Comp Sci Fortran variable value changing randomly

AI Thread Summary
The discussion revolves around the unexpected changes in the value of the variable 'seed' in a Fortran program. The issue arises from the function congmult, which modifies the value of 'seed' as it generates random numbers, leading to discrepancies between iterations. Participants suggest checking whether function parameters are passed by reference or value, clarifying that Fortran uses pass-by-reference. It is emphasized that the random number generator (RNG) modifies the seed with each call, which is crucial for maintaining the sequence of generated numbers. Understanding this behavior is essential for effective programming in Fortran.
cpburris
Gold Member
Messages
37
Reaction score
4
This isn't really a question for which the template is useful..
Can anyone explain to me why the value of the variable 'seed' in location 1 is different from the value in location 2? The only thing happening in between is the defining of the varaible v(1) using the real function. I don't see where in that 'seed' is given a new value. I've tried google searching 'Fortran seed' and such and can't find any answer.
Code:
      JRDM = 83721563
      call baysprep(JRDM)
! -----------------------------------------
      subroutine baysprep(first)
          integer first,seed
          common /tobays/ v(53),oldbays,seed
          save /tobays/
          seed = abs(first)
       if (seed.gt.2147483647) seed = 24287
 
       do i = 1,53
!Location 1
        print*,'seed into do loop iteration=',seed
        v(1) = congmult(seed)
!Location 2
        print*,'seed coming out of do loop iteration=',seed
      end do
!
 real function congmult(ix)
!
       integer a,p,ix,b15,b16,xhi,xalo,leftlo,fhi,k
       data a,b15,b16,p/16807,32768,65536,2147483647/
      ! assigns variables these values
       xhi = ix / b16
       xalo = a*(ix - xhi*b16)
       leftlo = xalo / b16
       fhi = xhi*a + leftlo
       k = fhi / b15
       ix = (((xalo-leftlo*b16) - p) + (fhi-k*b15)*b16) + k
       if (ix.lt.0) ix = ix + p
       congmult = real(ix)/2.147483647e+09
       end
 
Last edited by a moderator:
Physics news on Phys.org
cpburris said:
This isn't really a question for which the template is useful..
Hi cpburris. Please use the template for posts to the homework forums regardless of whether you believe it is "suitable" or not. It's just one of those rules we're picky about :smile:

Check to see if your function parameters are call-by-reference or call-by-value. I see that seed is passed to function congmult which takes takes parameter ix and modifies it, so try printing out its value from inside that function just before it returns. See if the value shown matches that of 'seed coming out of do loop iteration'.
 
gneill said:
Hi cpburris. Please use the template for posts to the homework forums regardless of whether you believe it is "suitable" or not. It's just one of those rules we're picky about :smile:

Check to see if your function parameters are call-by-reference or call-by-value. I see that seed is passed to function congmult which takes takes parameter ix and modifies it, so try printing out its value from inside that function just before it returns. See if the value shown matches that of 'seed coming out of do loop iteration'.

Ah, yes you were correct. The modification of the value by the function was being passed back. Thank you very much.
 
You'e welcome. Glad I could help.
 
gneill said:
Check to see if your function parameters are call-by-reference or call-by-value.
The latter doesn't exist in Fortran. All variables are passed by reference.

@cpburris: Note that this is how this RNG keeps track of where it is in the sequence. You give it an initial seed, but then it modifies it every time a new random number is generated. It is important not to change the value of seed between calls to congmult.
 
DrClaude said:
The latter doesn't exist in Fortran. All variables are passed by reference.
It's something that programmers need to be aware of for whatever language they are using. Discovering the hard way is one path to enlightenment :smile:
 

Similar threads

Replies
11
Views
6K
Replies
2
Views
6K
Replies
13
Views
3K
Replies
3
Views
7K
Replies
1
Views
2K
Replies
7
Views
2K
Replies
12
Views
3K
Back
Top