Question

I’m a bit puzzled by the value of h in this experiment. In cpp,

int h,J=3,n=200,p=3,h_m=(n+p+1)/2;
float rt=(float)h_m/n;

for(int j=0,j<J,j++){
   h=floor((j+1)/J*rt*(n-p-1)+p+1);
   std::cout<<h<<" "<<j<<" "<<rt<<" "<<n-p-1<<" "<<h_i<<" "<<J<<std::endl;
}

gives:

4 0 0.51 196 7  3
4 1 0.51 196 7  3
103 2 0.51 196 7  3

I’m trying to get (this is in R):

n<-200
p<-3
h_m<-as.integer((n+p+1)/2)
J<-3
j<-0:(J-1)
rt<-h_m/n
floor((j+1)/J*rt*(n-p-1)+p+1)
[1]  37  70 103

what is wrong in the cpp construct?

Answer 1

This

(j+1)/J

is integer division, and therefore gets truncated. Expicitly cast one of the operands to float:

(j+1)/(float)J