I'm pulling my hair (and there's not much left!) on this one. Basically I'm
not getting the same result t when I "step" through the program and evaluate
each element separately than when I use the outer() function in the
FindLikelihood() function below.
Here's the functions:
Dk<- function(xk,A,B)
{
n0 *(A*exp(-0.5*(xk/w)^2) + B)
}
FindLikelihood <- function(Nk)
{
A <- seq(0.2,3,by=0.2)
B <- seq(0.2,3,by=0.2)
k <-7
L <- outer(A, B, function(A,B) sum( (Nk*log(Dk(seq(-k,k),A,B))) -
Dk(seq(-k,k),A,B) ))
return(L)
}
where Nk <- c(70 , 67 , 75 , 77 , 74 ,102, 75, 104 , 94 , 74 , 78 , 79 , 83
, 73 , 76)
Here's an excerpt from my debug session..
> Nk
[1] 70 67 75 77 74 102 75 104 94 74 78 79 83 73 76
> debug(FindLikelihood)
> L<-FindLikelihood(Nk)
debugging in: FindLikelihood(Nk)
debug: {
A <- seq(0.2, 3, by = 0.2)
B <- seq(0.2, 3, by = 0.2)
k <- 7
L <- outer(A, B, function(A, B) sum((Nk * log(Dk(seq(-k,
k), A, B))) - Dk(seq(-k, k), A, B)))
return(L)
}
Browse[1]> n
debug: A <- seq(0.2, 3, by = 0.2)
Browse[1]> n
debug: B <- seq(0.2, 3, by = 0.2)
Browse[1]> n
debug: k <- 7
Browse[1]> n
debug: L <- outer(A, B, function(A, B) sum((Nk * log(Dk(seq(-k, k),
A, B))) - Dk(seq(-k, k), A, B)))
Browse[1]> sum((Nk * log(Dk(seq(-k, k),0.2,0.2))) - Dk(seq(-k, k), 0.2,
0.2)) # WHY DOES THIS LINE GIVE ME THE CORRECT RESULT WHEN I SUBSTITUTE
0.2, 0.2 FOR A AND B
[1] 2495.242
Browse[1]> outer(A, B, function(A, B) sum((Nk * log(Dk(seq(-k, k),
+ A, B))) - Dk(seq(-k, k), A, B)))
[,1] [,2] [,3] [,4] [,5] [,6] [,7]
[,8]
[1,] 58389.48 58389.48 58389.48 58389.48 58389.48 58389.48 58389.48
58389.48 # BUT ELEMENT (1,1) WHICH SHOULD ALSO BE (A,B) = (0.2, 0.2),
GIVES THE INCORRECT RESULT????
[2,] 58389.48 58389.48 58389.48 58389.48 58389.48 58389.48 58389.48
58389.48
[3,] 58389.48 58389.48 58389.48 58389.48 58389.48 58389.48 58389.48
58389.48
[4,] 58389.48 58389.48 58389.48 58389.48 58389.48 58389.48 58389.48
58389.48
[5,] 58389.48 58389.48 58389.48 58389.48 58389.48 58389.48 58389.48
58389.48
[6,] 58389.48 58389.48 58389.48 58389.48 58389.48 58389.48 58389.48
58389.48
[7,] 58389.48 58389.48 58389.48 58389.48 58389.48 58389.48 58389.48
58389.48
[8,] 58389.48 58389.48 58389.48 58389.48 58389.48 58389.48 58389.48
58389.48
[9,] 58389.48 58389.48 58389.48 58389.48 58389.48 58389.48 58389.48
58389.48
[10,] 58389.48 58389.48 58389.48 58389.48 58389.48 58389.48 58389.48
58389.48
[11,] 58389.48 58389.48 58389.48 58389.48 58389.48 58389.48 58389.48
58389.48
[12,] 58389.48 58389.48 58389.48 58389.48 58389.48 58389.48 58389.48
58389.48
[13,] 58389.48 58389.48 58389.48 58389.48 58389.48 58389.48 58389.48
58389.48
[14,] 58389.48 58389.48 58389.48 58389.48 58389.48 58389.48 58389.48
58389.48
[15,] 58389.48 58389.48 58389.48 58389.48 58389.48 58389.48 58389.48
58389.48
[,9] [,10] [,11] [,12] [,13] [,14] [,15]
[1,] 58389.48 58389.48 58389.48 58389.48 58389.48 58389.48 58389.48
[2,] 58389.48 58389.48 58389.48 58389.48 58389.48 58389.48 58389.48
[3,] 58389.48 58389.48 58389.48 58389.48 58389.48 58389.48 58389.48
[4,] 58389.48 58389.48 58389.48 58389.48 58389.48 58389.48 58389.48
[5,] 58389.48 58389.48 58389.48 58389.48 58389.48 58389.48 58389.48
[6,] 58389.48 58389.48 58389.48 58389.48 58389.48 58389.48 58389.48
[7,] 58389.48 58389.48 58389.48 58389.48 58389.48 58389.48 58389.48
[8,] 58389.48 58389.48 58389.48 58389.48 58389.48 58389.48 58389.48
[9,] 58389.48 58389.48 58389.48 58389.48 58389.48 58389.48 58389.48
[10,] 58389.48 58389.48 58389.48 58389.48 58389.48 58389.48 58389.48
[11,] 58389.48 58389.48 58389.48 58389.48 58389.48 58389.48 58389.48
[12,] 58389.48 58389.48 58389.48 58389.48 58389.48 58389.48 58389.48
[13,] 58389.48 58389.48 58389.48 58389.48 58389.48 58389.48 58389.48
[14,] 58389.48 58389.48 58389.48 58389.48 58389.48 58389.48 58389.48
[15,] 58389.48 58389.48 58389.48 58389.48 58389.48 58389.48 58389.48
Browse[1]>
As "commented" above, when I evaluate a single A,B element (i.e. A=0.2,
B=0.2) I get a different result than when I use OUTER() which should also be
evaluating at A=0.2, B=0.2??
Any help appreciated. I know I'm probably doing something overlooking
something simple, but can anyone point it out???
Thanks!
-Scott
Scott Norton, Ph.D.
Engineering Manager
Nanoplex Technologies, Inc.
2375 Garcia Ave.
Mountain View, CA 94043
www.nanoplextech.com
[[alternative HTML version deleted]]
______________________________________________
[EMAIL PROTECTED] mailing list
https://www.stat.math.ethz.ch/mailman/listinfo/r-help
