需要将以下FORTRAN代码转换为C ++

Question

Im a very poor programmer, and i was given a program to supposedly help me on my aerodynamic hw. but its in fortran, and im trying to use MATLAB to run this program. any help on converting it to a language matlab understands? (preferabbly c++)

      program joukow
c

c   computes joukowski airfoil and finds pressure coefficient

c   currently set up for symmetric airfoil with sharp trailing edge

c   and chord length equal to one.

c   profile is written onto prof.dat and cp onto cp.dat

c      implicit real*8(a-h,o-z)

      complex z,zeta,cw
      dimension uz(100),vz(100),xi(100),eta(100),cp(100)
      dimension xout(100),yout(100)
         open(unit=8,file='prof.dat',status='unknown')
         open(unit=9,file='cp.dat',status='unknown')
      b=1.d0
      write(6,98)
      format(2x,'input the radius of the a-circle in z plane')
      read(5,99)a
      format(f10.0)
      xl=2.*a-1.+1./(2.*a-1.)

c      xl=a+1./a

c      chord=2.*xl

      chord=2.+xl
      del=a-b

c      del =0.1d0
      do 50 i=1,100
      ri=i
      theta=6.2832d0*ri/101.d0
      x=-del+a*cos(theta)
      y=a*sin(theta)
      z=cmplx(x,y)
      zeta=z+b**2/z

c

c  xi and eta are coordinates of points on airfoil

c

      xi(i)=real(zeta)

      eta(i)=aimag(zeta)

      cw=(1.-a**2/(z+del)**2)/(1.-b**2/z**2)
c

c  uz and vz are velocity components on the airfoil assuming the free-stream

c  speed is one.
c
      uz(i)=real(cw)
      vz(i)=-aimag(cw)

c

c  xout and yout are airfoil coordinates where the leading edge is at (0,0)

c  and the chordlength is one.

c

      xout(i)=(xl+xi(i))/chord
      yout(i)=eta(i)/chord
      write(8,100)xout(i),yout(i)
      format(2x,2f10.4)
      continue

c

c  now calculate the pressure coefficient cp

c

      write(6,200)
      format(2x,'pressure coefficients')
      do 70 i=1,50
      cp(i)=1.-(uz(i)**2+vz(i)**2)
      write(9,100)xout(i),cp(i)
      continue
      stop
      end

Answer 1

Matlab understands Fortran just fine -- check the documentation. And if that doesn't satisfy you, most of the lines in the program which do any computation could be typed into the Matlab console with very little modification. If you are a poor programmer, I suggest that you spend your time modifying the program into Matlab rather than into C++. I'll write more later if you don't get any better help than I have time for right now.

EDIT: first off, some information on using Fortran source files from Matlab. If you really don't want to (or can't or have performance reasons for not doing so) rewrite the Fortran into Matlab then turn it into a MEX file. Using f2c (or anything else, including your own time and effort) to first translate the Fortran into C or C++ seems pointless to me.

If you don't like that idea, here are some ideas on turning Fortran into Matlab.

First, all lines beginning with C or c are comments so you don't need to translate them. Start with your code:

  complex z,zeta,cw
  dimension uz(100),vz(100),xi(100),eta(100),cp(100)
  dimension xout(100),yout(100)

These lines declare a number of variables. You don't have to declare variables before you use them in Matlab but, there are sometimes good reasons to do so. You don't have to in Fortran either, though this is universally considered a bad idea these days. You could 'declare' these variables in Matlab with statements such as:

uz = zeros(100,1); 
vz = zeros(100,1);

By declaring these in advance in your Matlab you allocate memory for them once, and avoid some performance-reducing problems.

The next 2 lines:

     open(unit=8,file='prof.dat',status='unknown')
     open(unit=9,file='cp.dat',status='unknown')

open a couple of files for output. They are used later in write statements - forget them, write Matlab statements such as save xout instead.

The next line is Fortran but identical in Matlab:

  b=1.d0

The next lines get a value for the radius from the console:

  write(6,98)
  format(2x,'input the radius of the a-circle in z plane')
  read(5,99)a
  format(f10.0)

again, I suggest you forget these, just use the Matlab console to set the value of a . More Fortran that doesn't need to be translated (though I suggest you either drop the decimal points without following 0s or put a space between them and the subsequent * -- .* is a specific operator in Matlab):

  xl=2.*a-1.+1./(2.*a-1.)

  chord=2.+xl
  del=a-b

A Fortran do loop is the same as a Matlab for loop. Rewrite:

  do 50 i=1,100

as

for i = 1:100

As one of the other respondents has noted it's not clear where the matching end statement goes, you'll have to figure that out. Note that I'm just offering a line-by-line translation of Fortran into Matlab. It's not well-written Fortran, and I'm not offering well-written Matlab, I'll leave that to you.

This lot doesn't need to be translated:

  ri=i
  theta=6.2832d0*ri/101.d0 
  x=-del+a*cos(theta)
  y=a*sin(theta)

cmplx is a Fortran function which returns a complex number which has real part x and imaginary part y:

  z=cmplx(x,y)

In Matlab this would be z = x + y * i. Fortran uses ** for exponentiation, Matlab uses ^

  zeta=z+b**2/z

and so on and so on.

Hope that helps.

Answer 2

I used f2matlab and a little touching up afterward. Here is the cleaned up and compilable fortran90 code:

program joukow
 !
 !   computes joukowski airfoil and finds pressure coefficient
 !   currently set up for symmetric airfoil with sharp trailing edge
 !   and chord length equal to one.
 !   profile is written onto prof.dat and cp onto cp.dat
 !      implicit real*8(a-h,o-z)
 complex z,zeta,cw
 dimension uz(100),vz(100),xi(100),eta(100),cp(100)
 dimension xout(100),yout(100)
 open(unit=8,file='prof.dat',status='unknown')
 open(unit=9,file='cp.dat',status='unknown')
 b=1.d0
 write(6,98)
98 format(2x,'input the radius of the a-circle in z plane')
 read(5,99)a
99 format(f10.0)
 xl=2.*a-1.+1./(2.*a-1.)
!      xl=a+1./a
!      chord=2.*xl
 chord=2.+xl
 del=a-b
!      del =0.1d0
 do i=1,100
  ri=i
  theta=6.2832d0*ri/101.d0
  x=-del+a*cos(theta)
  y=a*sin(theta)
  z=cmplx(x,y)
  zeta=z+b**2/z
  !
  !  xi and eta are coordinates of points on airfoil
  !
  xi(i)=real(zeta)
  eta(i)=aimag(zeta)
  cw=(1.-a**2/(z+del)**2)/(1.-b**2/z**2)
  !
  !  uz and vz are velocity components on the airfoil assuming the free-stream
  !  speed is one.
  !
  uz(i)=real(cw)
  vz(i)=-aimag(cw)
  !
  !  xout and yout are airfoil coordinates where the leading edge is at (0,0)
  !  and the chordlength is one.
  !
  xout(i)=(xl+xi(i))/chord
  yout(i)=eta(i)/chord
  write(8,100)xout(i),yout(i)
100 format(2x,2f10.4)
 end do
!
!  now calculate the pressure coefficient cp
!
 write(6,200)
200 format(2x,'pressure coefficients')
 do  i=1,50
  cp(i)=1.-(uz(i)**2+vz(i)**2)
  write(9,100) xout(i),cp(i)
 end do
 stop
end program joukow

Here is the resulting matlab code:

function hw1(varargin)
%
%   computes joukowski airfoil and finds pressure coefficient
%   currently set up for symmetric airfoil with sharp trailing edge
%   and chord length equal to one.
%   profile is written onto prof.dat and cp onto cp.dat
%      implicit real*8(a-h,o-z)

format_99=['%10.0f'];
format_100=[repmat(' ',1,2),repmat('%10.4f',1,2),'\n'];
format_200=[repmat(' ',1,2),'pressure coefficients \n'];

fid_8=fopen('prof.dat','w+');
fid_9=fopen('cp.dat','w+');
b=1.0d0;
a=input('input the radius of the a-circle in z plane');
xl=2..*a-1.+1../(2..*a-1.);
%      xl=a+1./a
%      chord=2.*xl
chord=2.+xl;
del=a-b;
%      del =0.1d0
for i=1:100;
 ri=i;
 theta=6.2832d0.*ri./101.0d0;
 x=-del+a.*cos(theta);
 y=a.*sin(theta);
 z=complex(x,y);
 zeta=z+b.^2./z;
 %
 %  xi and eta are coordinates of points on airfoil
 %
 xi(i)=real(zeta);
 eta(i)=imag(zeta);
 cw=(1.-a.^2./(z+del).^2)./(1.-b.^2./z.^2);
 %
 %  uz and vz are velocity components on the airfoil assuming the free-stream
 %  speed is one.
 %
 uz(i)=real(cw);
 vz(i)=-imag(cw);
 %
 %  xout and yout are airfoil coordinates where the leading edge is at (0,0)
 %  and the chordlength is one.
 %
 xout(i)=(xl+xi(i))./chord;
 yout(i)=eta(i)./chord;
 fprintf(fid_8,format_100,xout(i),yout(i));
end; i=100+1;
%
%  now calculate the pressure coefficient cp
%
fprintf(1,format_200);
for  i=1:50;
 cp(i)=1.-(uz(i).^2+vz(i).^2);
 fprintf(fid_9,format_100, xout(i),cp(i));
end;  i=50+1;
end %program joukow

They both give the same results for me. I didn't check the algorithm for correctness, though, just converted the code.

Answer 3

我不知道它的支持程度如何--但是以前最简单的方法是f2c，它可以将fortran直接转换为c代码。