Newton Raphson Code for R Error with Diff Function

Question

I am having issues running my R code. I keep getting the error: Error in while (Diff > 1e-06) { : missing value where TRUE/FALSE needed

Here is the code I am using:

     Dose     with.tumor   total.animals
  1    0             3            50
  2  188             6            50
  3  375             6            50
  4  750            12            50

  X1=as.matrix(rep(1,4))
  X2a=as.matrix(c(0,188,375,750))
  X2=as.matrix(log(X2a))
  X=cbind(X1,X2)

a=1
b=1

B=matrix(c(a,b),nrow=2,ncol=1,byrow=F)

Diff=1
while(Diff>.000001){
P=(1-exp(-exp(a+b*X)))^-1
ni=c(rep(50,4))
V=diag(4)
for(i in 1:4)
{V[i,i]=diag(ni[i]%*%P[i]%*%(1-P[i]))}
R=c(3,6,6,12)
S=matrix(data=1,nrow=4,ncol=1)
for (i in 1:4)
{S[i]=R[i]-ni[i]*P[i]}

new.B=B+solve(t(X)%*%V%*%X)%*%(t(X)%*%S)
Diff=max(abs(new.B-B))
B=new.B}

Any insight would be appreciated.

Welcome to SO. This code is not reproducible. Please read [this](http://stackoverflow.com/questions/5963269/how-to-make-a-great-r-reproducible-example). — agstudy, Mar 11 '14 at 21:56
Thanks for the comment. I inserted the data set I have for the problem. — user3408107, Mar 11 '14 at 23:07
I apologize. I saved the wrong draft in my editing. This is the code with the data outlined. — user3408107, Mar 11 '14 at 23:26

score 1 · Answer 1 · answered Mar 11 '14 at 23:26

You have NaN when computing new.B. Usually, this error pops up because you are comparing NaN with a number.

Your t(X) has Inf as one of its elements. This arises from the fact that you are taking log(0) when you do X2a=as.matrix(c(0,188,375,750)) and then X2=as.matrix(log(X2a))

Newton Raphson Code for R Error with Diff Function

1 Answers1