Fortran Document not compiling

[Fairfield33]

This document on Integer FFT's is not compiling for me. I am a beginner in Fortran and am having some problems. I would greatly appreciate any Help with this document.


SUBROUTINE FASTI (IRE,IMG,ISIZE)
C PROVIDE INTEGER WORD LENGTH AND PIE TO AVAILABLE PRECISION
DATA NBITTS/16/,PIE/3.14159/
DIMENSION IRE (1) ,IMG(1)
IBIG=2**(NBITS-2)
IBG2=IBIG/2
XBIG=FLOAT(IBIG)
N=IABS(ISIZE)
C SET UP INITIAL TRIG FUNCTIONS
Z=PIE/FLOAT(N)
IAS=SIN(Z)*XBIG+0.5
IAS=COS(Z)*XBIG+0.5
C SET UP INTIAL VALUES OF TRANSORM SPLIT
IFA = N/2
C IF AN INVERSE TRANSFORM, CONJUGATE THE DATA
IF(ISIZE)1,3,3
1 DO 2 K=1,N
2 IMG(K)=-IMG(K)
3 I TIME=0
4 IFAB=IFA*2
ITIME=ITIME+1
C DOUBLE THE BASIC ANGLE
IBSIN = 2*MPY(IAS,IAC)
IBCOS=MPY(IAS,IAS)
IAS=IBSIN
IAC=(IBG2-IBCOS)*2
CALL COMP(IAS,IAC)
C MAIN CALCLATION OF RADIX 2 TRANSFORMS
ISW=0
ICW=IBIG
DO 11 LA = 1,IFA
DO 9 IP=LA,N,IFAB
IQ=IP+IFA
C EXTRACT DATA AND SCALE
K1=IRE(IP)
K2=TRE(IQ)
L1=IMG(IP)
L2=IMG(IQ)
IF(ISIZE)6,18,5
5 K1=K1/2
K2=K2/2
L1=L1/2
L2=L2/2
C FLUTTER BY
6 IRE(IP)=K1+K2
K1=K1-K2
IMG(IP)=L1+L2
L1=L1-L2
IF(LA-1)18,7,8
7 IRE(IQ)=K1
IMG(IQ)=L1
GO TO 9
C MULTIPLY BY TWIDDLE FACTORS IF REQUIRED
8 IRE(IQ)=MPY(K1,ICW)+MPY(L1,ISW)
IMG(IQ)=MPY(L1,ICW)+MPY(K1,ISW)
9 CONTINUE
IF(LA-IFA)10,11,18
C CALCULATE A NEW SET OF TWIDDLE FACTORS
10 IT=ICW-MPY(IBSIN,ISW)-2*MPY(IBCOS,ICW)
ISW=ISW+MPY(IBSIN,ICW)-2*MPY(IBCOS,ISW)
ICW=IT
CALL COMP(ISW,ICW)
11 CONTINUE
IFA=IFA/2
IF(IFA)12,12,4
C TRANSFORM COMPLETE - CONJUGATE RESULT IF INVERSE
12 IF(ISIZE)13,18,15
13 DO 14 K=1,N
14 IMG(K)=-IMG(K)
15 DO 17 J20=1,N
IZ=J20-1
CALL BITRV(IZ,II,ITIME)
IF(II-J20)17,17,16
16 II=II+1
IT=IRE(II)
IRE(II)=IRE(J20)
IRE(J20)=IT
IT=IMG(II)
IMG(II)=IMG(J20)
IMG(J20)=IT
17 CONTINUE
18 RETURN
END
FUNCTION MPY(I,J)
C FUNCTION TO MULTIPLY FIXED POINT FRACTIONS
C THIS VERSION INCORPORATES ROUNDING
C
C PROVIDE CONSTANTS FOR DECOMPOSISTION FO PRODUCT
DATA IMAD,IMID/64,128/
ILEF=I/IMID
IRIT=I-ILEF*IMID
JLEF=J/IMID
JRIT=J-JLEF*IMID
C DO NOT REARRANGE THE FOLLOWING MULTIPLICATION
II=(IRIT*JRIT)/IMID+IRIT*JLEF+JRIT*ILEF
MPY= (II+ISIGN(IMAD,II))/IMID+ILEF*JLEF
RETURN
END

FUNCTION MPY(I,J)
C
C FUNCTION TO MULITPLY FIXED POINT FRACTIONS
C THIS VERSION DOES NOT ROUND
C
C PROVED CONSTANTS FOR DECOMPOSITION OF PRODUCT
DATA IMID/128/
ILEF=I/IMID
IRIT=I-ILEF*IMID
JLEF=J/IMID
JRIT=J-JLEF*IMID
C DO NOT REARRANGE THE FOLLOWING MULTIPLICATION
MPY=((IRIT*JRIT)/IMID+IRIT*JLEF+JRIT*ILEF)/IMID+ILEF*JLEF
RETURN
END

SUBROUTINE BITRV(I,J,NBITS)
C
C SUBROUTINE WHICH RETURNS NBITS BITS OF I REVERSED IN J.
C
IZ=I
IJ=0
DO 1 K=1,NBITS
C DO NOT REARRANGE THE FOLLOWING STATEMENT
IJ=IJ+IJ+IZ/2*2
1 IZ=IZ/2
J=IJ
RETURN
END

 

[xwb]

If you are pasting code, enclose it withing [ CODE] and [/CODE]. Looks like F77 so you have to follow the F77 rules.

1) Labels from cols 1-5
2) Continuation in col 6
3) Code begins in col 7 - there are a few errors in this area. If you can do not use tabs. Some compilers take a tab as 8 characters, others take it as one. eg

      DO 1 K=1,NBITS
C DO NOT REARRANGE THE FOLLOWING STATEMENT
         IJ=IJ+IJ+IZ/2*2
   1     IZ=IZ/2

The IZ=IZ/2 was in the wrong column.

By default, integers begin with letters from I to N, everything else is real. If IAS is meant to be real, it has to be declared as such. Try that for starters: see how you get on.

 

[xwb]

Which version of Fortran are you using. Just realized that the program has 3 way jumps which aren't supported in Fortran 90/95.

 

[Fairfield33]

I only know it is F77. The errors are pretty much fixed other than --->In function `main':
: undefined reference to `MAIN__'

Any ideas?

 

[xwb]

Have you got something like

program main
   call fasti (1, 2, 3)
   stop
end program

It is probably telling you that you are missing a program module.

 

[Fairfield33]

        SUBROUTINE FASTI (IRE,IMG,ISIZE)
C PROVIDE INTEGER WORD LENGTH AND PIE TO AVAILABLE         C PRECISION
       DATA NBITS/16/,PIE/3.14159/
       DIMENSION IRE (1) ,IMG(1)
       IBIG=2**(NBITS-2)
       IBG2=IBIG/2
       XBIG=FLOAT(IBIG)
       N=IABS(ISIZE)
C SET UP INITIAL TRIG FUNCTIONS
       Z=PIE/FLOAT(N)
       IAS=SIN(Z)*XBIG+0.5
       IAS=COS(Z)*XBIG+0.5
C SET UP INTIAL VALUES OF TRANSORM SPLIT
       IFA = N/2
C IF AN INVERSE TRANSFORM, CONJUGATE THE DATA
       IF(ISIZE)1,3,3
  1    DO 2 K=1,N
  2    IMG(K)=-IMG(K)
  3    ITIME=0
  4    IFAB=IFA*2
       ITIME=ITIME+1
C DOUBLE THE BASIC ANGLE
       IBSIN = 2*MPY(IAS,IAC)
       IBCOS=MPY(IAS,IAS)
       IAS=IBSIN
       IAC=(IBG2-IBCOS)*2
       CALL COMP(IAS,IAC)
C  CALCLATION OF RADIX 2 TRANSFORMS      
       ISW=0
       ICW=IBIG
       DO 11 LA = 1,IFA
       DO 9 IP=LA,N,IFAB
       IQ=IP+IFA
C EXTRACT DATA AND SCALE
       K1=IRE(IP)
C CHANGED TO IRE FROM TRE
       K2=IRE(IQ)
       L1=IMG(IP)
       L2=IMG(IQ)
       IF(ISIZE)6,18,5
 5     K1=K1/2
       K2=K2/2
       L1=L1/2
       L2=L2/2
C FLUTTER BY
 6     IRE(IP)=K1+K2
       K1=K1-K2
       IMG(IP)=L1+L2
       L1=L1-L2
       IF(LA-1)18,7,8
 7     IRE(IQ)=K1
       IMG(IQ)=L1
       GO TO 9
C MULTIPLY BY TWIDDLE FACTORS IF REQUIRED
 8     IRE(IQ)=MPY(K1,ICW)+MPY(L1,ISW)
       IMG(IQ)=MPY(L1,ICW)+MPY(K1,ISW)
 9     CONTINUE
       IF(LA-IFA)10,11,18
C CALCULATE A NEW SET OF TWIDDLE FACTORS
 10    IT=ICW-MPY(IBSIN,ISW)-2*MPY(IBCOS,ICW)
       ISW=ISW+MPY(IBSIN,ICW)-2*MPY(IBCOS,ISW)
       ICW=IT
       CALL COMP(ISW,ICW)
 11    CONTINUE
       IFA=IFA/2
       IF(IFA)12,12,4
C TRANSFORM COMPLETE - CONJUGATE RESULT IF INVERSE
 12    IF(ISIZE)13,18,15
 13    DO 14 K=1,N
 14    IMG(K)=-IMG(K)
 15    DO 17 J20=1,N
       IZ=J20-1
       CALL BITRV(IZ,II,ITIME)
       IF(II-J20)17,17,16
 16    II=II+1
       IT=IRE(II)
       IRE(II)=IRE(J20)
       IRE(J20)=IT
       IT=IMG(II)
       IMG(II)=IMG(J20)
       IMG(J20)=IT
 17    CONTINUE
 18    RETURN
       END
C       FUNCTION MPY(I,J)
C FUNCTION TO MULTIPLY FIXED POINT FRACTIONS
C THIS VERSION INCORPORATES ROUNDING
C
C PROVIDE CONSTANTS FOR DECOMPOSISTION FO PRODUCT
C       DATA IMAD,IMID/64,128/
C       ILEF=I/IMID
C       IRIT=I-ILEF*IMID
C       JLEF=J/IMID
C       JRIT=J-JLEF*IMID
C DO NOT REARRANGE THE FOLLOWING MULTIPLICATION
C     II=(IRIT*JRIT)/IMID+IRIT*JLEF+JRIT*ILEF
C       MPY= (II+ISIGN(IMAD,II))/IMID+ILEF*JLEF
C       RETURN
C       END

       FUNCTION MPY(I,J)
C
C FUNCTION TO MULITPLY FIXED POINT FRACTIONS
C THIS VERSION DOES NOT ROUND
C
C PROVED CONSTANTS FOR DECOMPOSITION OF PRODUCT
        DATA IMID/128/
        ILEF=I/IMID
        IRIT=I-ILEF*IMID
        JLEF=J/IMID
        JRIT=J-JLEF*IMID
C DO NOT REARRANGE THE FOLLOWING MULTIPLICATION
        MPY=((IRIT*JRIT)/IMID+IRIT*JLEF+JRIT*ILEF)/IMID+ILEF*JLEF
        RETURN
        END

        SUBROUTINE BITRV(I,J,NBITS)
C
C SUBROUTINE WHICH RETURNS NBITS BITS OF I REVERSED IN J.
C
        IZ=I
        IJ=0
        DO 1 K=1,NBITS
C DO NOT REARRANGE THE FOLLOWING STATEMENT
        IJ=IJ+IJ+IZ/2*2
 1      IZ=IZ/2
        J=IJ
        RETURN
        END
       SUBROUTINE COMP (IS,IC)
C
C SUBROUTINE TO COMPENSATE SIN AND COS FOR MODULUS ERRORS
C
       IT=MPY(IS,IS)/2+MPY(IC,IC)/2
       IS=(IS-MPY(IS,IT))+IS/2
       IC=(IC-MPY(IC,IT))+IC/2
       RETURN
       END
C
C MAIN METHOD
C
       PROGRAM MAIN
       DIMENSION IRE(1),IMG(1)
       CALL FASTI(IRE,IMG,2)
       STOP
       END PROGRAM

Thanks to your help I have this code compiling correctly. However I am not certain of how exactly to print out results I need. I believe I need the results of IRE and IMG. I am actually using a converter for this specific program f2j and if anyone knows anything about an error that states "Ok... compiling 'original.f' to JAVA
original.f:
fasti:
DIMENSION is not implemented."

I believe it has something to do with my declaration of the one dimensional arrays in the main method but i am not certain.

 

[ArkM]

1. You MUST define real extents for arrays in PROGRAM (not IMG(1) etc).
2. Try declare INTEGER IMG(set_real_extent),... instead of DIMENSION...

 

[Fairfield33]

But dont you need to declare one dimensional arrays that need to be passed into the FASTI subroutine