument.

Replace


Jon Moroney wrote:
> 
> #Create the log likelihood function
> LL<-function(x) {(trials*log(x))-(x*sumvect)}
> 
> optim(1,LL,method="BFGS")
> 
with

LL<-function(x) {-trials*log(x)+x*sumvect}
optimize(LL,c(0,1e5))

or better yet, to avoid errors and make generalizeable, do this

LL<-function(x) {-sum(dexp(vector,x,log=TRUE))}
optimize(LL,c(0,1e5)) 

-- 
View this message in context:
http://n4.nabble.com/Trouble-with-optim-function-tp1559382p1560118.html
Sent from the R help mailing list archive at Nabble.com.

optim really isn't intended for [1D] functions. And if you have a
constrained search area,
it pays to use it. The result you are getting is like the second root of a
quadratic that
you are not interested in.

You may want to be rather careful about the problem to make sure you have the
function you
intend.

> I'm trying to make a little script to determine an "unknown"
rate for a
> number of known exponential trials.
> 
> 
> 
> My Code:
> #Set Trials and generate number
> trials=100
> rand<-runif(1,0,1)
> vector=0
> 
> #Generate vector of 100 random exponentials and sum them
> for (i in 1:100) {
> vector<-rexp(trials,rate=rand)
> }
> sumvect=sum(vector)
> 
> #Create the log likelihood function
> LL<-function(x) {(trials*log(x))-(x*sumvect)}
> 
> optim(1,LL,method="BFGS")
> 
> 
> 
> The "rand" variable should be between 0 and 1 and is what I am
trying to
> approximate. However, as it is I generally get a value in the tens of 
> thousands. I'm sure it's something simple but I can't find my
error.
How about?

 > ans2<-optimize(f=LL,interval=c(0,1))
 > ans2
$minimum
[1] 6.610696e-05

$objective
[1] -962.4322

 >

J Nash