In the program below, the variable x is always a tiny number close to 0. It
never gets to be more than about 1e-314. Variable y does get the whole range
of rand() (0 - .9999999)
This leads to all estimates of PI being 4.00000.
If I add another variable t by uncommenting the line t=rand(), then t is
incorrect but both x and y get the full range of rand() and the program
works OK.
I'm using the 64-bit version of gfortran with MSYS2-64 on Win 10.
! Program to compute Pi using monte carlo methods
program monte_pi
implicit none
integer niter,i,j
integer seed
real*4 count
real *8 x,y,pi(100),z,t
real*8 rand
! initialize random numbers
seed = 35791246
call srand (seed)
do j= 1,100
niter = niter+100
count =0
do i=1,niter
! t = rand()
x=rand()
y=rand()
z= x*x +y*y
if (z .le. 1) count =count+1
end do
pi(j)= count/niter*4.
write(*,10) niter,pi(j)
10 format('Number of trials is: 'i5,' estimate of pi is:',f8.5)
end do
end
