[Wien] possibly solved bug : spaghetti SIGSEV with large supercells and SOC

[Muechler, Lukas]

Hey everyone,

I've bee having an issue with plotting the bandstructure of large supercells with SOC, as can also be found in the Mailingslist for several cases.

The error that occurs is as follows

" number of k-points read in case.vector=  (something)
forrtl: severe (174): SIGSEGV, segmentation fault occurred"

I noticed that this only occurs when systems with a large number of bands are used, so I looked into spagh.f and found this:
-------------------------------------------------------------------------------------------------
!
!.....ALL K-VECTORS HAVE BEEN READ; SEARCH FOR K-POINT WITH SMALLEST
!     NUMBER OF EIGENVALUES
!
 200  continue
      nu_min=999
      write(*,*) 'number of k-points read in case.vector=',n_kpt
      do 205 j=1,n_kpt
         if (n_ene(j).lt.nu_min)  then
            nu_min=n_ene(j)
            k_min=j
         endif
 205  continue
      write(6,*) 'smallest number eigenvalues at k=',k_min,' (', &
       k_name(k_min),')'
      write(6,*) '         =',nu_min
-------------------------------------------------------------------------------------------------

Since the program stops at this point and the number of bands in my case (heavy atoms) is larger than 999, I increased it to 1500 and now it works.
The bandstructure looks good aswell, so I think this number should be higher than 999.

Best,

Lukas Muechler