[Wien] About kram!

[whoming]

Dear WIEN2K users,
     I am puzzled by the program kram.f in SRC_kram directory.
If  gamma (Lorentzian broadening with gamma) equals zero, the caluculation
of optical conductivity is  like this:

 sig(i,j)=-iimag*sigfac*(eps(i,j)-ione)*xe(i)

While if gamma is set a value, the calculation of optical conductivity becomes:

sig(i,j)=sigfac*eps(i,j)*xe(i)

The fomula of caluclation sig from epsilon should be:

The output of unit 12 and 13 (correspond to case.epsilon and case.sigmk,respectively) 
files is much more strange too:

The head of both like this:

         write(12,111) ecv,(hdre,hpol(j),hdim,hpol(j),j=1,ncol)
         write(13,112) ecv,(hsre,hpol(j),hsim,hpol(j),j=1,ncol)

take my case as an example:  
        Energy  [eV]  Re_eps_xx     Im_eps_xx     Re_eps_xy     Im_eps_xy        
        Energy  [eV] Re_sigma_xx   Im_sigma_xx   Re_sigma_xy   Im_sigma_xy

But the output of data of epsilon and optical conductivity in the source code is:
      write(12,113) xe(i),(DBLE(eps(i,j)),aimag(eps(i,j)),j=1,ncol)
      write(13,113) xe(i),(aimag(sig(i,j)),DBLE(sig(i,j)),j=1,ncol)
please pay attention to the output order of real and image part of epsilon and sigmak.
The header and the data is totaly mismatched in sigmak.

Would any one please give me some explaination on why it is so?
Thank you very much!

Yours,
whoming
2003/9/2