CODEWITHSUNDEEP
variablesfortranintegration
user3311029
user3311029

Reputation: 25

Simpson's rule, declaration of user requested input not accepted? fortran

I'm working on this problem of getting the area under a tan curve using Simpson's rule. However, I keep getting errors I don't understand - even on different compilers like gfortran and ifort.

I realize that I could make pi a parameter in the main body of the program, but I'd rather just find out where I'm going wrong. Any help would be very much appreciated.

module consts

        !Declare pi as a global variable 
     real(4),parameter :: pi=3.1415927
end module


PROGRAM simpson
  use consts
  implicit none
  REAL :: area, h, sumo, sume
  INTEGER (kind=4) :: i, j, k !Loop index, Counter, Number of points
  REAL (kind=4) ::  rad, TanTab(0) !Radian variable to be calculated in Subroutine, Result array
!k = number of steps
   WRITE(6,*) "Please enter number of steps required (odd number)"
   READ(5,*) k

    IF(MOD(k,2) ==  1) THEN

CONTINUE
    ELSE
WRITE(*,*) "Hey, I said as an odd number"
    READ(5,*) k

    ENDIF


h = (60.0)/(k-1) ! Defining step size in relation to number of sampling points

 DO i=1,61,1
! Get table of tan
     call degtorad((i-1)*1.0, rad)

     TanTab(j) = tan(rad)

     j=j+1
    write(*,*) "Tan(", i, ")", TanTab(i)
    write(*,*) "Tan(", j, ")", TanTab(j)
 ENDDO

  DO j=1,k-1,h 
    IF(MOD(k,2) == 1) THEN
    sumo = sumo + TanTab(j)  !sum of odd functions
    ELSE
    sume = sume + TanTab(j)   !sum of even functions

    area = (h/3)*( (4*sumo) + (2*sume) + tan(1.047))
WRITE(6,*) area
    ENDIF
  ENDDO





END PROGRAM simpson

Upvotes: 0

Views: 538

Answers (3)

Peter Petrik
Peter Petrik

Reputation: 10195

You have more mistakes in your code:

  • in calculation of h, you are using degs and not rads
  • upper bound of integral connected to number of items in TanTab array
  • not init variables to 0 (sumo, sume, ...)
  • calculation of area for each step, could be done outside do loop
  • out-of-bounds mistakes

You should start using debugger and learn how to find problems in your code. The working version is here:

  PROGRAM simpson
  implicit none
  REAL :: area, h, sumo, sume
  INTEGER (kind=4) :: i!Loop index
  integer, parameter :: k = 100
  REAL (kind=4) ::  rad, TanTab(k+1) !Radian variable to be calculated in Subroutine, Result array

  sumo = 0.0
  sume = 0.0
  area = 0.0
  h = 0.0174532925 * 60.0/real(k) ! Defining step size in relation to number of sampling points

  DO i=1,k+1
      rad = 0.0174532925 * 60.0 * real(i-1) /real(k)
      TanTab(i) = tan(rad)
      ! write(*,*) "Tan(", i-1, ")", TanTab(i)
  ENDDO

  DO i=2,k
      IF(MOD(i,2) == 1) THEN
        sumo = sumo + TanTab(i)  !sum of odd functions
      ELSE
        sume = sume + TanTab(i)  !sum of even functions
      ENDIF
  ENDDO

  ! by simson
  area = (h/3)*( TanTab(1) + (2*sume) + (4*sumo) + TanTab(k+1))
  WRITE(*,*) area

  ! by integration
  area = log(1.0) - log(cos(0.0174532925 * 60.0))
  WRITE(*,*) area
  END PROGRAM simpson

Upvotes: 1

M. S. B.
M. S. B.

Reputation: 29401

Some things to think about: The loop DO i=1,61,1 suggests that you should be setting TanTab(i), not TanTab(j). Why bother with the additional variable j that increments as i does? And since the loop goes from 1 to 61 suggests that you should declare TanTab as real, dimension (1:61) :: TabTab, otherwise its not going to be large enough to store the values. But then your next loop, DO j=1,k-1,h is different. Are you sure that it is accessing TabTab (j) in a consistent manner?

Upvotes: 0

High Performance Mark
High Performance Mark

Reputation: 78374

I can see one probably cause of a crash at run time, sparking a segmentation violation:

You use j in the line

 TanTab(j) = tan(rad)

before you assign a value to it. Fortran doesn't do any automatic initialisation of variables.

I can also see something very fishy. You declare tantab to have dimension 0:

 REAL (kind=4) ::  rad, TanTab(0)

Now Fortran will happily access elements of arrays outside their declared bounds and sometimes (sometimes for a very long time) the users of a program won't notice the problem. If you compile your code with the option -check bounds (that's the Intel version, other compilers have the same facility, consult your documentation) and then try running your code you should get a run-time error.

And I agree with @AlexanderVogt (I usually do) about your use of a real variable for the loop index. Don't do that.

Upvotes: 1

Related Questions